Fasting alters histone methylation in paraventricular nucleus through regulating of polycomb repressive complex 2. Ying Jiang

Transcription

1 Fsting lters histone methyltion in prventriculr nucleus through regulting of polycom repressive complex 2 Ying Jing Disserttion sumitted to the fculty of the Virgini Polytechnic Institute nd Stte University in prtil fulfillment of the requirements for the degree of Doctor of Philosophy In Animl nd Poultry Sciences Michel D. Denow Cynthi J. Denow Mrk A. Cline Pul B. Siegel Blcksurg, VA Keywords: Neuroscience, Epigenetics, Histone, Fsting, PVN, EZH2, PRC2, BDNF Copyright 2013, Ying Jing

2 Fsting lters histone methyltion in prventriculr nucleus through regulting of polycom repressive complex 2 Ying Jing ABSTRACT The developing rin is highly sensitive to environmentl influences. Unfvorle nutrition is one kind of stress tht cn cuse cute metolic disorders during the neontl period [1,2,3] nd severe diseses in lter life [4,5]. These erly life experiences occurring during heightened periods of rin plsticity help determine the lifelong structurl nd functionl spects of rin nd ehvior. In humns, for exmple, weight gin during the first week of life incresed the propensity for developing oesity severl decdes lter [5]. This susceptiility is, if not ll, relted to the dynmic reversile epigenetic imprints left on the histones [6,7,8], especilly during the prentl nd postprtum period [9]. Histones re highly dynmic nd responsive towrds environmentl stress [10,11]. Through covlent modifiction of the histone til, histones re le to direct DNA scffolding nd regulte gene expression [10,12]. Thus fr, vrious types of post trnsltionl modifictions hve een identified on vrious histones tils [12]. Among them, the methyltion nd cetyltion on lysine residue (K) 27 on histone 3 (H3) hs een tightly linked to gene repression [13,14] nd ctivtion [15], respectively. EZh2 (enhncer of zeste 2) in the polycom repressive complex 2 (PRC2) is the only methyltrnsferse tht hs een linked to ctlyze this methyltion rection. In ddition, SUZ (suppressor of zeste) nd EED (emryonic ectoderm development) re two other key proteins in PRC2 function core tht help EZH2. As previous reported, incresed H3K27 methyltion ws monitored fter fsting stress during neontl period in chicks' prventriculr nucleus (PVN). In this study, we investigted the detiled mechnism ehind chnges in H3K27 methyltion following fsting stress. After 24 hours fsting on 3 dys-of-ge (D3), chicks exhiited elevted mrna levels of PRC2 key components, including EZH2, SUZ nd EED, in the PVN on D4. Western lots confirmed this finding y showing incresed glol methyltion sttus t the H3K27 site in the PVN on D4. In ddition, until 38 dys post fsting, SUZ nd EZH2 remined inhiited. A newly identified norexigenic fctor, Brin-derived neurotrophic fctor (BDNF), ws used s n exmple of multiple hormones expressed in PVN to verify this finding. Both BDNF protein nd mrna exhiited comptile chnges to glol chnges of tri- (me3) nd di-methylted (me2) H327. Furthermore, y using chromtin immunoprecipittion ssys (ChIP), we were le to monitor the chnges of H3K27me2/me3 deposition long the Bdnf gene. Fsting significntly incresed H3K27me2/me3 s well s EZH2 t the Bdnf s promoter, trnscription strt site nd 3'-untrnslted region. These dt show tht fsting stress during the erly life period could leve epigenetic imprinting in PVN for long time. Next, we tried to understnd the function of this epigenetic imprinting in the chicks' PVN. Thus, we compred nive chicks (never fsted) to chicks tht received either single 24 hour fst on D3 or two 24 hour fst on oth D3 nd 10 dys-of-ge (D10). We found tht

3 the D3 fsted group significntly incresed the level of PRC2 key components nd its product H3K27me2/me3 compred to the nive group. However, D3 fsting nd D10 fsting together decresed the surges of H3K27me2/me3, SUZ nd EED (not EZH2) compred to the nive group. We clled this phenomenon "epigenetic memory". The Western lot, qpcr nd CHIP ssy results from BDNF ll confirmed the existence of "epigenetic memory" for PRC2. These dt suggested tht fsting stress during the erly period of rin development could leve long term epigenetic modifictions in neurons. These chnges could e eneficil to the ody, which keeps homeostsis of inner environment nd prevent mssive response to future sme stress. The EZH2 protein ws knocked down nd the H3K27 methyltion sttus chnges were monitored fter pplying the sme tretment. We first confirmed tht EZH2 ntisense oligonucleotides (5.5 ug), ut not EZH2 sirna nd rtificil cererospinl fluid (ACSF), inhiit EZH2 protein y 86 % in the PVN. Then, on D3, chicks were sujected to 24 hour fsting stress (D3-fsting) post either EZH2 ntisense or ACSF injection. The EZH2 ntisense locked the surge of oth EZH2 mrna nd H3K27 methyltion fter D3-fsting. At the sme time, BDNF exhiited elevted expression levels nd less methylted H3K27 deposition long the Bdnf gene. In ddition, we were lso interested in the chnges of "epigenetic memory" post EZH2 ntisense injection. We found tht fter EZH2 ntisense injection, chicks' PVN no longer exhiited ny "epigenetic memory" to repetitive fsting stress. While EZH2 mrna ws constntly inhiited, SUZ, EED nd H3K27me2/3 levels were unpredictle. These findings suggested tht neurons in the PVN utilized PRC2 s mjor H3K27 methyltion tool. Knockdown of EZH2 in the PRC2 impired the proper response in PVN to fsting stress nd PVN s ility to cclimte to repetitive fsting stresses. Thus, EZH2 is n importnt H3K27 methyltrnsferse inside chicken hypothlmus to mintin homeostsis. In conclusion, fsting stress during the erly life period could leve epigenetic mrkers on chromosomes of neurons in the feeding regultion center. These epigenetic mrkers will e left on chromosomes for long period of time nd hve eneficil role in keeping homeostsis when individuls fce future fsting stress gin. H3K27 methyltion is one of these epigenetic mrkers nd inhiits expression of vrious genes inside neurons. EZH2 is so fr the only detected methyltrnsferses for H3K27 tht form the PRC2. Thus EZH2 plys key function in the ody's response to fsting. Keywords: Neuroscience, Fsting, PVN, EZH2, PRC2, BDNF iii

4 ACKNOWLEDGEMENTS Time trvels y fst. Now, it is the sixth yers of my life in USA. I hve een working very hrd nd prying for this very dy to come. And now, it comes. I m officilly PhD in Neuroscience. All the ters nd sufferings during the pst men less tody nd I m very proud of myself, who could survive in this toughest period of my life till tody. Well, without ll these kinds of guidnce, encourgement nd support from mny individuls, I do not think I could chieve this honor y myself. It is only ecuse of you tht my PhD hs more fun nd rewrd. First, I would like to thnk my mjor professor, Dr. Michel Denow. He kindly ccepted me into his l on 2010 fter I otined my mster degree. He provided me lots of teching opportunities tht llow me to improve myself in teching nd presenttion skill. Dr. Denow led my reserch ck on the right trck gin nd gin during my difficult times. I ws lso llowed to explore some of my own reserch interests in his l. Additionlly, he spent countless hours proofreding my mnuscript. I feel very lessed to hve mjor professor like him. I would lso like to express my grtitude to Dr. Cynthi Denow for her generous support. I ws le to use equipments in her l nd finish this disserttion. She corrected mny mistkes in my reserches nd kept me in right direction. She spent countless hours proofreding my mnuscript. And her ig smile lwys keeps me wrm. I would like to extend my sincere pprecition to Dr. Pul Siegel for his guidnce nd humor. He rought lot of lugh into the l nd fun into the reserches. And thnk you for providing me the chnce to go to Poultry Science meeting in I m indeted to Dr. Mrk Cline for his guidnce, support nd dvice. I lerned mny reserch techniques from him. His knowledge rought mny things into my reserch. Brr Self is nother wonderful ldy I wnt to thnk. Without her perfect orgniztion, mngement nd tutoring, it is impossile for me to finish this study. And I enjoy the converstion with her in the l. I m lso grteful to Dr Elizeth Gilert. She is oth mentor nd friend to me. Her ig smile lwys ring fresh ir to me. Thnk you for llowing me to use your l equipments. I will miss you. I would like to thnk Dr. Pingwen Xu, who introduced me lots of things when I strted the l. More thnks go to the entire Animl nd Poultry Sciences Deprtment for the supporting environment. Grtitude lso goes to my friends: Lidn Zho, Guoho Xie, Yfei Zhng, Xiofei Cong, Dn Ji, Hio Zhu, Xiomei Ge, Shengcheng Su, Shui Zhng, Wei Zhng nd Ting Lu. My grtitude lso goes to Shnghi, the gret city tht rises me up. Finlly, thnks to my fntstic fmily. My Dd (Shouning Jing), my Mom (Ying Zho) nd my lovely wife (Chi-fng Tso) give me numerous supports nd keep me working hrd till tody. I miss my eloved grndm (Xiuyun Zhng) from my mother's side, who pssed wy when I ws in USA pursuing degree. You will lwys e in my hert. Much thnks to grndp iv

5 from my mother's side, who pssed wy yers go s well. Thnks to my eloved grndm, grndp nd grnd-grndm from my fther's side. Without you, my fther will not le to ccomplish this much nd I will never e le to stnd on my fther's shoulder, ccomplishing this much. I would never hve een le to mke ny ccomplishment without them. I love you ll. v

6 TABLE OF CONTENTS Title...i Astrct...ii Acknowledgements...iv Tle of Contents...v List of Tles...viii List of Figures...ix CHAPTER 1: Literture review Introduction of food intke nd feeding regultion Brief introduction to the history of identifying hypothlmus s feeding regulting center Su-Hypothlmic nucleus Brin derived neurotrophic fctor (BDNF) in feeding regultion Introduction to epigenetics Wht is epigenetics? Chromtin Structure Histone post-trnsltionl modifiction (PTM) Chromtin Remodeling...22 CHAPTER 2: Chnges of Polycom Repressive Complex 2 (PRC2) cused y neontl fsting stress protect chickens from future fsting stress Introduction Mterils nd methods Experimentl design Results Discussion...34 CHAPTER 3: EZH2 ntisense oligonucleotides inhiits EZH2 methyltrnsferse expression nd locks the fsting-tolernce cquisition Introduction Mterils nd methods...64 vi

7 3.3. Experimentl design Results Discussion...76 CHAPTER 4: Summry nd conclusion References vii

8 LISTS OF TABLES Tle 1. Potentil methyltion lysine residues on the H 3 N-terminl til nd their effect on gene expression...17 Tle 2. The primer design for different genes...38 Tle 3. Chnges in chick ody weight following D3-fsting...39 Tle 4. Chnges in ody weight of chicks following D10-fsting or D3/10-fsting...40 Tle 5. Chnges in ody weight of chicks following fsting...81 Tle 6. Chnges in ody weight of chicks following D10-fsting or D3/10-fsting...82 viii

9 LIST OF FIGURES Figure 1. Chnges in protein levels of H3, H3K27me2 nd H3K27me3 in the chick prventriculr nucleus (PVN) nd the forerin (FB) following fsting...41 Figure 2. Chnges in mrna levels of histone methyltrnsferses (HMTs) of the Polycom repressive complex 2 (PRC2) nd histone cetyltrnsferse (HAT) in the chick prventriculr nucleus (PVN) nd the forerin (FB) following fsting...43 Figure 3. Chnges in protein nd mrna levels of BDNF in the chick prventriculr nucleus (PVN) nd the forerin (FB) following fsting...45 Figure 4. Altertions in dimethyltion nd trimethyltion levels of histone H3 lysine 27 (H3K27) long the Bdnf gene in the prventriculr nucleus (PVN) following 24 hour fst on dy Figure 5. Altertions in dimethyltion nd trimethyltion levels of histone H3 lysine 27 (H3K27) long the Bdnf gene in the forerin (FB) following 24 hour fst on dy Figure 6. Altertions in EZH2 long the Bdnf gene in the PVN nd FB following 24 hour fst on dy 3. To ssess the histone modifictions present t the Bdnf gene, chromtin immunoprecipittion (ChIP) ssys were performed...51 Figure 7. Chnges in protein levels of H3, H3K27me2, H3K27me3 nd BDNF nd mrna level EZH2, EED, SUZ, CBP nd BDNF in the chick prventriculr nucleus (PVN) following D3/10-fsting...53 Figure 8. Chnges in protein levels of H3, H3K27me2, H3K27me3 nd BDNF nd mrna level EZH2, EED, SUZ, CBP nd BDNF in the chick forerin (FB) following D10-fsting or D3/10-fsting...55 Figure 9. Altertions in dimethyltion nd trimethyltion levels of histone H3 lysine 27 (H3K27) long the Bdnf gene in the prventriculr nucleus (PVN) nd forerin (FB) following D10-fsting or D3/10-fsting...57 Figure 10. Altertions in EZH2 long the Bdnf gene in the prventriculr nucleus (PVN) nd forerin (FB) following D10- fsting or D3/10-fsting...59 Figure 11. Chnges in protein nd mrna levels of GAPDH in the chick forerin (FB) following nti-gapdh sirna injection...83 Figure 12. Chnges in mrna levels of EZH2 in the chick's FB nd hypothlmus following EZH2 ntisense injection...85 Figure 13. Chnges in protein nd mrna levels of EZH2 in the chick's FB nd PVN following EZH2 ntisense injection...87 Figure 14. Chnges in protein levels of EZH2 in the chick's PVN following ACSF, EZH2 sense (1 ug/ul) nd EZH2 ntisense (1 ug/ul) injection...89 ix

10 Figure 15. Chnges in protein levels of H3, H3K27me2, H3K27me3 nd H3K27c in chick's prventriculr nucleus (PVN) nd the forerin (FB) following fsting...90 Figure 16. Chnges in mrna levels of key fctors in the Polycom repressive complex 2 (PRC2) nd histone cetyltrnsferse (HAT) in the chick prventriculr nucleus (PVN) nd the forerin (FB) following fsting nd injection...92 Figure 17. Chnges in protein nd mrna levels of BDNF in chick's PVN nd FB following fsting nd injection...94 Figure 18. Altertions in dimethyltion nd trimethyltion levels of H3K27 long the Bdnf gene in the PVN following 24 hour fst on dy 3 with injection...96 Figure 19. Altertions in dimethyltion nd trimethyltion levels of H3K27 long the Bdnf gene in the FB following 24 hour fst on dy 3 with injection...98 Figure 20. Altertions in EZH2 long the Bdnf gene in the PVN nd FB following 24 hour fst on dy 3 with injection Figure 21. Chnges in protein levels of H3, H3K27me2, H3K27me3 nd H3K27c in chick's PVN nd FB following fsting nd injection Figure 22. Chnges in mrna levels of key fctors in Polycom repressive complex 2 (PRC2) nd histone cetyltrnsferse (HAT) in the chick PVN nd FB following fsting nd injection Figure 23. Chnges in protein nd mrna levels of BDNF in chick's PVN nd FB following fsting nd injection Figure 24. Altertions in dimethyltion nd trimethyltion levels of H3K27 long the Bdnf gene in the PVN following 24 hour fst on dy 3 with injection Figure 25. Altertions in dimethyltion nd trimethyltion levels of H3K27 long the Bdnf gene in the FB following 24 hour fst on dy 3 with injection Figure 26. Altertions in EZH2 long the Bdnf gene in the PVN nd FB following 24 hour fst on dy 3 with injection x

11 CHAPTER 1: Literture review 1.1. Introduction of food intke nd feeding regultion Brief introduction to the history of identifying hypothlmus s feeding regulting center Pituitry insted of hypothlmus ws identified s feeding regulting center t erly time Food intke is regulted y the rin [16]. However, there ws time when extreme intke ws thought of s psychologicl rther thn physiologicl prolem. It ws not until 1840 tht this ide ws first chllenged. In 1840, Mohr reported out n oese womn who ecme extremely overweight nd died one yer lter [17]. An utopsy found tht rin tumor ws sitting on the sell turcic of the sphenoid one, which extended from the se of the frontl loe to the pons, which compressed the se of the rin including the optic nerve, optic chism, hypothlmus nd crus cereri. Lter, fter reviewing series of similr cses [17], scientists strted to ttriute this pthologicl ody weight increse to the destruction/interruption of the se region of the rin [18] nd Fröhlich ecme the first one to summrize tht the oesity ws directly relted to the mlfunction of pituitry glnd [17,19]. However, in 1904, Erdheim suggested opinion tht rther thn the pituitry glnd, the se region of the rin ws the reson for the oesity [20]. His rgument ws sed on his finding tht: 1) there were clinicl cses where oese ptients did not hve the pituitry glnd ffected, 2) oesity nd cromegly could occur together, nd 3) the pituitry tumor would not lwys cuse oesity. Susequent studies seemed to support Erdheim s hypothesis. For instnce, some studies showed tht pituitry glnd resection did not ffect food intke nd ody weight [21,22]. In ddition, hypothlmic destruction would result in oesity no mtter the sttus of the pituitry glnd [21]. In 1921, Biley et l. [23] ccidently found their dietes insipidus nimls, which were induced y somedil hypothlmus destruction, showed not only polyuri, ut lso hyperphgi nd incresed ody weight. From tht time, reserchers strted to focus on the hypothlmus s potentil trget for feeding regultion The competition inside hypothlmus: the "Stiety Center" nd the "Hunger Center" The Horsley Clrke pprtus ws monumentl invention for neuroscience. As introduced in the erly 1900s y Dr. Horsley [24], the Horsley Clrke pprtus ws used s stereotctic device which ensured ccess to specific re of the rin without interruption of other djcent rin regions [25]. Hetherington ws one of the first to use this pprtus in the study on feeding regultion. In 1941, Hetherington et l. [26] plced electrodes into specific rin regions guided y the Horsley Clrke pprtus nd induced destruction in different rin regions with nd without dmging the pituitry glnd. He showed tht in rts, oesity could not e induced y simple pituitry glnd dmge. In ddition, he nd his collegues lter reported tht oesity could e only induced y destroying the ventromedil hypothlmus ut not the nterior nd dorsl hypothlmus [27]. This finding ws lter confirmed y other scientists who conducted the sme procedures on cts [28] nd monkies [29]. 1

12 After this finding, more nd more focus ws plced on the hypothlmic re nd mny reserchers investigted different prts of the hypothlmus to find the reltionship etween su-hypothlmic nuclei, feeding, nd energy homeostsis. Finlly, fter the finding of stiety center [30] nd hunger center [28], the model of two-center theory ws introduced in the 1950s. In 1950, Kenndy reported his findings tht rts showed pprent hyperphgi fter destroying either the lterl prt of the tuerl region or the region lying ventrl-lterl to ventrl-medil nuclei within the hypothlmus, [30]. He elieved tht there ws stiety mechnism in these res of the hypothlmus which inhiited food-intke ehvior in nimls. After dmge, or with the incresing ge, the stiety mechnism ecme disrupted or wekened nd nimls strted to show incresed ppetite, resulting in dipose deposition nd oesity. Thus, he proposed the existence of the stiety center in the hypothlmus. This hypothesis ws supported y mny susequent studies. For exmple, Broeck et l. [17] showed tht ilterlly dmging the ventromedil hypothlmus induced hyperphgi in rts. Lter, Wyrwick et l. [31] conducted n experiment y putting electrodes into the ventromedil hypothlmus of sheep, which gretly inhiited food intke of hungry sheep. Only one yer fter Kenndy s stiety center hypothesis, in 1951, Annd nd Broeck purposed the "hunger center" hypothesis, which ws locted t the lterl side of the lterl hypothlmus. They reported tht fter ilterl, ut not unilterl, destruction of the "hunger center", nimls would stop eting completely (phgi) nd showed ody weight loss [28]. Additionlly, the "hunger center" could completely overrule the effect of ilterl "stiety center destruction, which cused opertion induced oese niml to e phgi nd weight loss. This hypothesis ws lter studied in sheep [31] nd cts [32]. All of them reported the sme phgi in these operted nimls Estlishing the function of hypothlmus in energy lnce nd oesity Bsed on the two center theory, reserchers were trying to find the mechnisms within the hypothlmus involved with feeding nd energy regultion. A lot of hypotheses were proposed. For exmple, some ppers suggested tht oesity ws induced y long-term hyperphgi which resulted in dipose deposition [17]. Some ppers ttriuted this to significntly reduced ctivity nd reluctnt running in nimls tht received either whole or ventromedil hypothlmus destruction [26,33]. The most promising hypothesis elieved tht there existed certin connection etween the centrl nerve system (hypothlmus) nd peripherl digestive orgns. As the informtion integrtion center, the hypothlmus mixed the signls from the peripherl digestive orgns nd mde finl decision to et or stop eting. Thus, there must e signl loop where signls were sent ck nd forth etween peripherl sites nd the hypothlmus to lnce ody energy homeostsis. So wht is the signl? Between 1940s to 1950s, this question ws discussed intensely nd three hypotheses were dopted, which postulted tht ody temperture [34], lood sugr level [35] nd lipid mount [36] were the stiety signls which served s the messenger to let the hypothlmus control food intke. 2

13 In 1948, Broeck first hypothesized tht food intke ws regulted y ody temperture [34]. He proposed tht t high environmentl temperture, nimls would stop eting. He elieved tht t high environmentl temperture, it would e very difficult for nimls to reduce their ody temperture. Food intke would cuse therml effect, which led to further incresing in ody temperture nd hrmful stress. In contrst, t low environmentl tempertures, food intke would help to rise ody temperture, thus preventing hypothermi nd cold stress. Broeck showed tht t 65 to 76 F, dult mle rts hd incresed food intke. However, t >92 F dult mle rts ppered hypophgi/phgi, pyrexil nd lost ody weight [34]. Thus he elieved tht food intke ws potentil mechnism of temperture regultion to void hrmful stress. In the erly 1940s, the thermotxic center ws lso found in the hypothlmus [37]. Thus the integrted hypothesis ws tht the environmentl temperture ws feeding signl, which ws sent nd processed in the hypothlmus nd then directed ltertions in food intke. However, this hypothesis ws discredited y lter reserch findings [36,38]. Kennedy showed tht the rts in Broeck s study hd lost more thn 30g in ody weight overnight fter het tretment, which ws more thn doule thn norml food strvtion could cuse [36]. As these rts were rised in high environmentl tempertures, Kennedy showed tht dehydrtion ws the mjor reson for ody weight loss. Incresed ody tissue ctolism under hyperthermi lso plyed n importnt role. Once the rts were cclimted to the high environmentl temperture, hypothlmic dmged rts gin weight nd ecme oese, just like their counterprts t norml room temperture. Myer lso mentioned tht fter dministrtion of therml regultion gents (thyroxine s hypertherml nd thyrodectomy nd thiourcil for hypotherml), results were contrdictory to Broeck s hypothesis [38]. The glucostsic-food regultion hypothesis ws proposed in 1953 y Myer [38]. In this hypothesis, lood glucose level ws elieved to hve key influence on food intke. The hypothlmic feeding center (including oth hunger nd stiety center) ws elieved to hve glucoreceptors, which could monitor rel-time lood glucose levels in the ody. If the lood glucose ws in the norml rnge, nimls would not feed. If there ws drop of the lood glucose level, n fferent signl would ctivte the hypothlmus nd the hunger center would e stimulted, which initited the feeding process. To support this hypothesis, Myer et l. showed tht glucose or fructose dministrtion, which mde rts hyperglycemic, inhiited feed intke significntly [39]. In ddition, the dministrtion of smll mounts of insulin induced hypoglycemi (the glucose level remined in the physiologicl rnge ut reltively lower thn the norml), which resulted in incresed feeding. Furthermore, Myer et l. [39] lso used specil strin of rts with lloxn induced dietes. Alloxn is clssicl type I dietes inducer [40]. It is known s n nlogue of glucose, which is selectively stored in the et cell in the rt pncres vi the GLUT2 trnsporter on its memrne [41,42]. The ccumultion of lloxn cuses necrosis of et-cells in the pncres nd thus induced hyperglycemi, which is similr to the insulin-dependent type I dietes mellitus [43]. Myer nd Btes [39] showed tht fter injection of glucose, the lloxn-dietic rts remined hyperglycemic for long time nd reduced their food intke. The glucostsic-food 3

14 regultion hypothesis gve good foundtion for short-term food intke regultion. It helps explin the how the ody regultes food intke in ech mel. However, it cnnot explin long-term ody weight regultion. The third hypothesis ws introduced y Kennedy, who elieved tht the lipids controlled feeding sttus [36]. He found tht young nimls regulted food intke to keep their ody ft in n lmost fixed percentge [30,36]. In certin situtions, when nimls were put in hot or cold environments, or during lcttion, the nimls mintined reltively fixed ody weight s well [36]. In ddition, even those nimls with dmged hypothlmus would mintined the ody weight t certin level [35]. Kennedy proposed the ide tht the dipose deposit in the niml's ody would regulte the food intke [36]. By sending out n inhiitory signl from the dipose tissue, the niml ws le to regulte feeding nd thus kept their ody weight reltively stle. During strvtion, the inhiitory signl gets weker, which motivtes the niml to et. However, Kennedy ws not le to determine the exct signl hormone tht ws used in this loop. Kennedy s ide ws lter supported y the finding of Hervey. In 1959, Hervey did priosis experiment, which he surgiclly pired two rts together [44]. This surgicl fusion of the two rts llowed 1% chnge in plsm per minute. Then he dmged the ventromedil hypothlmus of one of the pir. It would e expected tht the two nimls would grow oese together ecuse the hyperphgi from the operted rt would trnsfer the excessive mount of nutrition to the norml rt. On the contrry, while the operted rt ws oese s expected, the un-operted niml showed pprently decresed ppetite, weight loss, smller liver, less food in the GI trct (found during utopsy) nd eventully strved to deth. Only n identicl destruction of the hypothlmus of the norml rt could chnge the ppetite of the norml rt nd mke oth nimls oese. He even proposed tht some signls from the operted rts were sent to the hypothlmic food intke controlling center of the norml rts, which ffect the norml feeding regultion in the norml rt. Compred to the glucosttic regultion, which focused more on temporry food intke regultion, the lipostsis hypothesis explined how the hypothlmus could regulte the ody weight for long-term Su-Hypothlmic nucleus The hypothlmus is crucil structure in the rin, which is involved in the mny physiologicl roles of regultion, including feeding. Bsed on regionl destruction nd injection studies, five su-hypothlmic nuclei hve een shown to ply pivotl roles in feeding regultion. They re the rcute nucleus (ARC), ventromedil nucleus (VMN), dorsomedil nucleus (DMN), prventriculr nucleus (PVN), nd lterl hypothlmus (LH) Arcute Nucleus (ARC) The ARC is locted t the ottom of the hypothlmus nd is djcent to the medin eminence. The first study involving ARC function in feeding ws conducted in 1969, when Olney injected monosodium glutmte (MSG) sucutneously into 4

15 neontl mice [45]. The result showed tht MSG dministrtion induced pprent neuronl degenertion in severl prts of the rin, ut specificlly in the ARC of the hypothlmus. The treted nimls showed mrked oesity ut not hyperphgi. Lter experiments proved these results nd lso showed tht the effect of MSG ws species distinctive [46,47]. As mentioned ove, there re certin regions of the rin tht lck the lood rin rrier (BBB), which llow free ccess of lrge sustnces to pss freely into the CSF. The ARC receives hormonl signls directly from the peripherl circulting serum [48,49]. There re two mjor groups of neurons in the ARC, the Neuropeptide Y (NPY) -Agouti-relted peptide (AgRP) nd Amphetmine-Regulted Trnscript (CART) -Proopiomelnocortin (POMC) [50,51]. These two neuron popultions express orexigenic nd norexigenic signls, respectively, which regulte energy lnce in the ody. Insted of working seprtely, they re connected y γ-minoutyric cid (GABA) contining fiers from the NPY-AgRP neurons [52]. GABA is n inhiitory neurotrnsmitter nd cuses CART-POMC neuron inhiition upon NPY-AgRP neuron ctivtion [53] nd results in n incresed feeding ehvior. In ddition, oth NPY-AgRP [54] nd CART-POMC neuron [55] popultions hd leptin receptors on the memrne [56], which indictes their inter-connection with the leptin system. Thus, ARC is thought to e the most importnt nucleus involved in energy lnce regultion Ventromedil nucleus (VMN) VMN hs long een recognized s n importnt nucleus involved in feeding nd energy homeostsis regultion. Erly studies tht destroyed the VMN ll reported hyperphgi nd oesity [17,57,58] nd VMN electric ctivtion resulted in feeding suppression [31,59]. Mny injection studies lso confirmed the effectiveness of mny norexigenic fctors fter VMN microinjection, including histmine [60], serotonin [61], urocortin [62], GABA ntgonist [63], cholecystokinin (CCK) [64], leptin [65] nd insulin ntiody [66]. On the other hnd, thyroid hormone [67], GABA [63], norepinephrine [68] nd NPY [69,70] exhiited n opposite effect when microinjected into the VMN. The communiction etween VMN nd other nuclei in the hypothlmus confirmed its role in feeding nd energy homeostsis regultion. For exmple, studies using retrogrde trcer found smll mounts of fferent fiers to VMN from LHA [71,72], DMN nd PVN [73]. VMN lso projected to the LHA [73] nd PVN [74] in the hypothlmus nd the nucleus trctus solitrii (NTS) in the rin stem [74]. Brin-derived neurotrophic fctor (BDNF) is highly expressed in the VMN. The topogrphy of the BDNF receptor, TrkB, demonstrted highly expressed level not only in VMN, ut lso in other hypothlmic nuclei, which, however, did not include orexin nd MCH neurons in the LHA nor CART nd NPY neurons in the ARC [75]. This suggested n sence of innervtions of BDNF(+) projection from VMN to the LHA nd ARC. Contrrily, immunohistochemicl studies reveled the outons of γ-melnocyte-stimulting hormones (γ-msh) nd AgRP fiers from the ARC innervted the BDNF expression neurons in the VMN [75,76]. These results suggested unidirectionl innervtion to the VMN BDNF(+) neurons from the hypothlmic ARC. However, it ws not until recently tht one group using lser scnning 5

16 photo-stimultion method, reported tht the VMN did project to ARC [77]. Surprisingly, only CART/POMC neurons received strong excittory innervtions from VMN while the NPY/AgRP neurons did not. Insted, NPY/AgRP received wek inhiitory fiers within the ARC. This excittory signl to CART/POMC neurons from VMN could e diminished when the niml ws fsted [77]. Compred to other regions of the hypothlmus, the VMN contins much higher level of BDNF, memer of neurotrophin fmily [75,78,79], nd cholecystokinin (CCK) [80]. Becuse of the importnt role of VMN in feeding nd energy homeostsis regultion, the high mount of BDNF nd CCK were suspected to e the mjor stiety signls of VMN nd crried out n inhiitory role in feeding Dorsomedil nucleus (DMN) The first ever pper demonstrting the function of the DMN ws in 1943, when group put pir of electrodes into the DMN of cts nd stimulted the neurons in this re. The stimultion of the DMN in cts resulted in hyperphgi [81]. This study ws repeted in other nimls nd yielded consistent results [82,83]. While some studies ctivted the DMN re, others were trying to see the results of DMN lesion (DMNL). Electrolytic lesions [84], iotenic cid injection [85] nd surgicl cutting [86] t the DMN generted DMNL rts, who exhiited inhiited feeding nd slower growth s well. Interestingly, unlike other niml with injured hypothlmic nuclei, the DMNL nimls exhiited norml ody ft composition nd plsm ftty cid levels [87]. More surprisingly, there ws no interruption in the nimls' nolic nd growth-promoting hormone levels (insulin, thyroxine, triiodothyronine (T3), growth hormone (GH), nd somtomedin (SM)) [87,88]. However, these DMNL nimls hd decresed ody weight nd disrupted plsm corticosterone [89] nd prolctin [90] levels, which did not ffect food intke [81]. The decresed ody weight seemed to e pproprited to their decresed feeding ehvior [87]. Bsed on these findings, the function of the DMN seemed to e it confusing compred to the direct effect from other regions of hypothlmus. However, cholecystokinin (CCK) is n ctive plyer in DMN, which regultes the feeding nd energy homeostsis. Cholecystokinin (CCK) ws first discovered in the duodenum in 1930 [91] nd ws recognized y its stimultory involvement in the secretion of pncretic enzymes in gll ldder contrction nd inhiition GI trct emptying [92]. Now, CCK is recognized s n importnt neuropeptide in short-term feeding nd energy homeostsis oth in centrl nd peripherl tissues. In peripherl tissue, the CCK precursor ws secreted in GI incresingly fter ft or protein ut not crohydrte ingestion, suggesting its function in energy homeostsis [93,94]. In the CNS, CCK hd een clssified s the most undnt rin-gut neuropeptide [95] nd detected in severl mmmlin rins, such s rt [96] nd humn [97]. The extremely high mount of CCK in res, such s hippocmpus, olfctory ul, septum, mygdl, hypothlmus, etc, suggested tht CCK ws involved in vrious roles in CNS function. The reltively high mount of CCK in the hypothlmus, together with its role in peripherl 6

17 digesting tissues, indicted tht CCK my ply pivotl role in feeding nd energy homeostsis regultion t oth the CNS nd peripherl levels. Different ppers hd reported CCK s concentrtion in different res of the hypothlmus, which ws lmost consistent. The only discrepncy remined ws which nucleus/res contined the highest CCK concentrtion. One pper set the order from high to low s VMN, DMN, PVN nd ARC [80] while the other reported the highest mount of CCK expressing neurons ws gthering in the DMN [98]. Lter, y using rdioctive 125 I-CCK-8, the CCK receptors inside the hypothlmus were pinpointed, including VMN with the highest density, followed y the DMN [99]. Both results sed on the topogrphy of CCK nd the CCK receptor suggested the similr importnce of the VMN nd DMN in the CCK signling pthwy. In nother study, the reserchers microinjected CCK-8 into different nuclei in the rin, including six regions in the hypothlmus nd seven regions out of the hypothlmus [64]. While lmost ll res in the hypothlmus ffected feeding fter CCK microinjection, the NTS ws the only outer hypothlmic re tht responded to microinjection. Among those res, the DMN exhiited the most prominent feeding inhiition [64]. This implied tht lthough the concentrtions of CCK in the VMN nd DMN were similr, the DMN could e the effective center of the CCK signl pthwy. Susequent groups studied the distriution of the CCK receptor y using CCK-A receptor specific rdioctive gonist. CCK-A receptor ws known s the mjor conductor of CCK for its feeding regultion (referring to the following prgrph). They found tht the CCK-A receptor hd the highest concentrtion in the DMN in the hypothlmus nd NTS in the hinder rin [100]. On the contrry, the PVN contined much lower mount of CCK-A receptor nd there ws none in the VMN [100]. In ddition, VMN lesions did not olished the feeding inhiition induced y CCK IP injection in rt [101]. Comining the result of CCK concentrtion, CCK-A receptor distriution, the CCK-8 microinjection s effect nd lesion study, the DMN ws elieved to e the effective center for CCK in the CNS Lterl hypothlmic re (LHA) The LHA ws first relted to feeding nd energy homeostsis regultion in 1951 [28]. As previously mentioned, fter ilterl, ut not unilterl, destruction of LHA, nimls would ecome phgic, dipsic nd demonstrted significnt ody weight loss [28]. Surprisingly, even VMN destruction could not reverse this effect. More thn 35 different types of neurons hve een identified in the LHA, indicting its role in oth stiety nd hunger signl integrtion [102]. This ide is lso supported y the mtrix of fferent/efferent fiers connected to the LHA. The LHA neurons project into the VMN, DMN nd ARC in the hypothlmus nd mny other regions of the rin nd lso received the innervtion fiers from the limic system nd rin stem [103]. Three kinds of neuropeptides hve een detected in the LHA, melnin-concentrting hormone (MCH), orexin, nd excittory mino cid (EAA) glutmte. Unlike NPY- AgRP nd CART-POMC neurons tht co-express two neuropeptides in the ARC, MCH nd orexin were distinct from ech other [104]. 7

18 Prventriculr nucleus (PVN) Histology studies showed tht the PVN receives innervtion fiers from lmost ll res of the hypothlmus [105], which implies its pivotl role in physiologicl function regultion. Nerly ll known orexigenic fctors incresed feeding when microinjected into the PVN [50]. These orexigenic fctors included NPY [106], AgRP [107], GABA A receptor gonist (GABA nd muscimol) [108], glnin [109,110], nd orexin [111]. Feeding ehvior ws inhiited when the PVN ws loclly microinjected with norexigenic fctors, including Neuropeptide S (NPS) [112], α-msh [107], nicotine [113], nd corticotropin-relesing fctor (CRF) -2 receptor gonist (CRF nd urocortin-1) [114,115]. This informtion implies tht feeding nd energy homeostsis regultion must e one of the most importnt physiologicl roles tht PVN is involved in. Bsed on the nucleus size, the neurons in the PVN re divided into two groups, mgnocellulr nd prvocellulr neurons [116]. Although numerous neuropeptides hve een detected in the PVN, three of them were suggested s signls involved in feeding nd energy homeostsis regultion: oxytocin, corticotropin-relesing fctor (CRF) nd thyrotropin-relesing hormone (TRH) [56]. In generl, the mgnocellulr neurons express oxytocin, ut not vsopressin, inside the cell ody nd trnsport it through their xons to the posterior prt of the pituitry glnd [117]. On the other hnd, the prvocellulr neurons encode high level of TRH [118,119] nd CRF [120]. However, certin levels of oxytocin (OT) re lso expressed in the prvocellulr neurons, which hve different projections compred to the OT-expressing mgnocellulr neurons [121,122]. The xon of these prvocellulr neurons project to the medin eminence nd relese TRH, CRH nd OT into the systemic circultion through the pituitry glnd [119,123,124] Brin derived neurotrophic fctor (BDNF) in feeding regultion Before introducing the role of BDNF in feeding regultion, generl introduction of neurotrophins is presented. Neurotrophins re fmily of neuron growth fctors, which re cple of mediting mny physiologicl functions, such s neuron prolifertion, differentition, growth nd interction [125]. Memers of this fmily include nerve growth fctor (NGF) [126], rin-derived neurotrophic fctor (BDNF) [127], neurotrophin-3 (NT-3) [128], neurotrophin-4 nd -5 (NT-4/5) [129], neurotrophin-6 [130] nd neurotrophin-7 [131,132]. Most of these neurotrophic fctors were initilly discovered only in the rin which suggested limited physiologicl functions such s neuron survivl [133]. However, more recent studies reveled other functions of the neurotrophin fmily, such s feeding nd energy homeostsis regultion [134]. In this disserttion, the discussion would e focused on BDNF Brief introduction of BDNF protein discovery, function nd distriution 8

19 The discovery of BDNF ws first reported in 1982 [127]. In the study, the uthors reported smll protein (12.3 KD), which ws extrcted nd purified from the pig rin. Then this protein ws tested in n in vitro neuron culture ssy nd the results showed tht it incresed neuron survivl s much s NGF. The uthors nmed the protein BDNF. Susequent studies confirmed BDNF s effect on neuron survivl oth in vitro [135] nd in vivo [136]. Both the gene nd the mino cid sequence of BDNF were reported in 1989 y the sme group of scientists who discovered it [133]. The sequence lignment indicted similrity of mino cid sequence etween BDNF nd NGF. In ddition, they lso locted BDNF mrna expression y Northern lot nd showed tht BDNF ws exclusively expressed in neurons ut not other peripherl tissues. However, lter studies using more dvnced PCR method found BDNF expression in peripherl tissues, such s plcent [137], hert, muscle [138], liver nd spleen [78] Brief introduction of BDNF receptors As neuropeptide, the physiologicl effect of BDNF ws conducted vi its specific neurotrophin receptors. In 1988, pper reported tht leled 125 I-BDNF ws le to ind to cell memrne in n in vitro neurl culture [139], suggesting the existence of BDNF cell memrne receptor. Bsed on the dissocition constnt detected from kinetic ssy, the receptors could e su-divided into two groups, high-ffinity receptors (dissocition rte (K d ) = 1.7 x M) nd low-ffinity receptors (K d = 1.3 x 10-9 M) [139,140,141]. Bsed on their structure, high nd low ffinity neurotrophin receptors were defined s memers of either the tropomyosin-relted kinses (Trk) receptor fmily or the tumor necrosis fctor (TNF) receptor fmily, respectively [140]. The low ffinity neurotrophin receptor, nmed P75, ws n unspecific neurotrophin receptor nd cple of intercting with ll neurotrophins [142] nd tumor necrosis fctor (TNF) [143]. The Trk sutype, high ffinity neurotrophin receptors, showed high selectivity to their specific neurotrophins [142]. According to phrmcologicl studies, the Trk receptor ws further clssed into TrkA, TrkB nd TrkC, which were defined s the specific receptors for NGF (nd NT-6 nd NT-7), BDNF (nd NT-4/5) nd NT-3, respectively [140,141]. But this did not men tht there ws no interction etween non-specific Trk receptors nd neurotrophins. Insted, different neurotrophins could still ind to their non-specific Trk receptor ut the dissocition rte (K d vlue) ws much higher thn the specific neurotrophin [144]. For exmple, in order to decrese the inding rte of BDNF to TrkB y 50%, NGF needed to increse y 1000-fold concentrtion, which ws enough to inhiit the inding etween BDNF nd the P75 receptor [145]. P75 nd Trk receptors re oth cell memrne surfce receptors nd, in most cses, re expressed on the sme cell memrne [146]. Interestingly, the P75 nd Trk receptors exhiit opposite iologicl effects, which regulte the sttus of certin cells [140,147,148]. While the Trk receptor is the gte of the neurotrophic effect, the P75 receptor shred the "deth domins" of the tumor necrosis fctor receptor fmily [149] nd ws more involved in cell deth [150]. This suggests the different intrcellulr signling pthwys these two receptors utilize. This ide ws supported y mny studies nd, to dte, three signling 9

20 pthwys were discovered for Trk [140] nd two signling pthwys for P75 receptor [151,152], respectively. Trk conducts its iologicl effect through Rt srcom (RAS), Phosphtidylinositide 3-kinses (PI-3K) nd Phosphoinositide phospholipse C (PLC)-γ1 pthwys [140]. On the other hnd, P75 ctivtion cuses cermide production, c-jun N-terminl kinses (JNK) level incresing nd nucler fctor kpp-light-chin-enhncer of ctivted B cells (NF-κB) ctivtion in the cell [153]. Interestingly, the P75 signling pthwy could chnge the ffinity of different Trk receptors to its cognte lignd, BDNF [146,154], suggesting P75 could lso e regultor in BDNF s signl pthwy BNDF hs importnt role in feeding regultion in hypothlmus In the rin, BDNF is widely expressed in vrious regions nd the highest concentrtion ws detected in the hippocmpus nd hypothlmus [78,155]. As the importnce of the hypothlmus in feeding nd energy homeostsis regultion ws uncovered, the involvement of BDNF in this cpcity ws hypothesized. The involvement of BDNF in feeding regultion ws not unveiled until 1992 when Lpchk nd Hefti ccidently discovered the reltionship etween BDNF nd ody weight in neuron regenertion relted reserch [156]. Initilly, they were trying to rescue the hippocmpl cholinergic dysfunction fter prtil fimril trnsction y chronic BDNF ICV dministrtion. The BDNF filed to sve neurons, ut the BDNF dministrted mice ll showed decresed ody weight [156]. Simultneously, other memers of the neurotrophin fmily, such s NGF, were suggested s feeding regultors y showing decresed food intke nd weight loss fter ppliction [157,158]. Thus, more reserchers strted to look into the role of BDNF in feeding nd energy homeostsis regultion. In the hypothlmus, the highest BDNF concentrtion is detected in the VMN [75,79], which ntomiclly supports the role of BDNF s n importnt fctor involved in feeding nd energy homeostsis regultion. Almost ll studies supported the ide tht BDNF ws n norexigenic fctor. For exmple, fter 48 hours fsting, BDNF levels in the VMN [75] nd dorsl vgl complex (DVC) of the hinder rin (where NTS locted) [159] ws significntly lowered, suggesting n incresed feeding fter fsting. When nimls were re-fed, the BDNF level went ck to norml. Acute microinjection of BDNF in the VMN [160], DVC [159] nd PVN [161] inhiited the feeding nd incresed energy expenditure in test nimls. Following chronic ICV injection of BDNF, rts showed dose-dependent food intke suppression nd ody weight loss [162]. However, when pretreting nimls with TrkB-Fc fusion proteins, n ntgonist of TrkB, the BDNF ICV dministrtion no longer inhiited feeding [160]. This suggested tht BDNF used the TrkB receptor pthwy to perform its norexignic effect. In humns, recent study lso supports the ide of BNDF s n norexigenic fctor. In femle with either ulimi nervos or norexi nervos, serum BDNF levels were significntly lowered [163], with the norexi nervos ptients hving the lowest [163]. Just like other studies of orexigenic nd norexigenic fctors, these utilizing knockout muttion rodent strins offered the most direct insight into BDNF s effect. However, s n importnt peptide involved in neurl development, BDNF doule KO (BDNF / ) nimls were not le to e otined for reltively long-term oesity studies ecuse of their erly postntl mortlity 10

21 [164]. In 2000, BDNF heterozygous mice (with only one functionl BDNF llele) were creted with significntly lowered BDNF protein levels in the hypothlmus, including PVN, LHA, VMN, ARC nd DMN, when compred to the wild type (WT) [79]. Bsed on ody weight, two phenotypes of BDNF heterozygous mice were sudivided into ft (FBH) or non-ft (NBH). Interestingly, the FBH strin showed lte-onset ody weight gin, which ws significntly higher thn oth the WT nd NBH, while there ws no ody weight difference etween WT nd NBH. In ddition, FBH strin lso hd more dipose tissue nd hd endocrine normlities, including hyperleptinemi nd hyperinsulinemi, ut norml serum glucose nd corticosterone level. Additionlly, BDNF or NT4/5 ppliction into the third ventricle could trnsiently ut noticely inhiit feeding nd lower ody weight. The study lso reported tht oth leptin receptors nd other feeding relted neuropeptides, such s NPY nd CART, remined t norml level in the hypothlmus, which suggested unique pthwy of BDNF in feeding nd energy homeostsis regultion. Prt of the results from BDNF heterozygous mice ws confirmed y humn cse. In 2006, one cse report descrie n 8-yer-old girl who exhiited severe hyperphgi nd oesity [165]. The genotyping showed chromosoml inversion t 46,XX,inv(11)(p13p15.3), which ws the re coding the BDNF gene, nd resulted in BDNF heterozygous humn model. A more recent study lso reported the selective deletion of BDNF in the VMN nd DMN together in dult mice exhiited hyperphgi nd oesity strting t week 5 nd 7, respectively [166]. The role of TrkB in feeding nd energy homeostsis ws confirmed in lter studies y showing tht TrkB mlfunction mimics the effect of BDNF knock down. The TrkB doule KO (TrkB -/- ) ws s ftl s tht of BDNF KO. Thus, y using trnsgenic method, one group creted mouse strin tht exhiited mutnt TrkB receptor (functionl KO). These mice exhiited distinctively elevted food intke nd longer feeding ehvior strting t week 5 [75]. They lso hd pprent oesity t the mturity period nd incresed liner length [75]. As nother BDNF pthwy conductor, the P75 receptor lso plys role in feeding nd energy homeostsis regultion. In the "New Zelnd oese" mice strin, there ws n incresed P75 receptor mrna level in the hypothlmus [167]. Additionlly, compred to the mortlity of TrkB doule KO (TrkB -/- ), the P75 doule KO (P75 - /- ) mice were le to survive, ut exhiited lower ody weight strting t week 4 [168]. The food deprivtion of these nimls cused significntly decresed serum insulin nd leptin. Another study using P75 -/- mice confirmed the normlly incresed ltency to feed fter eing moved to new environment [169]. However in oese rodents, such s the o/o strin, the tumor necrosis fctors (TNF) level ws lso incresed [170]. This suggested the interction etween P75 with TNF rther thn BDNF, which cused feeding inhiition nd ody weight loss. The BDNF role in feeding nd energy homeostsis regultion ws lso connected to other signling pthwy, such s leptin, insulin, POMC, NPY, nd AgRP [134]. The receptors of leptin [171], insulin [172], orexin [173] nd MCH [174] re presented in the VMN. In norml mice, cute intrvenous leptin dministrtion induced up-regulted expression of BDNF in the VMN [175]. In ddition, one group found tht in n in vitro rel-time imging experiment, leptin dministrtion ctivted the PI- 3K pthwy in the CART/POMC neurons in the ARC while leptin removl ctivted the sme pthwy in NPY/AgRP neurons 11

22 [176]. Interestingly, nother dipose signl, insulin, elicited the sme result s leptin on CART/POMC neurons ut opposite results on NPY/AgRP neurons [176]. Becuse PI-3K ws one of the three mjor signling pthwys of the BDNF receptor, TrkB, PI-3K could e common pthwy for leptin, insulin nd BDNF. In two leptin resistnt mice strins, BDNF dministrtion corrected not only the hyperphgi nd oesity, ut lso glucose homeostsis [177]. Blood tests reveled tht BDNF ppliction noticely decresed serum insulin nd leptin level [177]. An immunohistochemistry study reveled the outons of γ-msh nd AgRP fiers from the ARC innervted the BDNF expression neurons in the VMN [75]. Both pro-opiomelnocortin (POMC) nd melnocortin receptor 4 (MC4R) doule KO nd gouti over expression mice exhiited decresed BDNF mrna levels in the VMN [75]. In ddition, s mentioned ove, food deprivtion could significntly decrese the BDNF level in the VMH. This phenomenon ws prtilly relieved y ICV MC4 receptor gonist (melnotn II) ICV ppliction [75]. NPY micro-injection into the VMN incresed feeding, which ws lso le to e reversed y BDNF co-ppliction [160]. Microinjection of BDNF in inhiited NPY mrna in the ARC nd PVN [161]. Additionlly, ICV infusion of BDNF incresed CRH nd urocortin levels in the PVN nd promoted feeding [178]. However, CRH nd urocortin receptor ntgonist ppliction olished the hyperphgi BDNF gene construction Presently, the humn BDNF gene is encrypted nd covers ~70 k regions on chromosome 11, which contins eleven exons nd nine of its promoters re tissue-specific [179]. The BDNF precursor (pro-bdnf) used trns-splicing method to splice different 5 non-coding exons with the 3 exon, which encoded the pro-bdnf mrna. Bsed on different tissue loctions or regultion requirements, different comintions of mrna from the 5 non-coding exons were fused to the pro-bdnf mrna, which generted t lest 17 trnscripts [179]. For exmple, I, II nd III 5 non-coding exons were undnt in the rin while IV ws highly expressed in the lung nd hert [180]. After kinic cid-induced seizure, the mrna of the I, II nd III 5 noncoding exons incresed significntly while there ws no chnge in the level of IV s level [180]. The 3 exon product of the BDNF gene, the pro-bdnf (~32 KD), ws lterntively spliced into either mture BDNF mrna (14 KD) or minor truncted form (28 KD) [134,181]. After trnsltion nd protese clevge [134], the BDNF protein ws stored in the vesicles in the somtodentritic domin of the neurons nd relesed upon neuron excittion [181]. For rodents, oth mice nd rts hd similr BDNF gene rrngement to tht of humn. Their pro-bdnf ws coded y one protein coding 3 exon nd eight regultory 5 exons [182]. Initilly, in the vin niml, the Bdnf gene ws elieved to e formed with no 5 non-coding exons ut only one 3 exon, whose trnscription ws regulted y the methyltion sttus of the stnd-lone C-phosphte-G (CpG) islnd upstrem of its promoter [183]. One recent study reported three 5 non-coding exons in the Bdnf gene in chicken [184]. These three 5 noncoding exons encoded three mrna trnscripts (BDNF1, BDNF2 nd BDNF3), which were spliced to the pro-bdnf mrna coded y the BDNF 3 exon. BDNF1 nd BDNF2 were found specificlly in rin (mong eight tissues tested y the uthors). 12

23 In the CNS, BDNF1 nd BDNF2 were detected in the hypothlmus while BDNF3 ws not [184]. This new finding reveled tht the chicken BDNF gene structure ws similr to mmmlin nimls Introduction to epigenetics Wht is epigenetics? The word epigenetics ws first coined y Conrd H. Wddington ( ). He defined epigenetics s the rnch of iology which studies the cusl interctions etween genes nd their products, which ring the phenotype into eing (Wddington, 1942). The prefix epi- mens ove nd genesis indictes the unit of heredity or gene, which in rod sense is like ridge connecting n orgnism s genotype nd phenotype thus chnging the finl phenotypic outcome without chnging the ctul DNA sequence. In other words, epigenetics is term used to descrie the events tht cnnot e explined y genes lone (Golderg et., 2007). In 1957, Wddington descried wht he clled the epigenetic lndscpe, in which he portryed fertilized emryo s ll sitting on the slope of vlley. The ll during its development cn t vrious points tke specific trjectories down the slope, which ultimtely could led to different phenotypic outcomes. It ws good visul representtion to illustrte tht cellulr differentition cn e governed y the cell s physicl environment in ddition to its genetic predisposition. A more modern definition of epigenetics is the study of the stle nd idelly heritle chnges tht re not relted to chnges in the ctul gene sequence, ut tht result in prticulr cellulr phenotype [185]. Gene expression in eukryotic orgnisms is regulted y severl epigenetic mechnisms including DNA methyltion nd histone modifictions. In this proposl, rief discussion of these underlying mechnisms eginning with some ckground on gene structure will e ddressed Chromtin Structure The humn genome contins out 6 illion se pirs (p) of DNA, which, if stretched out, would e close to 2 meters long [186]. So how is the 2-meters long DNA pckged into the nucleus of cell tht is only severl micrometers in dimeter? In cells, the DNA nd its ssocited proteins ind together nd re known s chromtin, which, fter nuclese tretment, demonstrted the ed-on- string s seen under the electron microscope (Olins et l., 1975; Oudet et l., 1975). This string is doule-strnded DNA nd pproximtely 2 nm in dimeter. The eds re the fundmentl units of chromtin, clled nucleosomes, which re 10 nm in dimeter. The nucleosome is the comintion of oth DNA nd the nucleosome core prticle (NCP). The NCP is pproximtely 206 K in moleculr weight, 6.5 nm in dimeter nd consists of disk-shped protein polymers, which is wrpped 1.65 times y 146 p of tight-superhelicl DNA in left-hnded mode [187,188,189]. Inside the NCP, four types of core cnonicl histones hve een identified, which re H 2A, H 2B, H 3 nd H 4. At physiologicl condition, (H 3 -H 4 )2 tetrmer loctes in the middle while one H 2A -H 2B dimer stys on ech side [190]. The nucleosomes re then connected y the linker DNA t different lengths with linker histones. This llows the chromtin fiers to form higher-order structure, 13

24 which increses pcking efficiency nd only unpcks chromtin prtilly during gene trnscription. The chrcters presented ove re from the humn α-stellite DNA in the nucleosome with cnonicl histone octmer. Different orgnisms my contin different lengths of DNA in nucleosome, vrying from 100 to 170 p [191]. One recent review pper mentioned the discovery of sturted nucleosome form, which could ccommodte DNA up to 173 p [191]. In ech nucleosome, there exists two kinds of onds etween the NCP nd wrpping DNA, including direct inding nd wter-ridge interction, which give pproximtely similr contriutions to the nucleosome stility [188]. There re 14 direct-inding sites in one nucleosome etween the minor groove of the DNA doule helix nd NCP [187,188]. Some reserchers reported the discovery of NCP preferred sequences, which explined the sitution of uneven NCP distriution on the DNA nd nucleosome depletion t the gene promoter nd til sites where trnscription fctors re ound [192]. Additionlly, totl of 3120 wter molecules exist in one nucleosome, mong which, 121 forms the hydrogen ond etween NCP nd DNA [188]. This nucleosome model offers the foundtion to explin the chromtin remodeling process, which llow cells to control their gene expression. Bsed on the density of the chromtin structure, two chromtin forms hve een proposed, which re the euchromtin nd heterochromtin [193]. The euchromtin is the chromtin in loose form nd ppers during the interphse of the cell cycle. It contins most of the coding genes in the cell, nd thus llows recognition nd inding of trnscription mchinery. On the contrry, the heterochromtin is constntly in condensed pttern throughout ll cell cycles, which is inccessile nd impossile for trnscription. Thus, the trnsformtion etween these two forms of chromtin drmticlly chnges gene expression nd DNA repliction in the cell. The nucleosome, which is key in chromtin trnsformtion pcking, hs een shown to e the most importnt method for euchromtin/heterochromtin trnsformtion oth in vitro [194] nd in vivo [195] for nerly two decdes. The NCP, which is n importnt component of the nucleosome, plys key role in this process Histone post-trnsltionl modifiction (PTM) All histone proteins in the NCP contin two motif domins, which re the N-terminl til domin nd the fold domin [196]. In ddition, the H 2A lso hs C-terminl til domin [196]. The histone fold domin is highly responsile for NCP formtion nd nucleosome stility mintennce. The N-terminl til domin, on the other hnd, extends out from the nucleosome core The function of histone N-terminl tils So wht is the iologicl function of the histone N-terminl til domin? Erly studies showed tht the NCP remined intct even though the N-terminl til domin ws removed y trypsin tretment [197]. Thus, the histone N-terminl til is not involved in the NCP stility. Biochemicl studies identified the constituents of the N-terminl til domin s contining minly 14

25 four different mino cid residues, including lysine, serine, threonine nd rginines [198]. On the til, ll these mino cids cn e post-trnsltionlly modified y histone modifiction enzymes, which either chnges the inding ffinity etween DNA nd the NCP or recruits the chromtin remodeling complex. Furthermore, unlike the co-trnsltionl N-terminl modifiction of mny other proteins, the histone N-terminl PTM is more reversile [199]. Thus, due to different histone N-terminl til PTMs, chromtin structure ecomes switchle etween tightly pcked heterochromtin nd loosely pcked euchromtin. If properly performed, cells cn decide when nd where to mke certin genes ccessile/resistile to DNA inding proteins nd the trnscriptionl mchinery, which indirectly ffects physiologicl processes (e.g. gene trnscription, DNA repliction nd DNA repir). It is worth mentioning tht some types of histone N-terminl til PTMs re more stle thn others. For instnce, cetyltion nd phosphoryltion tend to e less stle thn methyltion, which suggests more sustined effect from methyltion to control the gene [200,201]. This is ecuse the decetyltion only requires hydrolysis of single mide ond while the demethyltion needs more energy to rekdown the C-N ond [202]. After the very first report of core histone N-terminl til cetyltion promoting gene expression in vitro in 1964 [203], gene epigenetic regultion through histone N-terminl til PTM received more ttention. So fr, mny types of histone N- terminl til PTMs hve een identified, including cetyltion, methyltion, phosphoryltion, ADP riosyltion, proline isomeriztion, deimintion, glycosyltion, uiquityltion/uiquitintion, cronyltion nd sumoyltion [12,198]. Ech type of modifiction will e lnced with nother type nd overll they result in concerted comintoril effect on gene trnscription [204]. The two most importnt histone N-terminl til PTMs re cetyltion nd methyltion. In the following prgrphs, rief discussion will e ddressed out these two PTMs Histone methyltion By providing proper sptiotemporl inhiition on different genes, orgnisms estlish nd mintin the identity of different types of cell. Epigeneticlly speking, the function of histone methyltion is one of the most importnt regultion mechnisms in the orgnism to provide this process. Exmples of this type of regultion involve X chromosome inctivtion, heterochromtin formtion nd trnscriptionl inhiition/ctivtion [205]. Mlfunction of histone methyltion hs long een relted to increses in the rte of cell trnsformtion, thus incresing the chnce of oncogenesis [206]. However, recent studies lso showed tht histone methyltion cn ctivte genes, which suggests it hs dul effect in cells [205]. The methyltion could hppen on different mino cid residues, including oth lysine nd rginine residues [207]. At the sme mino cid residue, the methyltion could hppen to different degrees, including mono-, di-, or trimethyltion [208,209]. For exmple, the lysine methyltion could otin up to three methyl groups while rginine could only do one [210]. Numerous studies so fr, focused on the function of the lysine (K) residue, especilly on the H 3 N-terminl til. This is ecuse ech lysine methyltion on the H 3 N-terminl til induces unique effect on chromtin structure nd gene trnscription 15

26 [211]. Compred to H 4, which hs only one lysine methyltion site (H 4 K 20 ), there re more thn 20 methyltion sites on the H 3 N-terminl til tht hve een identified nd five lysine residues re thought to e the most functionlly importnt [202]. These sites re K 4, K 9, K 27, K 36 [201] nd K 79 [212]. Tle 1 lists the iologicl effect of methyltion t different lysine sites on the H 3 N-terminl til. Mny studies hve estlished close reltionship etween the H 3 N-terminl til methyltion nd norml development/neoplsm genesis [213]. Among them, K 27 is the most importnt methyltion site, which inhiits gene trnscription [214] H 3 K 27 Methyltion induced gene trnscription inhiition H 3 K 27 methyltion exists in three forms, which re mono-, di- nd tri-methylted [215]. In mmmlin cells, the distriution percentge for H 3 K 27 Me 0, H 3 K 27 Me 1, H 3 K 27 Me 2 nd H 3 K 27 Me 3 is 15%, 25%, 50% nd 10% [216]. In Drosophil nd mmmlin cells, ll H 3 K 27 is conducted y EZ nd EZH1/2 proteins (their properties will e discussed lter in this thesis). The mechnism of histone N-terminl til induced gene repression will e discussed in lter prgrphs. Compred to H 3 K 27 Me 2 nd H 3 K 27 Me 3, H 3 K 27 Me 1 function nd regultion remins poorly known. While the plnts use specific mono-hmts to conduct H 3 K 27 Me 0 to H 3 K 27 Me 1 [217], EZH1, EZH2 nd EED together re the responsile for oth Me 0 to Me 1 nd Me 2 to Me 3 [218,219]. In Drosophil, chromtin scnning reveled the distriution of H 3 K 27 Me 1 is prevlent t oth heterochromtin (e.g. pericentromeric region) nd euchromtin (ut selectively depleted round trnscription strt sites of ctive genes) [14,220]. This suggests tht H 3 K 27 Me 1 might e involved in heterochromtin structure mintennce nd reservoir for inhiitory DNA inding proteins t ctive gene loci. Additionlly, H 3 K 27 Me 1 could lso prevent the cetyltion of H 3 K 27 nd other PTMs. One most current review pper suggested the existence of H 3 K 27 Me 1 inding protein, which could either directly induce chromtin structure chnge or recruit other proteins [217]. A potentil cndidte for such protein is the Decrese in DNA Methyltion 1 (DDM1) protein identified in the plnt, Aridopsis [221,222]. Although no experiment so fr hs directly shown correltion etween H 3 K 27 Me 2 nd gene expression level, some studies hve indirectly suggested it s n inhiitory iomrker. For exmple, in Drosophil, H 3 K 27 Me 2 coincides with H 3 K 27 Me 1 t chromocenter regions [14], which re highly pcked re nd hve no ccess to polymerse. Similr results were otined in Aridopsis [223]. Similr to H 3 K 27 Me 3, the H 3 K 27 Me 2 is methylted from H 3 K 27 Me 1 y EZH2 in the PRC2 complex in Drosophil [224], nd indicted y the knock-down of EZH2 which decresed the levels of oth H 3 K 27 Me 2 nd H 3 K 27 Me 3 [215]. Recently, H 3 K 27 Me 2 is suggested s n intermedite sttus [215], which prevents gene ctivtion through oth supplying H 3 K 27 Me 3 rw mteril nd preventing H 3 K 27 cetyltion [213]. The H 3 K 27 Me 3 is the only methylted H 3 K 27 tht is clerly demonstrted s iomrker for gene repression. In Drosophil cells, lthough lso detected t limited euchromtin loci, H 3 K 27 Me 3 loctes constntly t pericentric heterochromtin t high concentrtions [14]. Similr condensed distriution of H 3 K 27 Me 3 is lso oserved in inctivted X chromosomes in 16

27 mmmlin cells [13]. In Aridopsis defective in H 3 K 27 Me 3 nd H 3 K 27 Me 3, specific demethylses (JMJ12) cused hundreds of genes to fil to e silenced nd extensively silenced, respectively [225]. EZH2 knock-down [215] nd knock out [218] cused significnt decreses, ut not totl loss of H 3 K 27 Me 3. The explntion is the existence of EZH1, which is homolog of EZH2 nd replces EZH2 nd keeps miniml necessry H 3 K 27 Me 3 for emryo development. Additionlly, noticele increses of H 3 K 27 Me 3 were recently reported fter novel point muttion in the SET domin of EZ protein, which resulted in sustntilly lowered Ad-B gene expression in Drosophil [226]. Tle 1. lists ll lysine residues on H 3 N-terminl til, which so fr hs een found s potentil methyltion sites. Their function ws indexed with some relevnt references. Histone Methylted Lysine Residue Level of Methyltion Effect on gene trnscription Reference H 3 Lys 4 Me 1 ctivte [227,228] Me 2 Me 3 [208,229] Lys 9 Me 1 found within silent domins of euchromtin [230], [231,232] Me 2 mye relted to ctivtion found enriched within silent domins of euchromtin, function uncler. [230] Me 3 inhiit [14,216,226,230,231,233] Lys 27 Me 1 Function unknown. Mjority of ppers tend to suggest [14,216,220,230] Me 2 it s n inhiitory mrker. But one suggests it s mrker for ctive gene [214]. Function unknown. One pper suggested it s n intermedite stte etween Me 1 nd Me 3 [215]. [14,215] Me 3 inhiit [13,234] Lys 36 Me 1 inhiit [235,236] Me 2 ctivte Me 3 Tle 1. Potentil methyltion lysine residues on the H 3 N-terminl til nd their effect on gene expression 17

28 Introduction to the PRC2 Histone N-terminl til methyltion is crried out y the histone methyltrnsferses (HMTs), whose key chrcteristic is the SET domin. The SET domin is short for the Su(vr)3-9; E(z); Trithorx, which is defined s the conserved protein domin tht hs methyltrnsferse cpcity [237]. So fr, mny proteins contining the SET domin hve een found to e memers of the HMTs fmily, including the polycom repressive complex (PRC). The PRC is highly conserved from Drosophil to humn nd two sutypes of the PRC hve een identified, which re PRC1 [238] nd PRC2 [239]. These two sutypes consist of different components nd methylte histone N-terminl tils. It remins elusive s to which protein serves s the HMT in PRC1 [240,241]. On the other hnd, PRC2 hs so fr the only specific HMT for H 3 K 27 di- nd tri-methyltion. In humns, its HMT function is conducted vi the key component in the functionl core, clled EZH [213]. Additionlly, three more proteins re identified in the PRC2 functionl core s well, which re EED, SUZ12 nd RAp46/48 (or clled RBBP7/4) [213,242,243]. In Drosophil, PRC2 is lso known s the ESC-EZ complex. The ESC/ESCL, histone H4 inding protein p55 nd EZ of the ESC-EZ complex re the ortholog of EED, RAp46/48 nd EZH protein in humns, respectively. All the protein in the PRC2 nd PRC1 functionl cores elong to the Polycom group (PcG) fmily [244]. In erly Drosophil studies, the PcG fmily served s n importnt repressor to the Hox gene nd controlled Drosophil ody-segmentl development. Now, the PcG fmily is proven to e required during physiologicl processes including emryonic development nd cell pluripotency [245]. Selective knocking-out of one of the PcG proteins in the PRC2 nd PRC1 functionl cores leds to emryonic lethlity nd developmentl normlities, respectively [246]. The following prgrphs will shortly ddress the known knowledge of these four PcG proteins in the PRC2 functionl core Components in the functionl core of the PRC2 In 2002 study, the structure of the SET domin in the EZH2/EZ protein ws determined [247] In this study, the EZH2/EZ ws determined s the ctlytic site of the PRC2, which conducted methyltion on H 3 K 27 nd H 3 K 9 residues [213,247]. Although susequent studies reported different findings for which EZ hd no effect on H 3 K 9 residue [242,248]. It is with no dout tht its methyltion on H 3 K 27 is specific since the mono-methyltion process on H 3 K 27 grdully ccumultes methyl groups from Me 0 to Me 1, Me 1 to Me 2, Me 2 to Me 3 [249]. Mny studies hve ridged the reverse reltionship etween EZH2 nd gene trnscription vi H 3 K 27 methyltion on genes. After overexpression of EZH2 in cncer cells n incresed H 3 K 27 Me 3 nd histone decetylses (HDAC) were oserved t the promoter site of Runt-relted (RUNX) fmily genes [250]. On the other hnd, the EZH2 knockdown exhiited oth decresed H 3 K 27 Me 3 nd HDAC t the promoter site of the RUNX gene, which llowed the recovery of RUNX gene expression [250]. In the EZ knockout, Drosophil lrv methylted t H 3 K 27 were rely detectle [14]. If the SET domin inside the 18

29 EZH2/EZ is disrupted, its HMT function is olished s well. For exmple, y replcing conserved histidine residue (H689) with n lnine (H689A) inside the SET domin of EZ, the HMT ctivity of PRC2 ws decresed in vivo [247]. Similrly, switching the cysteine-545 (C545Y) to tyrosine within the SET domin not only decresed the HMT ctivity, ut lso inhiited the inding etween the PRC2 nd specil DNA regions clled PcG response elements (PREs), thus inhiiting the histone methyltion s well [247]. However, even with the existence of norml EZH2/EZ, the PRC2 still hs no HMT function t ll if EZH2/EZ lcks coopertion from the other three proteins inside the PRC2 functionl core. The explntion for this phenomenon is tht EZH2/EZ requires DNA inding efore it performs the HMT function. Becuse EZH2/EZ itself does not hve DNA inding domin, it needs help from other proteins to locte the position, nchor to the chromosome nd potentite the methyltion. So fr, EZ hd een shown to hve direct connection with the ESC in Drosophil [251] nd SUZ12 [252], or mye RAp46/48 s well [253]. These three proteins inside the functionl core, together with other ccessory proteins, complete the well regulted methyltion process. The EED is conserved protein found in the PRC2 complex nd is essentil for the PRC2 silencing function. In Drosophil, the EED hs two homologues, which re the ESC nd ESCL [254], nd oth re coded y the Esc gene [255]. The function of EED is elieved to e connecter inside the PRC2, which ridges the histone with the HMTs inside the PRC2. For exmple, in oth Drosophil [256] nd mice [257], the Esc gene knock-out in the emryo inhiits the homeotic genes, just s other PcG proteins in the PRC2. Susequent Esc gene knock-in could rescue the emryo, if only the supplement ws given in the first severl hours of development [256]. This suggests the unique time-sensitive iologicl effect of the EED protein, which is different from other PcG proteins [258]. The PcG protein suunits inside the PRC could recognize the polycom response elements (PREs). RNA interference specificlly knocking down the ESC led to oviously decresed levels of the EZ- PRG inding, which inhiited the methyltion nd susequent regultory protein inding [242]. Susequent studies showed tht this ws prtilly due to the filure of the ridge effect of EED/ESC nd proper EED/ESC ssocition. Biochemicl studies showed tht EED/ESC hd C-terminl with seven WD regions, which could directly ind to oth the N-terminl of EZH2/EZ [251] nd H 3 [255]. After proper ridging y EED/ESC, the EZH2/EZ then performed its methyltrnsferse effect nd conducted the glol H 3 K 27 methyltion [259]. Interruption of the EED/ESC protein ridging resulted in glol H 3 K 27 methyltion filure, which cused emryo development filure in oth Drosophil.[259] nd mmmls [260]. Only in the presence of EED/ESC, H 3 K 27 methyltion ws properly conducted [259]. Additionlly, the EED showed connection to the HDAC in the PRC2 s well [261,262]. Histone decetyltion is criticl step in gene repression. One study reported the finding of highly specific interction etween HDAC1/2 proteins nd EED [262]. The SUZ12 protein is reltively new component found in the PRC2 complex. Gene lignment demonstrted tht it ws highly conserved protein from plnts to mmmls [263]. Although its detiled function remined uncler, the loss of the 19

30 Suz12 gene resulted in the totl sence of the H 3 K 27 Me 3 in Drosophil [264]. Biochemicl nlysis identified two regions inside the SUZ12 protein, the zinc finger motif domin nd the VEFS (VRN2-EMF2-FIS2-Suz12) domin [252]. The function of zinc finger motif is unknown, nd one pper reported tht this region did not hve DNA inding ility [263]. On the other hnd, the VEFS hd direct physicl contct with EZH2 nd heterochromtin protein 1α (HP1α) oth in vitro nd in vivo [252], which ws confirmed y n immunohistochemistry study [264]. VEFS domin mlfunction induced y Suz12 gene missense muttion cused the PRC2 to lose its methyltrnsferse ility while still mintining proper PRC2 complex structure [259]. RAp46 nd 48 re two proteins discovered inside the PRC2 core in humns [242,247] nd their homolog is clled p55 in nucleosome remodeling fctor NURF (NURF55) in Drosophil nd p55 in fly [265]. Both RAp46/48 [266,267] nd NURF55 could directly ind to H 4 nd H 2A [253,268]. Biochemicl nlysis reported the structure of RAp46/48 contined seven WD repet motif, which ws similr to tht of EED [253]. Previous studies reported tht the RAp46/48 ws directly ssocited with SUZ12 [269] nd HDAC1 [253]. These dt suggested tht RAp46/48 functions s ridge to connect its inding histone to other ctlytic units in the PRC2. The exct function of RAp46/48 in PRC2 remined uncler due to conflicting studies. Two recent studies found p55 ws either essentil [270] or not essentil t ll for PRC2 functioning [271]. However, its function in Drosophil chromtin ssemly fctor 1 (dcaf-1) might give clue out how it works. A previous study showed tht p55 ws lso detected in dcaf-1, in which p55 hd direct contct with RPD3, HDAC [253]. The deletion of p55 in dcaf totlly olished the function of dcaf-1 [272]. This indicted tht p55 might conduct its function through setting up ridge etween the histone protein nd functionl unit inside the dcaf-1, including RPD3. Thus, some ppers proposed the sme model for p55 in the PRC2 sed on the following experiment results: 1) the HDAC1/2 proteins were identified in the PRC2 [262]; 2) p55 hd direct contct with HDAC1/2 in PRC2 [253]; nd 3) p55 mutnt Drosophil strin hd reduced levels of oth PRC2 nd its product, H 3 K 27 me 3 [270] Proteins out of the functionl core of PRC2 There re lso some other proteins, which re not in the functionl core, tht help in PRC2 functioning. So fr, severl sequencespecific DNA-inding proteins hd een identified. For exmple, in oth PRC1 nd 2, PREs recognition proteins were identified [273], which were the pleiohomeotic (PHO) in Drosophil nd its mmmlin orthologue, Yin Yng (YY) 1 [274]. One study showed tht the YY1 nd PHO were crucil for recognizing the PREs region, thus llowing the PRC complex to conduct methyltion t well defined res [273]. Additionlly, the Zeste protein co-loclized with oth PRC1nd PRC2 on the chromtin t mny sites, suggesting the ssocition of Zeste with the PRC [275]. Becuse the Zeste protein could recognize nd ind to specific DNA sequence (T/CGAGT/CG), it is elieved tht Zeste lso helps the PRC to locte sites [276]. Other proteins include GAGA fctor (GAF), pipsquek (Psq), dorsl switch protein (Dsp1), griny hed (Grh) nd SP1/KLF [277]. Nevertheless, how ll these proteins help the PRC to locte the position is still not cler. 20

31 Histone Acetyltion Although the core histone nd certin C-terminl tils re sujected to cetyltion modifiction s well, they seem to e much less common compred to tht of the N-terminl til cetyltion [12]. Histone cetyltion is overll relted to incresed gene expression [193,278,279]. For exmple, in chicken emryo erythrocytes, the lph-d-gloin gene is n ctively trnscried gene, which contins 15- to 30-fold more cetylted histone compred to those genes tht re not ctively trnscried [280]. After deleting H 4 N-terminl til, H 4 til could no long e cetylted, which resulted in genes trnscription inhiition [281]. Among ll cnonicl histones, H 3 is the most populr post trnsltionlly modified protein in epigenetic reserch regultion. Incresed levels of cetylted H 3 were found t the gene promoter re, which is highly correlted with gene trnscription ctivtion [282]. The H 3 N-terminl til hs the most vrieties nd sites of PTM compred to its counterprt s well [12]. The lysine residue is the mjor site for enzymtic PTM [201]. Although in eukryotic cells, there remined some differences mong smple sources (e.g., humn orgn cellc vs yest) most of the time, the lysine sites on the H 3 N-terminl til re either methylted or cetylted [201]. So fr, in eukryotic cells, five lysine sites on the H 3 N-terminl til hve een identified to e cetyltion modifile, which re the Lys (K) 9, K14, K18, K23 nd K27 [201]. The H 3 K 27 cetyltion (Ac) hs een reported in yest [283], humn, nd mouse [201]. Using CHIP-Seq techniques, H 3 K 27 Ac ws confirmed to e positively correlted with gene expression nd locted round trnscription strt sites [15]. It is iomrker to distinguish ctive enhncers nd poised enhncer, which re enriched nd sent in H 3 K 27 Ac, respectively [284,285]. Additionlly, H 3 K 27 Ac hs complimentry effect with methylted H 3 K 27, nd cn replce or e replced y H 3 K 27 Me 2 nd H 3 K 27 Me 3 [15,284,286,287], thus switch the gene trnscription sttus. Histone N-terminl til cetyltion is conducted y cetyltrnsferse (HATs) first reported in yest [288] nd, now, mny HATs hve een identified in vrious nimls [198]. In Drosophil, the most well known HAT-ering complex is trithorx cetyltion complex (TAC), whose functionl core hs trithorx protein (TRX) nd CREB inding protein (CBP) [289]. TRX elongs to one su-fmily of trithorx group (TrxG) gene, which is cple of recognizing specific DNA domin clled TrxG response elements (TREs) [290]. Additionlly, TRX lso ers HMTs SET domin nd methyltes H3K4 specificlly s n excittory gene iomrker [291]. TREs serve the exct sme function s tht of PREs for the PcG complex. This llows the TrxG complex ering TRX to e le to recognize specific gene locus. On the other hnd, CBP is the true HAT, which directly nd selectively conducts cetyl group trnsfer to H 3 K 27 nd H 3 K 18 [287,292]. The function of the TrxG fmily protein is involved in multiple cell developmentl/ging processes [293]. Although it remins controversil, the current hypothesis is tht the TrxG protein countercts the effect of the PcG [294,295]. For exmple, in Drosophil, the TrxG nd PcG regultes the on nd off sttus of the Hox gene, which ffects Drosophil ody segment development [294]. The TREs usully cluster or even coincide with the PREs [295,296]. With the help of some shred 21

32 components inside the TrxG nd PcG (e.g. PRC2), these two complexes could e recruited to the sme DNA regions [293], ind simultneously [297] nd compete for the sme residue, H 3 K 27 [255]. Without ffecting CBP level, decresed TRX protein ws correlted with reduced H 3 K 27 Ac levels nd vise vers [255]. Trx gene or TRX protein muttion gve the sme results [289]. CBP hs long een identified s HAT [298], which is s essentil s HMTs during erly life development. For exmple, CBP knock-out cused developmentl rrest in Drosophil emryoes [299]. On the contrry, regionl overexpression of CBP in Drosophil cused ody developmentl normlity nd glol histone cetyltion, while uiquitous overexpression induced mortlity [300]. Recently, H 3 K 27 ws reported s specific cetyltion site for CBP [255]. In the study, the uthors found tht prtil CBP knock-out llowed Drosophil to develop ut with significnt decresed H 3 K 27 Ac, incresed H 3 K 27 Me 3, nd no chnges in H 3 K 27 Me 2. Conversely, modertely overexpressed CBP levels exhiited incresed H 3 K 27 Ac nd modertely reduced H 3 K 27 Me 3 in Drosophil. Ltely, the sme group reported the physicl ssocition of CBP with H 3 K 27 -specific demethylse UTX nd chromtin-remodeling fctor BRM [301]. This suggests further support for the counterctive effects etween cetyltion nd methyltion on H 3 K 27 nd chromtin structure chnge fter H 3 K 27 Ac Chromtin Remodeling As mentioned ove, lthough the presence of nucleosomes helps the 2 m DNA to e well folded into the eukryotic cell, it stericlly occludes the ccessiility of DNA to the DNA inding molecules from smll proteins (DNse I) nd to huge trnscriptionl mchiner. Thus gene trnscription would e inhiited if there is no wy to relese the nucleosomes on the chromtin [302,303]. The iologicl system is smrt enough to incorporte severl mechnisms, such s chromtin structure remodeling, to llow ccess of polymerse to DNA. But, in 6 million p DNA pool, how is the cell le to know the exct site nd time to remodel chromtin structure? The histone N-terminl til PTM is one of the most importnt iomrkers tht cells utilize to ccurtely remodel chromtin structure Brief introduction of chromtin structure remodeling nd remodeling complex The chromtin structure remodeling is process tht cells utilize to regulte gene expression without chnging their DNA sequence. So fr, five types of chromtin dynmic chnges hve een demonstrted, including histone switching, histone ejection nd recruitment, histone replcement nd histone sliding. The overll iologicl roles of these chnges mke chromtin switchle etween heterochromtin nd euchromtin. These remodeling processes re conducted y the chromtin remodeling complex. Chromtin remodeling complex is protein polymer, which use ATP to mnipulte nucleosomes on the chromtin [304]. This complex usully consists of motor proteins nd ccessory proteins [305]. The ccessory proteins re minly involved in nucleosome inding, recognizing histone covlent modifiction, intercting with trnscriptionl fctors or other 22

33 regultory proteins nd regulting the motor protein. On the other hnd, the motor protein contins the rel ATP hydrolysis domin, which utilizes the ATP to rek down the interction etween NCP nd DNA, nd cuse chromtin remodeling. So fr, inside the chromtin remodeling complex, four fmilies of SF2 hve een identified, which re SWI/SNF fmily, imittion SWI (ISWI) fmily, chromodomin, helicse, DNA inding (CHD) fmily nd inositol requiring 80 (INO80) fmily [305] Histone cetyltion induced chromtin remodeling Histone cetyltion ctivtes gene trnscription y chnging chromtin structure into less pcked sttus Two mechnisms for histone cetyltion induced chromtin structure chnge Firstly, histone cetyltion hs een shown to weken/decrese the contcts in the nucleosome etween DNA nd NCP [198,306]. The histone til cetyltion cn neutrlize histones positive chrge, nd thus decrese their ffinity for DNA. The reduced ffinity or even direct inding etween DNA nd NCP led to unfolded chromtin nd out-looped or even nked DNA strnds, which fcilitte the inding etween DNA nd the trnscription mchinry [198,204]. Opposite to the cetyltion, histone decetyltion is elieved to restore the positive chrge on the nucleosome, which strengthens the inding nd results in gene inhiition. Although this model is the most populr hypothesis nd supported y mny studies, it is chllenged y lter study, which reported tht histone cetyltion did not weken the DNA-histone inding [307]. Additionlly, histone cetyltion chnges the inding ffinity etween chromtin nd trnscription-relted proteins. In S. cerevisie, severl silencing relted trnscription fctors (TFs), such s type 3 nd 4 of silent informtion regultors (Sir) [308] nd Tup1p [309], decresed their ffinity to cetylted H 3 nd H 4, which resulted in reduced gene-inhiition. The H 4 decetyltion, on the other hnd, will recruit the Sir3 [310]. Secondly, histone cetyltion could recruit chromtin-remodeling fctor, which fcilittes the heterochromtin-toeuchromtin trnsformtion [198]. An erly study oserved tht histone cetyltion disrupted the higher-order of chromtin nd ctivted gene expression [311]. However, the decresed inding etween DNA nd NCP lone is not supposed to llow such chnge. Lter, more nd more reserchers relized tht histone cetyltion is lso le to recruit chromtin-remodeling fctors, such s SWI/SNF [312,313,314], to help them fcilitte chromtin remodeling Two models of histone cetyltion induced gene ctivtion Comining ll these mechnisms together, two models hve een proposed to explin the overll gene ctivtion effect induced y histone cetyltion [315,316]. The first model elieves tht gene ctivtion is sed on the ility of the chromtin-remodeling fctor/tf to recognize locus-specific cetyltion/decetyltion [316]. There remins cis-element t the up-strem of gene promoter. After inding of sequence-specific DNA-inding proteins to this region, protein-recruiting cscde occurs. Tking the repressing-process s n 23

34 exmple, during this cscde, HADCs rech this region, followed y chromtin-remodeling complex. The chromtin-remodeling complex fcilittes the histone moving to the vicinity of the promoter core fter the HADCs decetylte the promoter. In this cse, the trnscriptionl mchinery could no longer gin ccess to the promoter re of the down-strem gene, which is how trnscription gets inhiited [198,317]. If the repressing DNA-inding protein does not ind to the cis-element, the HADCs will not e recruited, which is followed y no chromtin-remodeling protein. Thus, if the nucleosome histones re cetylted, the chromtin remodeling complex will e ttrcted to this locus y its suunit clled romodomins [318]. After series of nucleosome dynmic chnges, the chromtin structure t the promoter region is wide enough for trnscriptionl mchinery to strt trnscription. The second model presents more non-specific nd glol effect s compred to tht of the first pthwy. Insted of just promoter vicinity regions, the whole genome is getting cetylted y HAC [198,315,316]. The chromtin remolding complex will then e ttrcted to the cetylted chromtin res. In this cse, the whole genome ecomes looser, which offers more ccessiility to the trnscriptionl mchinery. This sounds like tht histone cetyltion will cuse unspecific gene ctivtion. Well, different genes locte t different chromosome regions, where chromtin structure, nucleosome numers nd protein ccessiility re different. These properties mkes the cetyltion to e t different level nd time period which indirectly mkes it gene specific. Additionlly, HADCs is constntly in compny with the HAC, which lnce the trnscription of different genes [319]. This mkes the gene trnscription regultion even more precise Histone methyltion induced chromtin remodeling As mentioned ove, the histone-cetyltion neutrlizes the positive chrge of the histone. However, for methyltion, no chnge in electricl chrge hs een detected [205]. Specificlly, for H 3 K 27, no evidence so fr suggest tht its methyltion is directly relted to chnges in the chromtin structure [320]. Thus, histone methyltion must use other mechnisms to conduct its inhiitory effect. So fr, two models hve een proposed. The first model hypothesizes tht the histone til methyltion is to recruit or offer docking plce for some chromtin remodeling fctors, which mnipulte the downstrem chromtin structure [205]. One exmple is the oservtion tht methylted H 3 K 27 recruits the PRC1 [247,321,322]. The PRC1 hs protein constituent clled Polycom, which ers chromodomin [321,322]. Becuse chromodomin specificlly recognizes H 3 K 27 Me 3, it is resonle to suspect tht the PRC1 is recruited to this locus. To support this hypothesis, one study showed fter H 3 K 27 de-methylse UTX knock-down, oth methylted H 3 K 27 nd PRC1 significntly incresed inding to chromtin [323]. Another iochemicl study reveled tht PRC1 otined n extended recognition groove which specificlly recognized H 3 K 27 methyltion [322]. However, opposite results were lso pulished lter y showing PRC2 independent recruitment of PRC1 [324]. Thus, this hypothesis remins inconclusive. 24

35 Recent studies lso suggest tht H 3 K 27 Me 3 cn stop nery trnscriptionl mchinery pssing loclly. In humn emryonic stem (ES) cells, some genes re enriched with inry modifictions (two coexisting PTMs on the sme histone N- terminl til). For exmple, oth gene-ctivting H 3 K 4 Me 3 nd gene-inhiiting H 3 K 27 Me 3 hve een detected t severl genes promoters [325,326,327]. These genes re clled ivlent genes. Although the function of these remins uncler, the previling ide is tht these genes remin in poised stte, which is redy for rpid trnscriptionl ctivtion t the presence of specific signls [328]. One study found tht t the promoters of these ivlent genes, high level of trnscriptionl mchinery ws detected, ut with low trnscription levels [329]. A following study repeted nd greed with their results [330]. Thus, it suggested tht H 3 K 27 Me 3, put together with H 3 K 4 Me 3 t promoter res, is le to prevent trnscriptionl mchinery from pssing, nd thus inhiit gene trnscription. Additionlly, one pper demonstrted tht the inry modifiction on H 3 til (dimethyltion nd phosphoryltion) lso llows chromtin to fold into specific conformtion [331], which might prohiit the ssocition of chromtin-remodeling fctor from otining chromodomin nd preserve its heterochromtin structure. Additionlly, newer novel model ws introduced for HMT induced gene repression. This study reported tht histone til methyltion y HMTs prevented unlignded nucler receptors from inding to their trget gene, which prohiited gene conformtionl chnges nd kept them inctivted [332]. However, detiled informtion remins uncler. 25

36 CHAPTER 2: Chnges of Polycom Repressive Complex 2 (PRC2) cused y neontl fsting stress protect chickens from future fsting stress Ying Jing, Cynthi J. Denow, Michel D. Denow Astrct Unfvorle nutritionl conditions during the neontl criticl-period cn cuse oth cute metolic syndromes nd severe diseses in lter life. It is due to epigenetic chnges on different genes controlling feeding regultory fctors. Erly studies reported tht 24 hr fsting t 3 dy-of-ge (D3) cused chnges in inhiitory epigenetic mrks on neurons in the hypothlmic prventriculr nucleus (PVN), n re ssocited with energy homeostsis. In the current study, we further investigted the effects of fsting t D3 on epigenetic chnges in the rin of chickens. Methyltion of histone 3 t lysine 27 (H3K27) nd H3K27 methyltrnsferse (HMTs) were incresed fter fsting in PVN neurons, nd this effect ws lso evident t 10 (D10) nd 40 dys-of-ge (D40). By compring to the significntly elevted methylted H2K27 nd HMTs cused y single lte dy fsting (D10 or D40), pre-treted D3-fsting were le to prevent these surges nd ring these prmeters ck or close to control. Chnges in rin-derived neurotrophic fctor, which is n importnt protein for neuron survivl nd energy homeostsis, further supported these dt. These findings show tht neurons in the PVN re cple of cclimting to repetitive fsting stresses due to the erly exposure to fsting. It suggests tht the epigenetic chnges cn induce moleculr memory to protect individuls ginst future similr stresses. 26

37 2.1. Introduction The developing rin is sensitive to environmentl influences such s unfvorle nutrition during the neontl period tht my result in cute metolic disorders [1,2,3] nd severe diseses in lter life [4,5]. These erly life experiences occurring during heightened periods of rin plsticity help determine the lifelong structurl nd functionl spects of rin nd ehvior. In humns, for exmple, weight gin during the first week of life fter irth incresed the propensity for developing oesity severl decdes lter [5]. This my e due to extreme susceptiility of the developing rin, whose function is esily disrupted nd conditioned y nutritionl stress [3,333,334]. Long-term consequences of dverse conditions in erly life my e reflection of trnscriptionl chnges for genes within the centrl nervous system (CNS) tht occur s result of epigenetic modifictions [335,336,337]. Epigenetics refers to genome expression regultion tht is not due to DNA sequence chnges [338]. Chemicl modifictions mde to histone proteins on the chromtin is one of the most importnt options to perform epigenetics [338]. Although it hs een shown tht cute nutritionl stress chnged inhiitory modifictions on histones t genome-wide level [6] nd ltered the CpG methyltion sttus of specific genes [8], the longevity of these epigenetic mrks induced y nutritionl conditioning during rin development period remin unknown. Neither the function nor purposes of these chnged epigenetic mrks is known. N-terminl til post-trnsltionl modifiction (PTM) of histone proteins is one of the most importnt epigenetic chnges utilized in ll cells to regulte gene expression. Thus fr, vrious types of PTMs hve een identified on vrious histones [12]. Among them, the methyltion nd cetyltion on lysine residue (K) 27 on histone 3 (H3) hs een linked to gene repression [13,14] nd ctivtion [15], respectively. H3K27 methyltion is crried out y histone methyltrnsferses (HMTs) in multiprotein complex, clled polycom repressive complex 2 (PRC2) [339]. Three polycom group (PcG) proteins mke up the PRC2 functionl core [213], including emryonic ectoderm development (EED) protein, enhncer of zeste 2 (EZH2) nd suppressor of zeste (SUZ) [213,242,340]. While EED nd SUZ recognize specific histone site on chromtin nd ridge it with the PRC2 [242,259], EZH2 is the only HMT within the PRC2 tht produces di- (H3K27Me2) nd tri-methylted H3K27 (H3K27Me3) [259]. No experiment to dte hs studied the dynmic chnges in these suunits of PRC2 during norml development nd under nutritionl stress in chickens. Brin-derived neurotrophic fctor (BDNF) is n importnt hormone involved in neuron survivl nd rin development, nd is highly expressed in the hypothlmus [78]. As n norexigenic fctor, BDNF induces feeding suppression fter cute ppliction [161,341] nd weight loss during long-term ppliction in nimls [162]. On the other hnd, fsting inhiited BDNF mrna level in severl res of the hypothlmus [75]. The prventriculr nucleus (PVN) is key integrting nucleus for vrious metolic signls nd mintins norml energy homeostsis y coordinting other hypothlmic nuclei [342]. It contins high level of BDNF [79] nd responds to locl BDNF microinjection y instntly inhiiting ppetite nd feeding ehvior [161]. 27

38 The BDNF gene hs een the focus of mny developmentl studies imed t understnding the reltionship etween erly-life stress, rin responses, nd ehviorl outcomes. Dt demonstrte tht erly-life events influence the Bdnf gene nd ehviorl outcome [343,344]. A previous study reported the genome-wide chnges of inhiitory epigenetic mrks (H3K27me2, H3K27me3 nd EZH2) fter fsting in neontl chicks [6]. However, no study hs correlted these chnges t the Bdnf locus to gene expression. In our prior study fsting induced chnges in methylted H3K27 levels in neontl chickens [6]. In the current study, we monitored genome-wide chnges in the di- nd tri-methyltion sttus of H3K27 nd expression of key components of HMTs (EED, SUZ nd EZH2) in PVN neurons following 24 hr fst in oth neontl nd mture chickens. We lso investigted the effects of preconditioning chicks t n erly ge y fsting nd looking t the chnges in methyltion sttus of H3K27 following susequent fst. The di- nd trimethyltion levels of H3K27 long the dnf gene during fsting of 3-dy-old chicks nd during susequent fst fter the criticl period of feed-intke control estlishment were evluted y chromtin immunoprecipittion ssys. 28

39 2.2. Mterils nd methods Animls The dy of htching ws considered postntl dy 0. One dys-of-ge (D1) mle white Plymouth Rock chickens were otined from the Co-Vntress Htchery in Wdesoro, NC, USA. They were trnsported to VT nd initilly housed in Petersime tteries in groups of ten prior to twenty-one dys-of-ge (D21) nd in groups of four fterwrds. The room temperture ws 30 ± 2 C nd 50 ± 5% reltive humidity. Wter nd food were ville d liitum. All experimentl procedures were performed in ccordnce with the Virgini Tech Animl Cre nd Use Committee. Chemicls Antiodies were purchsed from Cell Signling Technology (Dnvers, MA, USA) nd included Histone 3 (H3), tri- Methylted Histone H3 t Lysine 27 (H3K27Me3), di-methylted Histone H3 t Lysine 27 (H3K27me3) H3K27Me2, β- ctin ntiodies nd chromtin immunoprecipittion ssy kits. BDNF ntiody ws purchsed from Acm Inc. (Cmridge, MA, USA). Anti-rit IgG horserdish peroxidse-conjugted ntiody ws from Snt Cruz Biotechnology (Snt Cruz, CA, USA). SYBR Green ws from Life Technologies Corp (Crlsd, CA, USA). All dditionl chemicls were purchsed from Sigm-Aldrich (St. Louis, MD, USA). Protein isoltion nd western lotting Nucler protein extrctions were conducted s descried previously [6]. Before western lots, the protein concentrtion ws quntified y BCA protein ssy, ccording to the mnufcturer's instruction (Thermo Fisher Scientific, Rockford, IL, USA). All smples were then djusted to the sme concentrtion. Before loding, smples were mixed with Lemmli loding uffer [345] t 1:1 volume nd heted for 10 min t 99 C. Protein extrcts were seprted on 12% SDS-PAGE pre-cst gels (Lonz, Wlkersville, MD, USA) nd trnsferred to nitrocellulose memrnes (12 V for 1 h, 24 V for 1 h). The memrnes were then locked in 5% ovine serum lumin (BSA) overnight t 4 C with shking. On second dy, the memrnes were wshed y 1% BSA 5min * 3 times. Primry ntiodies (Ezh2 t 1:500, H3K27Me2 t 1:1000, H3K27Me3 t 1:1000, H3 t 1:1000, β-ctin t 1:1000 nd BDNF t 1:1000) were incuted with the memrnes overnight t 4 C with shking. On the second dy, memrnes were wshed nd incuted with nti-rit IgG horserdish peroxidse-conjugted ntiody t room temperture for 1 h with shking. A chemiluminescent signl ws detected following incution with SuperSignl West Pico sustrte (Pierce, Rockford, IL, USA) using Gel Doc XR System (Bio-Rd, Hercules, CA, USA). The Quntity One (Bio-Rd, Hercules, CA, USA) imge nlysis softwre ws used for nd density nlysis. The finl rtio vlue of ech protein ws equl to the rtio of the trget protein to the density vlue of β-ctin protein. Rel-time polymerse chin rection (qpcr) ssy design 29

40 The totl RNA ws isolted nd followed y reverse trnscription s descried previously [6]. Rel-time PCR ws performed in 7500 Sequence Detection System (Life Technologies Corp., Crlsd, CA, USA) with 2 μl smple nd SYBR Green PCR Mster Mix (1 μl DEPC H 2 O, 2 ul primers t 2 μm nd 5 μl SYBR Green). Primers re listed in Tle 2. The product specificity of every rection ws confirmed y melting curve test. The Ct method ws used for ll qpcr results. Chromtin immunoprecipittion ssys (ChIP) SimpleChIP Enzymtic Chromtin IP Kit (Cell Signling, Dnvers, MA, USA) ws used for ChIP ssys. Mnufcturer s protocol ws followed with some modifictions. Nucler proteins in PVN were first cross-linked with 1% formldehyde for 10 min nd quenched y glycine solution. Chromtin ws shered y soniction, using Vircell Sonix (mximl power 750 wtts; Sonics & Mterils Inc, Newtown, CT, USA) t 32% power for 9 rounds of 10 sec soniction with 350 ul SDS lysis uffer (SDS, 1%; EDTA, 10 mm; nd Tris, ph 8.1, 50 mm). Then, 50 ul shered chromtin were incuted with H3 (4 ul), H3K27Me2 (8 ul), H3K27Me3 (8ul), Ezh2 ntiody (5 ul) nd nti-igg (4 ul s mock ckground immunoprecipittion) overnight. Then, the ntiodies were retrieved y using Protein G Agrose Beds nd trget proteins were eluted t 65 C overnight. On the next dy, Protein G Agrose Beds were centrifuged down t 6,000 rpm nd incuted t 65 C overnight with Proteinse K. After purifiction s provided in the kit, ll smples were sujected to rel-time PCR using primers trgeting on different regions of the BDNF gene. Primer design ws listed in Tle 1. The dt were normlized to n input control tht consisted of qpcrs from 1% crosslinked chromtin efore immunoprecipittion nd finl results were reported y Ct methods. Sttisticl nlysis JMP 9.0 (SAS Institute Inc, Cry, NC, USA) ws used for ll dt nlysis. Dt presented were mens ± SEM. For experiment 1, comprisons etween control nd tretment group used unpired t-test for ech single time point. For experiment 2, control, tretment 1 nd tretment 2 comprison used nlysis of vrince (ANOVA), which used the generl liner modeling procedure (GLM, SAS Institute Inc.). Significnt differences were considered s P < 0.05 unless otherwise stted Experimentl design Experiment 1: single fsting on three dys-of-ge (D3) Thirty-six one-dy-old chicks were used to test the effect of single fsting on erly postntl dys. They were rndomly ssigned into two groups, which hd eighteen chicks per group nd mrked s tretment nd control. All chicks hve ccess to food nd wter d liitum. On D3, food ws withdrwn from ll 3 groups for 24 hr in tretment group (clled D3-fsting for short in following prgrph). On D4, six chicks were scrificed y decpittion in ech group while the rest were given d 30

41 liitum ccess to food. On D11 nd D41, six chicks ws scrificed from ech group, respectively. PVN nd forerin (FB) were collected. Experiment 2: doule fsting on oth D3 nd ten dys-of-ge (D10) For D3 nd D10 doule fsting (clled D3/10-fsting in the following prgrph) group, eighteen chicks were rndomly picked nd divided into three groups (six per group). All of them were pplied with food nd wter ville d liitum. One group ws rndomly picked s control group, which would never een fsted. For the other two groups, on D3, food ws withdrwn from them for 24 hr. On D4, food ws ck to norml supply for oth of the fsted groups. On D10, one group of previous fsted chicks were fsted gin for 24hr. All three groups of chicks were scrificed on D11 nd PVN nd FB smples were collected Results Elevtion of glol methyltion t H3K27 in the PVN ut not in the FB ws detected 24 hours following fsting stress. Previous reserch reported tht chicken BDNF gene hs four exons, which generted three vrints with the sme coding re, Exon IV [346]. In our study, we first ligned BDNF gene sequence (GenBnk, ccession numer NM_ ) in the entire chicken genomic sequence (2006, UCSC), which verified the testing region in this study ws Exon IV. For esy of wording, we use BDNF gene insted of BDNF gene, Exon IV for short in rest of the pper. We first mesured the ody weight chnges t different time point etween nive group nd D3-fsting group. We found tht 24 hour fsting on dy 3 posthtch resulted in significnt decrese in ody weight on D4 nd D11, ut not D41 (Tle 1). Then, we were interested in evluting the role of covlent histone modifictions occurring t n erly criticl ge in which the hypothlmic control of feed intke is thought to e estlished. It is known tht chnges in chromtin structure contriute to the regultion of neuronl plsticity [347]. In this study, we concentrted on the methyltion of lysine 27 on histone 3 (H3K27). Using western lot nlyses with specific ntiodies ginst H3K27me2, H3K27me3, nd histone H3, we found tht 24 hour fsting on dy 3 posthtch, resulted in significnt increse in totl histone H3 levels (Fig 1A), H3K27me2 (Fig 1B), nd H3K27me3 (Fig 1C) in the PVN on D4. Totl H3 levels were significntly incresed on D11, ut reduced y Dy41 fter the 24 hour fst on D3. However, no significnt chnges were oserved on D11 or D41 for H3K27me2 or HeK27me3. H3K27 modifictions were lso evluted in n dditionl rin re: the forerin (FB). Totl H3 levels (Fig 1D) were significntly incresed in the D3 fsted chicks on D4, D11, nd D41. However, no significnt differences in H3K27me2 (Fig 1E) or H3K27me3 (Fig 1F) were oserved, s result of fsting on D3. 31

42 Since PRC2 is specific methyltrnsferses for covlent modifiction t H3K27, we next tested the mrna level of key grdients of this complex. As shown in Fig 2, in PVN, the expression of the histone methyltrnsferses SUZ (Fig 2A), EZh2 (Fig 2B), nd EED (Fig 2C) were significntly incresed y D4. While ll these fctors ecme insignificnt in D3-fsted group on D11, SUZ nd EZh2 expression levels were significntly decresed t D41 post fsting. The mrna expression level of these fctors greed with the chnges of H3K27Me2 nd H3K27Me3 in PVN s shown in Fig 1. Compred to the notle chnges of PRC2 nd TRX in PVN, FB remin "quite" to the fsting stress. In FB, no significnt differences ws detected etween D3- fsting nd nive group in SUZ (Fig 2D), EZH2 (Fig 2E), nd EED (Fig. 2F). It greed with the chnges of H3K27Me2 nd H3K27Me3 s shown in Fig 1. BDNF protein nd mrna expression greed with the incresed glol methyltion t H3K27 in the PVN following 24 hrs fsting stress The glol chnges of methyltion t H3K27 could ffect multiple genes' expression. As mentioned erly, BDNF is n norexigenic fctor nd inhiits food intke. In the current study, we were interested in evluting the expression level of BDNF in PVN. Using western lot nlyses with specific ntiodies ginst BDNF protein, we found tht significnt decrese in BDNF protein in the PVN in D4 chicks following the 24 hr fst s compred to nïve chicks (Fig 3A). This follows the pttern of mrna expression which ws significntly decresed in tht group (Fig 3B). Although no significnt decrese in BDNF mrna y D11 or D41, there ws significnt decrese t D11 of the BDNF protein ut no chnges t D41. Compred to PVN, in the FB, there ws lso significnt decrese in the BDNF protein on D4 following the fst (Fig 3C), while BDNF mrna expression ws significntly incresed t D4 (Fig 3D). ON D11, BDNF protein went ck to control level ut BDNF mrna remined significntly decresed on D11 in the fsted chicks. No significnt difference ws detected for oth BDNF protein nd mrna on D41. The inhiited BDNF expression level ws controlled y the distriution of H3K27me2 nd H3K27me3 on Bdnf gene Although the glol methyltion t H3K27 hd pprently ffect the BDNF mrna expression, we hd no ide out the detiled chnges of H3K27 methyltion sttus t Bdnf gene on chromosome. Thus, we used CHIP ssy to detect the dimethyltion nd trimethyltion levels of H3K27 long the Bdnf gene in oth PVN nd FB following D3-fsting on D4, D11, nd D41. Three different regions long the gene were evluted: the Bdnf promoter re (-869 to -801 p upstrem of the coding region); the Bdnf trnscription strt region (+91 to +190 p); nd the Bdnf 3 untrnslted region (+1623 to p). In the PVN, H3K27me2 ws significntly incresed on D4 t the trnscription strt site ut not trnscription strt region nd 3'-UTR (Fig 4B, E& H). By D11, however, the H3K27me2 level ecme significntly incresed t ll res long the Bdnf on D11 post fst. This pttern of H3K27me2 remined on D41. On the other hnd, the ptterns of H3K27me3 vried from tht 32

43 of the H3K27me2 long the Bdnf. After D3-fsting, H3K27me3 hd lredy incresed noticely t ll res long the Bdnf on D4 (Fig 4C, F & I). By D11, long the Bdnf, H3K27me3 level ecme insignificnt in D3-fsted chicks when compred to the nïve chicks. On D41, H3K27me3 gin pprently elevted t Bdnf promoter nd trnscription strt region ut not 3'-UTR. The significnce of totl H3 spred in PVN, which is only shown on D4 t Bdnf promoter nd 3'-UTR nd D41 t Bdnf trnscription strt region. As H3K27 specific methyltion ctlyzer, EZH2 hs een shown to e involved in the methyltion of H3K27 in chicks [7]. We hypothesized tht EZH2 distriution long Bdnf should e comptile to methyltion sttus t H3K27 site. By using CHIP ssy, we first found tht the dynmic chnge ptterns of EZH2 remined constnt long Bdnf in PVN: incresed on D4, dropped on D11 ut elevted gin on D41 (Fig 6A, B & C). This ws comptile to overll similr to chnging pttern in H3K27 methyltion s detected. As shown in Fig 6A, B & C, deposition of EZh2 in the PVN long the Bdnf followed the sme pttern s H3K27me3 on D4 nd D41. However, the EZH2 level showed significnt drop on D11 while H3K27me3 ws insignificnt etween treted nd nive groups. On the other hnd, the H3K27me2 showed constnt comptile pttern to EZH2 on D41 long Bdnf. However, only t Bdnf promoter region tht H3K27me2 nd EZH2 exhiited incresed level. On D11, H3K27me2 exhiited constnt incresed level in D3-fsting group while EZH2 exhiited the reversl. In FB, H3K27me2 remin unchnged long Bdnf fter D3-fsting (Fig 5B, E& H). It ws until D41 tht H3K27me2 showed decresed level t Bdnf promoter nd trnscription strt region ut not 3'-UTR. On the other hnd, the ptterns of H3K27me3 vried from tht of H3K27me2. After D3-fsting, H3K27me3 exhiited n instnt elevtion long Bdnf gene on D4 nd lst till D11 (Fig 5C, F & I). On D41, H3K27me3 ecme indifferent etween D3-fsted nd nive groups. Totl H3 exhiited very similr pttern s tht of methylted H3K27 except totl H3 ws significntly inhiited on D41 long Bdnf in D3- fsted group. Next, we lso compred EZH2 distriution long Bdnf to methyltion sttus on H3K27 in FB. Similr to the PVN, EZH2 overll exhiited very similr distriution ptterns long Bdnf gene in FB: incresed on D4 nd D11, ut went ck to control level on D41 (except D4 t Bdnf promoter re nd D11 t Bdnf 3'-UTR) (Fig 6D, E & F). Agin, similr to the PVN, EZH2 in FB exhiited perfect mtch to H3K27me3. Preconditioned chicks y erly fsting reduced homeostsis disturnce in PVN cused y lte fsting Preconditioned nimls/tissues tends to hve stress tolernce towrds future sme type stress [7,348]. As PVN exhiited significnt nd long-lsting chnges of epigenetic mrkers fter 24 hr on D3, we suspected tht these chnges serve s protective mechnism towrds potentil future fsting stress. Thus, we were interested in seeing if preconditioned chicks tht received n erly fsting chllenge t D3 would hve reduced homeostsis disturnce effects when susequently chllenged gin on dy

44 First, we mesured chnges in ody weight of chicks following D10-fsting or D3/10-fsting. We found tht compred to nïve group, non-pre-conditioned chicks showed higher ody weight level while pre-conditioned chicks showed lower ody weight on D11 (Tle 2). Single fsting on D10 resulted in significntly elevted glol H3, H3K27me2, nd H3K27me3 levels in the PVN compred to nïve irds (Fig 7A). The SUZ, EED, nd EZH2 mrna expression were lso significntly elevted fter 24 hr fsting on D10 when compred to nïve irds (Fig 7B). However, the chicks previously fsted on D3 nd then chllenged gin on D10, showed vried results. Previously conditioned chicks showed further elevted H3 (Fig 7A) nd EZH2 (Fig 7B) compred to the chicks receiving only one fsting chllenge t D10. However, levels of H3K27me2, H3K27me3, SUZ nd EED ecme insignificntly different compred to tht of the nive group (Fig 7A & B). D10-fsting significntly reduced the level of BDNF protein nd its mrna in the PVN (Fig 7C & D), ut previously conditioned chicks tht were susequently fsted gin on D10 showed return of BDNF protein nd mrna ck to control levels (Fig 7C & D). FB exhiited very different story from tht of the PVN. In the FB, fter D10-fsting, only glol H3 levels were significntly incresed, which ws similr to tht seen in previously conditioned nd re-fsted chicks (Fig 8A). Contrry to wht ws oserved in the PVN, SUZ, EED, H3K27Me3 nd EZH2 exhiited significnt increse fter D3/10-fsting compred to tht of D10-fsted chicks (Fig 8A & B). Although D3/10-fsting induced significnt inhiition of BDNF mrna compred to the nive group, no BDNF protein chnge oserved (Fig 8C & D). The distriution of epigenetic mrkers on the Bdnf gene were significntly chnged etween previously conditioned chicks nd those fsted only once on D10. The CHIP results for the PVN greed with the ove western-lot nd qpcr results. Throughout three different tested dnf loci, the pttern remined lmost identicl in the PVN (Fig 9 A, B & C). Totl H3 levels t ech site on the Bdnf gene fter D3/10-fsting ws more elevted from nive level compred to tht of D10-fsting. H3K27me2 nd H3K27me3 levels t ll three sites remined t similr level s nïve controls fter D3/10-fsting while D10- fsting exhiited significntly incresed levels compred to controls in the PVN (except H3K27me3 t Bdnf 3'-UTR). The distriution of EZH2 long Bdnf gene greed with the chnge of H3K27me2 nd H3K27me3 (Fig 10A). All detected epigenetic mrkers on the Bdnf gene in the FB hd lmost the sme distriution ptterns. H3K27me2 ws not ffected y either D3-fsting or D3/10-fsting t ll (Fig 9 D, E & F). However, fter D10-fsting, H3 exhiited significnt decrese while D3/10-fsting significntly incresed the H3 deposition on Bdnf. In ddition, H3K27me3 levels were lso elevted fter D3/10-fsting ut not D10-fsting compred to controls. Although EZH2 ws not ffected y D10-fsting, D3/10- fsting elicited significnt increse compred to the nive group, which ws consistent with tht of H3K27Me3 (Fig 10 B) Discussion 34

45 The studies descried here showed tht fsting erly in life ws le to: (1) elevted the gene expression of PRC2 key suunits nd thus (2) incresed the methyltion modifiction on H3K27 in PVN; (3) Some of these chnges would lst for long period of time; (4) oth BDNF protein nd mrna level were inhiited post fsting tretment, which ws tightly relted to the H3K27me2/me3 deposition long Bdnf gene; (5) helps chicks to develop epigenetic memory towrds future fsting stress. Together, ll these results support the hypothesis tht erly-life fsting would hve long-term effect in re-setting neurochemistry set-point. This effect hd protective mechnism for the ody to preserve endocrine homeostsis. Feeding regultion is determined y the neurl-networking in the five hypothlmic nuclei [51]. Among them, the PVN is considered one of the most importnt nuclei mong them. A histology study showed tht the PVN received innervting fiers from lmost ll res of the hypothlmus [105], which implied pivotl role in regulting physiologicl functions. In ddition, nerly ll known orexigenic [50,106,107,111] nd norexigenic fctors [107,112,113,114] incresed nd decresed feeding, respectively when micro-injected into the PVN. In ddition, multiple numerous neuro-peptides, including BDNF [79], ws lso detected in the PVN. Thus, in the current study, we chose the PVN for our min focus. In previous study of our l, we fsted the Co mle chicks for either 6 hours or 24 hours on 3 dy-of-ge. We reported tht reported tht compred to the non-fsted chick, 6 hours-fsting incresed, ut not significntly, EZH2 protein level in PVN ut not in FB [6]. If fsting stress ws extended to 24 hours, the elevtion of EZH2 protein ecme pprent (P < 0.05) while FB remined no chnged. These chnges were ccompnied y n incresed trend (P > 0.05) of H3K27me2/me3 in PVN ut not in FB. These dt suggested potentil upcoming surge of glol methyltion t H3K27 site. In current study, we further extended the previous study nd otined overll consistent result. We found tht fter 24-hour fsting on D3, oth the FB nd PVN responded showing chnged levels of different epigenetic mrkers glolly. As expected, in the FB, fsting hd limited cpcity to chnge glol epigenetic mrker expression levels. This greed with the fct tht the FB is not feeding regultory center nd not directly involved in feeding regultion. However, for the PVN, single 24 hr fst on D3 induced significntly incresed methyltion modifiction on H3K27. Firstly, this finding suggest tht histone modifiction is highly responsive rther thn sttic towrds nutritionl stresses. Histones re now known to e highly dynmic nd responsive towrds environmentl stress [10,11]. Through covlent modifiction of the histone til, histones regulte the DNA scffold nd regultes gene expression [10,12]. So fr, to the uthor's knowledge, only limited numer of pulictions reported the chnges in histone [6,349] nd nucleotide [350,351] modifictions nd their relted modifiction enzymes [349] in the hypothlmus fter unfvorle nutritionl stress. Thus, our finding first provided evidences to estlish the reltionship etween fsting stress nd direct histone modifiction. Secondry, the mssive inhiitory H3K27me2/me3 deposition on chromtin fter fsting suggested mssive inhiitory chnge in neuropeptides production. This phenomenon could chnge neurochemistry set-point in the hypothlmus nd feeding ehvior. This hypothesis ws supported y the finding of dropped BDNF protein nd mrna level inside PVN fter fsting tretment. Furthermore, 24 hour fsting t D3 surprisingly resulted in long term decresing of PRC2 key 35

46 suunits (SUZ nd EZH2, ut not EED) t lest till 38 dys post fsting. The drop of SUZ nd EED potentilly up-regulte gene trnscription of multiple genes, which cuse imlnce of neurpeptides production inside PVN. One erly study showed seprtion of the pups from their mothers left long term effect on neurochemistry in PVN [352], it is possile tht fsting could hve the sme effect in PVN nd reset the neurochemistry set-point in CNS for long term. Although we report lsting epigenetic modifictions, it remins to e seen whether these chnges re eneficil or hrmful to the chicks. Previous studies in humns reveled tht unfvorle nutritionl conditioning during the neontl criticl-period cn induce cute metolic syndromes [1,2,3] s well s severe diseses in lter life [4,5]. Other studies in nimls suggested tht it helps individuls to cclimte to the sme environmentl chnges nd mintin homeostsis y otining "moleculr memory" [353,354]. In the current study, we chllenged chicks with susequent fsting on D10 in order to determine whether these chnges might e eneficil or hrmful. We found tht erly-life fsting induced eneficil response: "memory" hd een developed inside PVN to the PRC2, ut not FB, to kept ody wy from hrmful response from fsting t lter life period. Interestingly, this memory ws only seen in certin fctors in PRC2: while D10- fsting significntly incresing EED nd SUZ mrna expression, pre-fsted irds did not elevted EED nd SUZ t ll; on the other hnd, EZH2 kept incresing in oth D10- nd D3/10-fsting sitution. PRC2 is huge H3K27 methyltion mchinery [213] nd EZH2, SUZ nd EED re ll locted in its functionl core [213]. If we just monitor EZH2 mrna expression, it would suggest constnt incresing of H3K27 methyltion in oth fsting stress we pplied. On the contrry, H3K27me2/me3 develop the similr "memory" pttern s well fter repetitive fsting stress. Literture review suggested tht EZH2 hd rely no function if presented lone without other non-ctlytic suunits in PRC2 [248]. This is ecuse EZH2 does not hve DNA inding domin, which prevent EZH2 from histone inding [259,355]. However, EED cn ind to histones [255,356] while SUZ connects proteins on chromosome, such s heterochromtin protein 1lph [252]. And EZH2 ws directly linked to EED [251,259,357] nd SUZ [252,259,358] inside PRC2, which forms triplet model tht EZH2 sits in the middle [259]. These linkges insure the EZH2 to hve full contct with histones nd proceed methyltion. Muttion or knocking-out SUZ [259,357,358] nd EED [259,277,359] depleted the function of PRC2. These dt suggest tht oth ctlytic (EZH2) nd non-ctlytic suunits (EED nd SUZ) re pivotl to PRC2. Thus, lthough ody responds to the fsting stress through EZH2-methylting-H3K27, it is indeed vi regulting of EED nd SUZ. However, we re not cler how the ody instills memory towrd SUZ nd EED nd why EZH2 does not. BDNF is initilly known for its function in neuron prolifertion, differentition, growth nd interction [125]. Recently, it is lso ctegorized s n norexigenic fctor [75,159]. Both cute injection [159,160] (including t PVN [161]) or long term intrcereroventricμlr injection [162] of BDNF inhiited feeding nd incresed energy expenditure in tested nimls. While fsting inhiited BDNF level inside in the ventromedil hypothlmus [75] nd dorsl vgl complex of hinder rin [159], refed led recovered BDNF [159]. In humn, femle with either ulimi nervos (BN) or norexi nervos (AN) hd significntly 36

47 lowered serum BDNF level [163]. In ddition, it ws lso reported to regulte other neuropeptide levels in the PVN [360] nd to chnge the ody s metolic rte [161]. All these dt suggested the pivotl role of BDNF in feeding regultion. In the current study, we identify n instnt drop in oth BDNF protein nd mrna level fter fsting stress. This finding greed with the finding from previous studies. But more importntly, we were le to comply BDNF mrna chnges with the H3K27me2/me3 distriution long Bdnf gene. It is widely ccepted tht H3K27 methyltion compct nucleosomes nd thus lock trnscriptionl elongtion on gene templtes [361,362]. Thus, tremendously incresed H3K27me2/me3 fter fsting indicted mssive gene trnscription inhiition in PVN. However, it remined out of clue tht when nd where of H3K27me2/me3 deposition will led to induce specific gene inhiition. We picked three res on Bdnf gene, which re promoter (-869 to -801 p), coding strting region (+91 to +190 p ) nd 3'-UTR side (+1623 to p). Sme s the glol chnges in PVN, instnt incresed level of H3K27 me2/me3 ws deposited long Bdnf gene. EZH2, on the other hnd, ws lso significntly chnges nd comply tht tht of H3K27 methyltion level. This indictes tht fsting could inhiit Bdnf trnscription vi deposition of PRC2 long Bdnf gene. Once PRC2 methyltes H3K27, locl chromtin structure ecome much more compcted, which inhiited gene trnscription. However, we do not understnd how PRC2 could recognize these specific sites. Neither do we know out other modifiction on other residues of other histones. In summry, the present study showed tht fsting in chickens t different life stges could ll elicit cute neurochemicl chnges due to epigenetic imprinting chnges. For the neontl period of chickens, fsting cn lso chroniclly cuse genome-wide epigenetic imprinting chnges. These chnges provide evidences to support the finding from humn epidemiologicl study. In ddition, this epigenetic imprinting serves s moleculr memory in the neurons, which protects cells from mssive interruption of genes trnscription nd keeps individuls in homeostsis fter sme type of fsting stress in lter life. BDNF, which is key norexigenic nd rin-developmentl fctor, verified this conclusion. 37

48 Primer Nme Gene ccession numer Sequence PCR products (p) BDNF NM_ F: 5'-ACTGGCGGACACTTTTGAAC-3' 312 R: 5'-GTTGCACCAGACATGTCCAC-3' BDNF promoter site NM_ F: 5'-TGGTTTTCATGAGGAGCCCT-3' 69 R: 5'-TTTCCCAGAGCCCCATATCA-3' BDNF trnscription strt site NM_ F: 5'-GCTTGGCTTACCCAGGTCTTC-3' 100 R: 5'-TCAAAAGTGTCCGCCAGTG-3' BDNF 3 -untrnslted region NM_ F: 5'-GTCCCCTCCCCTTTCCTCTC-3' 76 R: 5'-CAAGCTCCAGTTGTATGCTGAGTG-3' β-ctin NM_ F: 5 -CCGCAAATGCTTCTAAACCG R: 5 -AAAGCCATGCCAATCTCGTC-3 CBP XM_ F: 5'-GATCCTGAGCCGTTTGTTGACT-3' 139 R: 5'-TTTTCCTTCCGTGTTCTACCAGTT-3' EED NM_ F: 5'-ATACCCCAACGAACACTCCTAATG-3' 138 R: 5'GGACTCCAAACAAAGGCTGACC-3' EZH2 XM_ F: 5'-GCCCTGACTGCTGAGCGTAT-3' 168 R: 5'-GTTTTCTCCTCCAGTTTCTGTTCC-3' SUZ XM_ F: 5'-TTTATCATCCGAAAGGTGCTAGG-3' 127 R: 5'-TTGACTGGTCCATTGCGACTG-3' Tle 2. The primer design for different genes. 38

49 D4 D11 D41 D3-fsting Body weight * * SEM Tle 3. Chnges in chick ody weight following D3-fsting. At three dy-of-ge, irds were fsted for 24 hours (D3-fsted) or fed d liitum (nïve). Nïve controls nd fsted chicks were scrificed on dy 4 (D4), dy 11 (D11), nd dy 41 (D41). Body weight ws mesured efore scrifice. Ech vlue is the men stndrd error of the men of 22 to 24 individul chicks. ****P < indicted significnt difference reltive to D4, D11, nd D41 nïve chicks. There ws n overll significnt decrese of ody weight etween nïve nd D3-fsted chicks on 24 hrs nd 8 dys post fst, ut no difference t 38 dys post fsting. 39

50 D11 D3-fsting D10-fsting Body weight c +SEM Tle 4. Chnges in ody weight of chicks following D10-fsting or D3/10-fsting. Chicks were divided into three groups. Nive groups were fed d liitum nd never fsted. D3/10-fsted irds were pre-conditioned with fsting for 24 hours on 3 dys-of-ge in ddition to 24 hours fsting on 10 dys-of-ge. The non-pre-conditioned irds were fsted for 24 hours only on 10 dys-of-ge (D10-fsting). All groups were scrificed on dy 11 (D11). Body weight were mesured efore scrifices. Ech vlue is the men stndrd error of the men of 22 to 24 individul chicks. Dt red different letters indicte significnt different. There ws n overll significnt increse in the levels of ody weight etween nïve nd non-pre-conditioned chicks on D11. However, ody weight ecme significntly decresed in preconditioned irds compred to non-pre-conditioned irds nd nïve irds. 40

51 Reltive protein quntity of H 3 / - c tin R eltive protein quntity of H 3 K 2 7 m e 2 / - c tin R eltive protein quntity of H 3 K 2 7 m e 3 / - c tin Reltive protein quntity of H 3 / - c tin Reltive protein quntity of H 3 K 2 7 m e 2 / - c tin Reltive protein quntity of H 3 K 2 7 m e 3 / - c tin A B C D 4 D 1 1 D 4 1 D 4 D 1 1 D 4 1 D 3 -f stin g H 3 - c tin + _ + _ + _ D 3 -f stin g H 3 K 2 7 m e 2 - c tin + _ + _ + _ D 3 -f stin g H 3 K 2 7 m e 3 - c tin + D 4 D 1 1 D 4 1 _ * * * * * * * * * * * D 4 D 1 1 D 4 1 D 4 D 1 1 D 4 1 D 4 D 1 1 D 4 1 nive group D 3 -fsted group nive group D 3 -fsted group nive group D 3 -fsted group D E F D 4 D 1 1 D 4 1 D 4 D 1 1 D 4 1 D 3 -f stin g + _ + _ + _ D 3 -f stin g + _ + _ + _ D 3 -f stin g H 3 H 3 K 2 7 m e 2 H 3 K 2 7 m e 3 + D 4 D 1 1 D _ - c tin - c tin - c tin * * * * * D 4 D 1 1 D 4 1 D 4 D 1 1 D 4 1 D 4 D 1 1 D 4 1 nive group D 3 -fsted group nive group D 3 -fsted group nive group D 3 -fsted group 41

52 Figure 1. Chnges in protein levels of H3, H3K27me2 nd H3K27me3 in the chick prventriculr nucleus (PVN) (A, B & C) nd the forerin (FB) (D, E & F) following fsting. At three dys of ge, irds were fsted for 24 hours (D3-fsted) or fed d liitum (nïve). Nïve controls nd fsted chicks were scrificed on dy 4 (D4), dy 11 (D11), nd dy 41 (D41). Totl protein ws isolted nd evluted y Western lot using the ntiody ginst H3, H3K27me2 nd H3K27me3. Blot density were compred with -ctin. Ech vlue is the men stndrd error of the men of 6 to 12 individul chicks. *P 0.05, **P 0.01, **P < nd ****P < indicted significnt difference reltive to D4, D11, nd D41 nïve chicks. (A, B & C) There ws n overll significnt increse in the levels of glol H3 etween nïve nd D3-fsted chicks on 24 hrs nd 8 dys post fst, ut with reversl t 38 dys post fsting. Glol di- nd tri-methylted lysine 27 levels on H3 were significntly incresed t 24 hours post fst, ut y D11 nd D41 no significnt differences were found. (D, E & F) The glol histone 3 levels in the FB were significntly incresed t D4, D11, D41 post fst. However, there ws no significnt differences etween the glol di- nd tri-methylted lysine 27 levels on H3 levels in D4, D11, or D41 post fst. 42

53 r e l tiv e m R N A e x p r e s sio n S U Z / - c tin r e l tiv e m R N A e x p r e s sio n E Z H 2 / - c tin r e l tiv e m R N A e x p r e s sio n E E D / - c tin r e l tiv e m R N A e x p r e s sio n S U Z / - c tin r e l tiv e m R N A e x p r e s sio n E Z H 2 / - c tin r e l tiv e m R N A e x p r e s sio n E E D / - c tin A B C * * * 2.0 * 4 4 * * * D 4 D 1 1 D 4 1 D 4 D 1 1 D 4 1 D 4 D 1 1 D 4 1 n iv e g r o u p D 3 -f ste d g r o u p n iv e g r o u p D 3 -f ste d g r o u p n iv e g r o u p D 3 -f ste d g r o u p D E F D 4 D 1 1 D 4 1 D 4 D 1 1 D 4 1 D 4 D 1 1 D 4 1 n iv e g r o u p D 3 -f ste d g r o u p n iv e g r o u p D 3 -f ste d g r o u p n iv e g r o u p D 3 -f ste d g r o u p 43

54 Figure 2. Chnges in mrna levels of histone methyltrnsferses (HMTs) of the Polycom repressive complex 2 (PRC2) nd histone cetyltrnsferse (HAT) in the chick prventriculr nucleus (PVN) (A, B & C) nd the forerin (FB) following fsting (D, E & F). At 3 dys of ge, chicks were fsted for 24 hours (D3-fsted) or fed d liitum (nïve). Nïve controls nd fsted chicks were scrificed on dy 4 (D4), dy 11 (D11), nd dy 41 (D41). Totl mrna ws isolted nd evluted y rel-time polymerse chin rection (PCR) using the Syr green method with SUZ, EZh2 nd EED gene-specific primers (see Tle 1). Gene expression levels were compred with -ctin. Ech vlue is the men stndrd error of the men of 6 to 12 individul chicks. *P 0.05, **P 0.01, **P < nd ****P < indicte significnt difference reltive to D4, D11, nd D41 nïve chicks. Arevitions: SUZ (suppressor of zeste), EZh2 (enhncer of zeste 2); EED (emryonic ectoderm development). (A, B & C) In PVN, there ws n overll significnt increse in the expression of SUZ, EZH2 nd EED mrna in D3-fsted chicks compred to nïve group on 24 hrs post fst. On 8 dys post fst, no significnt differences were found. However, reversl t 38 dys post fsting ws found. (D, E & F) FB did not respond to the D3-fsting y showing no chnges in glol SUZ, EZH2 nd EED mrna level till 38 dys post fst. 44

55 R e l tiv e p r o te in q u n tity o f B D N F / - c tin r e l tiv e m R N A e x p r e s sio n B D N F / - c tin R e l tiv e p r o te in q u n tity o f B D N F / - c tin r e l tiv e m R N A e x p r e s sio n B D N F / - c tin A B D 4 D 1 1 D 4 1 D 3 -f stin g B D N F + _ + _ + _ c tin 1.5 * * * D 4 D 1 1 D 4 1 n iv e g r o u p D 3 -f ste d g r o u p 0.0 D 4 D 1 1 D 4 1 n iv e g r o u p D 3 -f ste d g r o u p C D 4 D 1 1 D 4 1 D D 3 -f stin g B D N F + _ + _ + _ 4 - c tin * * * * * D 4 D 1 1 D 4 1 n iv e g r o u p D 3 -f ste d g r o u p 0 D 4 D 1 1 D 4 1 n iv e g r o u p D 3 -f ste d g r o u p 45

56 Figure 3. Chnges in protein nd mrna levels of BDNF in the chick prventriculr nucleus (PVN) (A & B) nd the forerin (FB) (C & D) following fsting. At three dys of ge, irds were fsted for 24 hours (D3-fsted) or fed d liitum (nïve). Nïve controls nd fsted chicks were scrificed on dy 4 (D4), dy 11 (D11), nd dy 41 (D41). Totl protein ws isolted nd evluted y Western lot using the ntiody ginst BDNF. Blot density were compred with -ctin. Totl mrna ws isolted nd evluted y rel-time polymerse chin rection (PCR) using the Syr green method with Bdnf gene-specific primer (see Tle 1). Gene expression levels were compred with -ctin. Ech vlue is the men stndrd error of the men of 6 to 12 individul chicks. *P 0.05 nd ***P < indicted significnt difference reltive to D4, D11, nd D41 nïve chicks. (A & B) In PVN, there ws n overll significnt decrese in the levels of BDNF protein nd mrna in D3-fsted chicks compred to nïve group on 24 hrs post fst. While BNDF protein remined decresed in D3-fsted chicks on 8 dys post fst, BDNF mrna in D3-fsted chicks ecme indifferent compred to nïve group. On D41, no significnt differences were found in oth BDNF protein nd mrna. (C & D) In FB, there ws n overll significnt decrese nd increse in the levels of BDNF protein nd mrna in D3-fsted chicks, respectively, compred to nïve group on 24 hrs post fst. While no difference ws found in BNDF protein on 8 dys post fst, BDNF mrna in D3-fsted chicks ecme significntly decresed in D3-fsted chicks compred to nïve group. On D41, no significnt differences were found in oth BDNF protein nd mrna. 46

57 P C R undnce reltive to input P C R undnce reltive to input P C R undnce reltive to input P C R undnce reltive to input P C R undnce reltive to input P C R undnce reltive to input P C R undnce reltive to input P C R undnce reltive to input P C R undnce reltive to input A B C H 3 t B d n f p r o m o te r r e H 3 K 2 7 m e 2 t B d n f p r o m o te r r e H 3 K 2 7 M e 3 t B d n f p r o m o te r r e * * * * 1.5 * 8 4 * * * * * * 6 * * * * * * D 4 D 1 1 D 4 1 D 4 D 1 1 D 4 1 D 4 D 1 1 D 4 1 nive group D 3 -fsted group nive group D 3 -fsted group nive group D 3 -fsted group D E F H 3 H 3 K 2 7 m e 2 H 3 K 2 7 m e 3 t th e e g in n in g o f th e B d n f c o d in g r e g io n t th e e g in n in g o f th e B d n f c o d in g r e g io n t th e e g in n in g o f th e B d n f c o d in g r e g io n * * * * * * * * * * * 1.0 * * * * D 4 D 1 1 D 4 1 D 4 D 1 1 D 4 1 D 4 D 1 1 D 4 1 nive group D 3 -fsted group nive group D 3 -fsted group nive group D 3 -fsted group G H 3 H H 3 K 2 7 m e 2 I t th e B d n f 3 ' - u n tr n sl te d r e g io n t th e B d n f 3 ' - u n tr n sl te d r e g io n H 3 K 2 7 m e 3 t th e B d n f 3 ' - u n tr n sl te d r e g io n * * 2.0 * * * * * * * * 2.0 * * D 4 D 1 1 D 4 1 D 4 D 1 1 D 4 1 D 4 D 1 1 D 4 1 nive group D 3 -fsted group nive group D 3 -fsted group nive group D 3 -fsted group 47

58 Figure 4. Altertions in dimethyltion nd trimethyltion levels of histone H3 lysine 27 (H3K27) long the Bdnf gene in the prventriculr nucleus (PVN) following 24 hour fst on dy 3. To ssess the histone modifictions present t the Bdnf gene, chromtin immunoprecipittion (ChIP) ssys were performed. PVN smples of chicks fsted on dy 3 or nïve controls were tken from 4 dy old, 11 dy old nd 41 dy old chicks nd were immunoprecipitted with ntiodies ginst H3, dimethyl H3K27 (H3K27me2), trimethyl H3K27 (H3K27me3) nd IgG (s negtive control). (A, B & C). PCR results with Bdnf-specific primers (see Tle 1) ligning t the promoter region nd producing mplicon -869 to -801 p upstrem of the coding region. (D, E & F). PCR results with Bdnf-specific primers ligning t the trnscription strt site nd producing mplicon +91 to +190 p. (G, H nd I). PCR results with Bdnf-specific primers ligning t the 3 -untrnslted region (3 -UTR) nd producing mplicon to p. Ech vlue (PCR undnce reltive to input) is the men stndrd error of the men of 6 individul chicks deduct the ckground signl (IgG) nd then normlized to tht of ge comprle control nïve chicks. *P 0.05, ** P 0.01, nd *** P0.001 indicted significnt difference reltive to nïve chicks of the sme ge. 48

59 P C R undnce reltive to input P C R undnce reltive to input P C R undnce reltive to input P C R undnce reltive to input P C R undnce reltive to input P C R undnce reltive to input P C R undnce reltive to input P C R undnce reltive to input P C R undnce reltive to input A B C H 3 t B d n f p r o m o te r r e H 3 K 2 7 m e 2 t B d n f p r o m o te r r e H 3 K 2 7 M e 3 t B d n f p r o m o te r r e 2.5 * * * * * 2.0 * * * * 1.5 * 2 * * D 4 D 1 1 D D 4 D 1 1 D D 4 D 1 1 D 4 1 nive group D 3 -fsted group nive group D 3 -fsted group nive group D 3 -fsted group D E F H 3 H 3 K 2 7 m e 2 t th e e g in n in g o f th e B d n f c o d in g r e g io n t th e e g in n in g o f th e B d n f c o d in g r e g io n H 3 K 2 7 m e 3 t th e e g in n in g o f th e B d n f c o d in g r e g io n 2.5 * * 3 3 * 2.0 * * * * * D 4 D 1 1 D 4 1 D 4 D 1 1 D 4 1 D 4 D 1 1 D 4 1 nive group D 3 -fsted group nive group D 3 -fsted group nive group D 3 -fsted group G H 3 H H 3 K 2 7 m e 2 I t th e B d n f 3 ' - u n tr n sl te d r e g io n t th e B d n f 3 ' - u n tr n sl te d r e g io n H 3 K 2 7 m e 3 t th e B d n f 3 ' - u n tr n sl te d r e g io n * * * * * * * D 4 D 1 1 D 4 1 D 4 D 1 1 D 4 1 D 4 D 1 1 D 4 1 nive group D 3 -fsted group nive group D 3 -fsted group nive group D 3 -fsted group 49

60 Figure 5. Altertions in dimethyltion nd trimethyltion levels of histone H3 lysine 27 (H3K27) long the Bdnf gene in the forerin (FB) following 24 hour fst on dy 3. To ssess the histone modifictions present t the Bdnf gene, chromtin immunoprecipittion (ChIP) ssys were performed. FB smples of chicks fsted on dy 3 or nïve controls were tken from 4 dy old, 11 dy old nd 41 dy old chicks nd were immunoprecipitted with ntiodies ginst H3, dimethyl H3K27 (H3K27me2), trimethyl H3K27 (H3K27me3) nd IgG (s negtive control). (A, B & C). PCR results with Bdnf-specific primers (see Tle 1) ligning t the promoter region nd producing mplicon -869 to -801 p upstrem of the coding region. (D, E & F). PCR results with Bdnf-specific primers ligning t the trnscription strt site nd producing mplicon +91 to +190 p. (G, H nd I). PCR results with Bdnf-specific primers ligning t the 3 -untrnslted region (3 -UTR) nd producing mplicon to p. Ech vlue (PCR undnce reltive to input) is the men stndrd error of the men of 6 individul chicks deduct the ckground signl (IgG) nd then normlized to tht of ge comprle control nïve chicks. *P 0.05, ** P 0.01, nd *** P0.001 indicted significnt difference reltive to nïve chicks of the sme ge. 50

61 P C R u n d n c e r e l tiv e to in p u t P C R u n d n c e r e l tiv e to in p u t P C R u n d n c e r e l tiv e to in p u t P C R undnce reltive to input P C R undnce reltive to input P C R undnce reltive to input A B C 3 0 E Z H 2 t B d n f p r o m o te r r e * * * * 5 0 E Z H 2 t th e B d n f 3 ' - u n tr n sl te d r e g io n 2 0 E Z H 2 t th e B d n f 3 ' - u n tr n sl te d r e g io n 4 0 * * * * * * * * 2 0 * * * * * 2 0 * * * * * * * * * * * * * D 4 D 1 1 D 4 1 D 4 D 1 1 D 4 1 D 4 D 1 1 D 4 1 nive group D 3 -fsted group nive group D 3 -fsted group nive group D 3 -fsted group D E F E Z H 2 E Z H 2 t B d n f p r o m o te r r e t th e e g in n in g o f th e B d n f c o d in g r e g io n E Z H 2 t th e B d n f 3 ' - u n tr n s l te d r e g io n * * * * D 4 D 1 1 D 4 1 D 4 D 1 1 D 4 1 D 4 D 1 1 D 4 1 nive grou p D 3 -fsted grou p nive grou p D 3 -fsted grou p nive grou p D 3 -fsted grou p 51

62 Figure 6. Altertions in EZH2 long the Bdnf gene in the PVN nd FB following 24 hour fst on dy 3. To ssess the histone modifictions present t the Bdnf gene, chromtin immunoprecipittion (ChIP) ssys were performed. PVN (A, B & C) nd FB (E, E & F) smples of chicks fsted on dy 3 or nïve controls were tken from 4 dy old, 11 dy old nd 41 dy old chicks nd were immunoprecipitted with ntiodies ginst EZH2 nd IgG (s negtive control). PCR results with Bdnf-specific primers (see Tle 1) ligning t the promoter region nd producing mplicon -869 to -801 p upstrem of the coding region p in PVN (A) nd FB (E) smples.. PCR results with Bdnf-specific primers ligning t the trnscription strt site nd producing mplicon +91 to +190 p in PVN (B) nd FB (F) smples. PCR results with Bdnf-specific primers ligning t the 3 -untrnslted region (3 -UTR) nd producing mplicon to p in PVN (C) nd FB (G) smples.. Ech vlue (PCR undnce reltive to input) is the men stndrd error of the men of 6 individul chicks deduct the ckground signl (IgG) nd then normlized to tht of ge comprle control nïve chicks. *P 0.05, ** P 0.01, nd **** P indicted significnt difference reltive to nïve chicks of the sme ge. 52

63 R e l tiv e p r o te in q u n tity o f B D N F / - c tin r e l tiv e m R N A e x p r e s sio n B D N F / - c tin R e l tiv e p r o te in q u n tity o f H 3, H 3 K 2 7 m e 2 & H 3 K 2 7 m e 3 / - c tin reltive m R N A exp ression E E D, S U Z & E Z H 2 / - c tin A B 6 3 * * * * * * * H 3 H 3 K 2 7 m e 2 H 3 K 2 7 m e 3 n ive 24h -fsting on D 10 24h -fsting on D 3 nd D 10 0 n iv e E E D S U Z E Z H h -f stin g o n D h -f stin g o n D 3 n d D 1 0 C D D 3 -f stin g + D 1 0 -f s tin g _ + + B D N F - c tin n iv e 2 4 h -f s tin g o n D 1 0 n iv e 2 4 h -f s tin g o n D h -f s tin g o n o th D 3 n d D h -f s tin g o n o th D 3 n d D

64 Figure 7. Chnges in protein levels of H3, H3K27me2, H3K27me3 (A) nd BDNF (C) nd mrna level EZH2, EED, SUZ, CBP (B) nd BDNF (D) in the chick prventriculr nucleus (PVN) following D3/10-fsting. Chicks were divided into three groups. Nive groups were fed d liitum nd never fsted. D3/10-fsted irds were pre-conditioned with fsting for 24 hours on 3 dys-of-ge in ddition to 24 hours fsting on 10 dys-of-ge. The non-pre-conditioned irds were fsted for 24 hours only on 10 dys-of-ge (D10-fsting). All groups were scrificed on dy 11 (D11). Totl protein ws isolted nd evluted y Western lot using the ntiody ginst H3, H3K27me2, H3K27me3 nd BDNF. Blot density were compred with -ctin. Ech vlue is the men stndrd error of the men of 6 to 12 individul chicks. Totl mrna ws isolted nd evluted y rel-time polymerse chin rection (PCR) using the Syr green method with SUZ, EZh2, EED, nd CBP gene-specific primers (see Tle 1). Gene expression levels were compred with -ctin. Ech vlue is the men stndrd error of the men of 6 to 12 individul chicks. Arevitions: SUZ (suppressor of zeste), EZh2 (enhncer of zeste 2); EED (emryonic ectoderm development); CBP (CREB [camp response element inding] protein). α: P 0.05, β: P 0.01, χ: P < nd δ: P < , indicted significnt difference etween nive nd treted group. *P 0.05, **P 0.01, **P < 0.001, ***P < nd ****P < indicted significnt difference etween D10-fsting nd D3/10-fsting group. Column red different letters indicte significnt different. (A) There ws n overll significnt increse in the levels of glol H3, H3K27me2 nd H3K27me3 etween nïve nd non-pre-conditioned chicks on D11. Similrly, significnt increse of glol H3 lso existed etween nïve nd D3/10-fsted chicks on D11. However, glol H3K27me2 nd H3K27me3 ecme significntly decresed preconditioned irds compred to non-pre-conditioned irds nd exhiit no difference compred to nïve irds. (B) There ws n overll significnt increse in the expression of EED, CBP nd EZH2 mrna etween nïve nd non-pre-conditioned irds. Similrly, pre-conditioned irds lso showed increse of EZH2 mrna compred to nive nd ws even higher thn non-preconditioned irds on D11. However, EED nd SUZ ecme significntly decresed in preconditioned irds compred to nonpre-conditioned irds nd exhiit no difference compred to nïve irds. (C & D) In PVN, there ws significnt decrese in the levels of oth BDNF protein (P < ) nd mrna (P < 0.05) in D10-fsted chicks compred to nïve group on 24 hrs post fst. However, if pre-conditioned chicks with n dditionl 24 hours fsting on 3 dys-of-ge, oth BDNF protein nd mrna ecme significnt incresed compred to D10-fsted chicks nd no different compred to control. 54

65 R e l tiv e p r o te in q u n tity o f B D N F / - c tin r e l tiv e m R N A e x p r e s sio n B D N F / - c tin R e l tiv e p r o te in q u n tity o f H 3, H 3 K 2 7 m e 2 & H 3 K 2 7 m e 3 / - c tin reltive m R N A exp ression E E D, S U Z & E Z H 2 / - c tin A B * 4 * * H 3 H 3 K 2 7 m e 2 H 3 K 2 7 m e 3 n ive 24h -fsting on D 10 24h -fsting on D 3 nd D 10 0 n iv e E E D S U Z E Z H h -f stin g o n D h -f stin g o n D 3 n d D 1 0 C D D 3 -f stin g + D 1 0 -f s tin g _ + + B D N F - c tin n iv e 2 4 h -f s tin g o n D h -f s tin g o n o th D 3 n d D n iv e 2 4 h -f s tin g o n D h -f s tin g o n o th D 3 n d D

66 Figure 8. Chnges in protein levels of H3, H3K27me2, H3K27me3 (A) nd BDNF (C) nd mrna level EZH2, EED, SUZ, CBP (B) nd BDNF (D) in the chick forerin (FB) following D10-fsting or D3/10-fsting. Chicks were divided into three groups. Nive groups were fed d liitum nd never fsted. D3/10-fsted irds were pre-conditioned with fsting for 24 hours on 3 dys-of-ge in ddition to 24 hours fsting on 10 dys-of-ge. The non-pre-conditioned irds were fsted for 24 hours only on 10 dys-of-ge (D10-fsting). All groups were scrificed on dy 11 (D11). Totl protein ws isolted nd evluted y Western lot using the ntiody ginst H3, H3K27me2, H3K27me3 nd BDNF. Blot density were compred with -ctin. Ech vlue is the men stndrd error of the men of 6 to 12 individul chicks. Totl mrna ws isolted nd evluted y rel-time polymerse chin rection (PCR) using the Syr green method with SUZ, EZh2, EED, nd CBP gene-specific primers (see Tle 1). Gene expression levels were compred with -ctin. Ech vlue is the men stndrd error of the men of 6 to 12 individul chicks. Arevitions: SUZ (suppressor of zeste), EZh2 (enhncer of zeste 2); EED (emryonic ectoderm development); CBP (CREB [camp response element inding] protein). α: P 0.05, β: P 0.01 nd χ: P < 0.001, indicted significnt difference etween nive nd treted group. *P 0.05, **P 0.01, **P < 0.001, ***P < nd ****P < indicted significnt difference etween D10-fsting nd D3/10-fsting group. Column red different letters indicte significnt different. (A) There ws n overll significnt increse in the levels of glol H3, ut not H3K27me2 nd H3K27me3 etween nïve nd non-pre-conditioned chicks on D11. Significnt increse of glol H3 lso existed etween nïve nd D3/10-fsted chicks on D11. While, glol H3K27me2 ws not chnged y D3/10-fsting, H3K27me3 ecme significntly incresed in preconditioned irds compred to non-pre-conditioned irds nd nïve irds. (B) There ws overll no difference in the expression of EED, CBP nd EZH2 mrna etween nïve nd non-pre-conditioned irds. However, pre-conditioned irds incresed EZH2, EED nd SUZ mrna compred to nive. SUZ mrna level ws significnt incresed lso in pre-conditioned irds compred to non-pre-conditioned irds. (C & D) In FB, there ws no significnt chnges in the levels of either BDNF protein or mrna mong three groups. 56

67 P C R undnce reltive to input P C R undnce reltive to input P C R undnce reltive to input P C R undnce reltive to input P C R undnce reltive to input P C R undnce reltive to input A B C B d n f p r o m o te r r e B d n f c o d in g e g in n in g r e g io n B d n f 3 ' - u n tr n s l te d r e g io n * * * * * * * 4 * * * * * * 2 * * * * 4 3 * * 4 3 * * 1 * H 3 H 3 K 2 7 m e 2 H 3 K 2 7 m e 3 H 3 H 3 K 2 7 m e 2 H 3 K 2 7 m e 3 H 3 H 3 K 2 7 m e 2 H 3 K 2 7 m e 3 n iv e 2 4 h -f stin g o n D h -f stin g o n D 3 n d D 1 0 n iv e 2 4 h -f stin g o n D h -f stin g o n D 3 n d D 1 0 n iv e 2 4 h -f stin g o n D h -f stin g o n D 3 n d D 1 0 D E F B d n f p r o m o te r r e B dn f coding eginning region B d n f 3 ' - u n tr n s l te d r e g io n 3 * 3 3 * * * * * 2 * * * * 2 2 * * * * H 3 H 3 K 2 7 m e 2 H 3 K 2 7 m e 3 H 3 H 3 K 2 7 m e 2 H 3 K 2 7 m e 3 H 3 H 3 K 2 7 m e 2 H 3 K 2 7 m e 3 n iv e 2 4 h -f stin g o n D h -f stin g o n D 3 n d D 1 0 n iv e 2 4 h -f stin g o n D h -f stin g o n D 3 n d D 1 0 n iv e 2 4 h -f stin g o n D h -f stin g o n D 3 n d D

68 Figure 9. Altertions in dimethyltion nd trimethyltion levels of histone H3 lysine 27 (H3K27) long the Bdnf gene in the prventriculr nucleus (PVN) (A, B & C) nd forerin (FB) (D, E & F) following D10-fsting or D3/10-fsting. Chicks were divided into three groups. Nive groups were fed d liitum nd never fsted. D3/10-fsted irds were pre-conditioned with fsting for 24 hours on 3 dys-of-ge in ddition to 24 hours fsting on 10 dys-of-ge. The non-pre-conditioned irds were fsted for 24 hours only on 10 dys-of-ge (D10-fsting). All groups were scrificed on dy 11 (D11). To ssess the histone modifictions present t the Bdnf gene, chromtin immunoprecipittion (ChIP) ssys were performed. The PVN nd FB smples of chicks ll three groups were tken from 11 dys-of-ge chicks nd were immunoprecipitted with ntiodies ginst H3, dimethyl H3K27 (H3K27me2), trimethyl H3K27 (H3K27me3) nd IgG (s negtive control). (A & D). PCR results with Bdnf-specific primers (see Tle 1) ligning t the promoter region nd producing mplicon -869 to -801 p upstrem of the coding region. (B & E). PCR results with Bdnf-specific primers ligning t the trnscription strt site nd producing mplicon +91 to +190 p. (C & F). PCR results with Bdnf-specific primers ligning t the 3 -untrnslted region (3 -UTR) nd producing mplicon to p. Ech vlue (PCR undnce reltive to input) is the men stndrd error of the men of 6 individul chicks deduct the ckground signl (IgG) nd then normlized to tht of ge comprle control nïve chicks. α: P 0.05, β: P 0.01 nd χ: P < 0.001, indicted significnt difference etween nive nd treted group. *P 0.05, **P 0.01, **P < 0.001, ***P < nd ****P < indicted significnt difference etween D10-fsting nd D3/10-fsting group. 58

69 P C R u n d n c e r e l tiv e to in p u t P C R u n d n c e r e l tiv e to in p u t A B E Z H 2 d istr i u tio n lo n g B d n f E Z H 2 d istr i u tio n lo n g B d n f 3 * * * * 4 * * * * * * * * * p r o m o te r r e c o d in g e g in n in g r e g io n 3 ' - u n tr n sl te d r e g io n p r o m o te r r e c o d in g e g in n in g r e g io n 3 ' - u n tr n sl te d r e g io n n iv e 2 4 h -f stin g o n D h -f stin g o n D 3 n d D 1 0 n iv e 2 4 h -f stin g o n D h -f stin g o n D 3 n d D

70 Figure 10. Altertions in EZH2 long the Bdnf gene in the prventriculr nucleus (PVN) (A) nd forerin (FB) (B) following D10-fsting or D3/10-fsting. Chicks were divided into three groups. Nive groups were fed d liitum nd never fsted. D3/10-fsted irds were pre-conditioned with fsting for 24 hours on 3 dys-of-ge in ddition to 24 hours fsting on 10 dys-of-ge. The non-pre-conditioned irds were fsted for 24 hours only on 10 dys-of-ge (D10-fsting). All groups were scrificed on dy 11 (D11). To ssess the histone modifictions present t the Bdnf gene, chromtin immunoprecipittion (ChIP) ssys were performed. The PVN nd FB smples of chicks ll three groups were tken from 11 dys-of-ge chicks nd were immunoprecipitted with ntiodies ginst EZH2 nd IgG (s negtive control). Promoter region represents PCR results with Bdnfspecific primers (see Tle 1) ligning t the promoter region nd producing mplicon -869 to -801 p upstrem of the coding region. Coding eginning region represents PCR results with Bdnf-specific primers ligning t the trnscription strt site nd producing mplicon +91 to +190 p. 3'-untrnslted region (3 -UTR) represents PCR results with Bdnf-specific primers ligning t the 3 -UTR nd producing mplicon to p. Ech vlue (PCR undnce reltive to input) is the men stndrd error of the men of 6 individul chicks deduct the ckground signl (IgG) nd then normlized to tht of ge comprle control nïve chicks. α: P 0.05, β: P 0.01 nd χ: P < 0.001, indicted significnt difference etween nive nd treted group. *P 0.05, **P 0.01, **P < 0.001, ***P < nd ****P < indicted significnt difference etween D10-fsting nd D3/10-fsting group. 60

71 CHAPTER 3: EZH2 ntisense oligonucleotides inhiits EZH2 methyltrnsferse expression nd locks the fstingtolernce cquisition Ying Jing, Cynthi J. Denow, Michel D. Denow Astrct Both cute metolic syndromes nd severe diseses in lter life of humn hve een relted to the unfvorle nutritionl conditions during their neontl criticl-period. This oservtion hs een tightly relted to the epigenetic chnges on vrious genes controlling feeding regultory fctors in the feeding regultion centers, such s the prventriculr nucleus (PVN). A previous studies showed tht 24 hr fsting t 3 dy-of-ge (D3) cused significnt increse in methyltion modifiction t the lysine 27 residue (K27) of the histone 3 til (H3) in the PVN. This process is minly ctlyzed y enhncer of zeste 2 (EZH2) protein in polycom repressive complex 2 (PRC2). In the current study, we investigted the importnce of EZH2 in feed intke regultion y the knockdown of EZH2 expression. We found tht 24 hours post lterl ventricle injection, EZH2 ntisense oligonucleotides (1μg/μl injection concentrtion), ut not sirna, led to significnt inhiition of EZH2 protein in oth forerin (45% inhiition) nd PVN (86% inhiition) of 3 dy-of-ge (D3) chicks. Then, on D3, chicks were sujected to 24 hour fsting stress (D3-fsting) following either EZH2 ntisense or rtificil cererospinl fluid (ACSF) injection. EZH2 ntisense locked the surge of H3K27 methyltion fter D3-fsting. Brin-derived neurotrophic fctor (BDNF), n norexigenic fctor, exhiited elevted expression nd less methylted H3K27 deposition long the Bdnf gene. A previous studieslso reported fstingtolernce cquisition fter repetitive fsting stress on oth D3 nd D10, which we clled "epigenetic memory". However, fter EZH2 ntisense injection, chicks' PVN no longer exhiited "epigenetic memory" to repetitive fsting stress. These findings suggest tht neurons in the PVN utilize PRC2 s mjor H3K27 methyltion tool. Knockdown of EZH2 in PRC2 impired the response of the PVN to fsting stress nd the ility to cclimte to repetitive fsting stresses. Thus, EZH2 is n importnt H3K27 methyltrnsferse to mintin homeostsis. 61

72 3.1. Introduction The developing rin is highly sensitive to environmentl influences. Unfvorle nutrition is one kind of stresses tht cn cuse cute metolic disorders during the neontl period [1,2,3] nd severe diseses in lter life [4,5]. These erly life experiences occur during heightened periods of rin plsticity nd help determine the lifelong structurl nd functionl spects of rin nd ehvior. In humns, for exmple, weight gin during the first week of life fter irth incresed the propensity for developing oesity severl decdes lter [5]. This susceptiility is, if not ll, relted to the dynmic reversile epigenetic imprints left on the histones [6,7,8], especilly during the prentl nd postprtum period [9]. Histones re highly dynmic nd responsive to environmentl stress [10,11]. Through covlent modifiction of the histone til, histones lter the DNA scffold nd regulte gene expression [10,12]. Vrious types of post trnsltionl modifiction hve een identified on vrious histones tils [12]. Among them, the methyltion nd cetyltion on lysine residue (K) 27 on histone 3 (H3) hs een tightly linked to gene repression [13,14] nd ctivtion [15], respectively. The Polycom group (PcG) proteins re groups of evolutionrily conserved proteins, which re essentil for genes silencing vi histone methyltion [245]. SUZ (suppressor of zeste), EZh2 (enhncer of zeste 2) nd EED (emryonic ectoderm development) ll elong to this group. These three proteins form the functionl core of polycom repressive complex 2 (PRC2), which put di- nd tri-methyltion group onto H3K27 [242,243,247,248]. EZH2 otined the SET (Su[vr]3-9; E[z]; Trithorx) nd perform s methyltrnsferses [237]. On the other hnd, lthough lcking of ctlytic ility, SUZ [259] nd EED [256,257] ply key to help EZH2 to finish methyltion rection. On the other hnd, the Trithorx group (TrxG) proteins counterct the function PcG group [294,363], which ctivte gene trnscription. Similr to tht of EZH2, the CBP (CREB [camp response element inding] protein) elongs to TrxG nd is the ctlytic suunit of TRX, which cetyltes H3K27 [364]. Brin-derived neurotrophic fctor (BDNF) is widely expressed in vrious regions of rin ut hs the highest concentrtion in the hypothlmus [78,155]. Although initilly known for its pivotl role in rin development nd plsticity [78], BDNF hd lso een identified s n norexigenic fctor [156]. Previous studies showed tht nimls exhiited feeding suppression [161,341] nd weight loss [162] fter BDNF cute nd long term ppliction, respectively. From nother point of view, fsting lso inhiited BDNF mrna level in severl res of the hypothlmus [75]. Humn suffered severe hyperphgi nd oesity lso hd een reported to otin mutnt Bdnf gene [165]. Previous studies reported tht erly-life events ffect the Bdnf gene expression nd thus chnged ehviorl outcome of nimls, including different propensity for socil interction [365,366] nd levels for stress responsivity nd nxiety [343,344]. This Bdnf gene expression hd een tightly relted to the epigenetic imprints on Bdnf gene, including dynmic deposition of modified histones [7,343,344]. However, no studies hve shown ny chnges t the Bdnf locus fter fsting stress. The hypothlmus, crucil structure in the rin, is involved in mny physiologicl roles of regultion, including feeding. The hypothlmic prventriculr nucleus (PVN) is one of the most importnt feeding regulting nuclei [342]. The 62

73 micro-injection of orexigenic fctors [50] into the PVN could instntly increse feeding nd vice vers [107,112,113,114,115]. BDNF is lso pivotl feed regultor in the PVN. Not only the PVN hd high expression level of BDNF [79], ut microinjection of BDNF into the PVN inhiited feeding [161]. A previous study reported the genome-wide incresing of EZH2 in PVN fter fsting in neontl chicks [6]. However, little is known out the mechnism leding to chnges in PRC2 nd its products. The oligodeoxyrionucleotides is smll piece of DNA molecule [367,368]. Compred to norml DNA oligonucleotides, the oligodeoxyrionucleotide hd nonridging oxygen tom replced y sulfur tom t every single phosphorus site, which mkes them highly resistnt towrds endonucleses [368]. The dditionl phosphorothiote modifiction on the oligodeoxyrionucleotides mkes it more stle nd resistnt to endonucleses degrdtion [368]. In theory, fter uptken y cells, oligodeoxyrionucleotides hyridize to its specific complementry nucleotides inside cells. This hyridiztion could e used to inhiit trget gene either y triggering the RNse H to hydrolyze mrnas [369,370,371], locking trnscription inititor progression [372] or inhiiting mrna splicing [368]. Becuse of the effectives of oligodeoxyrionucleotides, multiple clinicl therpeutic trils hve een conducted to test effectiveness of oligodeoxyrionucleotides in cncer tretment [373,374,375] We previously demonstrted role for epigenetic regultion involved in fsting stress. We showed tht 24 hours fsting on 3 dy-of-ge chick significntly incresed PRC2 suunits nd its product, which lst t lest for 38 dys. In ddition, erly fsting stress ws le to induce "moleculr memory" in certin suunit of PRC2 (SUZ nd EED, ut not EZH2) inside PVN neurons, which provided protection from fsting lter in life. These dt confirmed the importnce of EZH2 in fsting relted histone methyltion regultion. In the present study, we investigted the importnce of EZH2 y knocking down EZH2 proteins. Our dt shown tht EZH2 knockdown locked the "norml" response towrds erly fsting stress in the PVN. In ddition, the PVN lost "moleculr memory" towrds fsting stress. 63

74 3.2. Mterils nd methods Animls In the current study, postntl dy 0 ws defined s the first dy of htching. Mle White Plymouth Rock chickens were otined on one dy-of-ge (D1) from the Co-Vntress Htchery in Wdesoro, NC, USA. When trnsported to VT, chicks were first housed in Petersime tteries in groups of ten. The room temperture ws set t 30 ± 2 C nd reltive humidity t 50 ± 5%. Wter nd food were fed d liitum. The Virgini Tech Animl Cre nd Use Committee hd pproved ll experimentl procedures. Chemicls Anti-Histone 3 (H3), nti-tri-methylted Histone H3 t Lysine 27 (H3K27Me3), nti-di-methylted Histone H3 t Lysine 27 (H3K27me3) H3K27Me2, nti-β-ctin ntiodies nd chromtin immunoprecipittion ssy kits were purchsed from Cell Signling Technology (Dnvers, MA, USA). Antiody purchsed from Acm Inc. (Cmridge, MA, USA) included BDNF ntiody. Antiody purchsed from Snt Cruz Biotechnology (Snt Cruz, CA, USA) included nti-rit IgG horserdish peroxidse-conjugted ntiody. SYBR Green nd Invivofectmine 2.0 were purchsed from Life Technologies Corp (Crlsd, CA, USA). The Silencer Select GAPDH sirna ws purchsed from Amion (Austin, TX, USA). Both EZH2 ntisense nd sense phosphorothiote oligodeoxyrionucleotides were purchsed from Eurofins MWG Operon (Huntsville, AL, USA). All dditionl chemicls were purchsed from Sigm-Aldrich (St. Louis, MD, USA). Protein isoltion nd western lotting Nucler protein extrctions were performed s previously descried [6]. After the extrction, the protein concentrtion ws quntified y BCA protein ssy efore western lot, ccording to the mnufcturer's instruction (Thermo Fisher Scientific, Rockford, IL, USA). Then, the Lemmli loding uffer [345] ws mixed with smples t 1:1 volume rtio nd heted t 99 C for 10 min. The mixes were loded onto nd seprted y the 12% SDS-PAGE pre-cst gels (Lonz, Wlkersville, MD, USA). Smples were trnsferred onto the nitrocellulose memrnes t 12 Volts for 1 hour followed y 24 Volts for 1 hour. After trnsferring, nitrocellulose memrnes were locked in the 5% ovine serum lumin (BSA) overnight t 4 C with shking. On the second dy, the memrnes were trnsferred into 1% BSA for 3 times' 5 minutes wsh followed y overnight incution with primry ntiodies t 4 C with shking. The dilution of these primry ntiodies re: Ezh2 t 1:500, H3K27Me2 t 1:1000, H3K27Me3 t 1:1000, H3 t 1:1000, β-ctin t 1:1000 nd BDNF t 1:1000. On the next dy, memrnes were trnsferred into 1% BSA for 3 times' 5 minutes wsh followed y incution with nti-rit IgG horserdish peroxidse-conjugted ntiody t room temperture for 1 h with shking. The SuperSignl West Pico sustrte (Pierce, Rockford, IL, USA), Gel Doc XR System (Bio-Rd, Hercules, CA, USA) nd Quntity One (Bio-Rd, Hercules, CA, USA) imge nlysis softwre were used to detecte 64

75 the chemiluminescent signl. The finl reltive rtio vlue of ech protein ws equl to the rtio of the trget protein to the density vlue of β-ctin protein. Rel-time polymerse chin rection (qpcr) ssy design The totl RNA ws isolted nd followed y reverse trnscription s previously stted [6]. The 7500 Sequence Detection System (Life Technologies Corp., Crlsd, CA, USA) ws used to perform the rel-time PCR. Totl loding in ech well included 2 μl smple nd one volume of SYBR Green PCR Mster Mix (2 μl primers t 2 μm, 1 μl DEPC H 2 O nd 5 μl SYBR Green). Primers re listed in Tle 1. A melting curve test ws performed fter every qpcr experiment to confirm the specificity of the finl products. The Ct method ws used for ll qpcr results. Chromtin immunoprecipittion ssys (ChIP) The CHIP ssy procedure followed modified mnufcturer s protocol of the SimpleChIP Enzymtic Chromtin IP Kit (Cell Signling, Dnvers, MA, USA). First, smples were fixed y 1% formldehyde for 10 min nd trnsferred into 350 μl SDS lysis uffer (SDS, 1%; EDTA, 10 mm; nd Tris, ph 8.1, 50 mm). The Vircell Sonix (mximl power 750 wtts; Sonics & Mterils Inc, Newtown, CT, USA) were then used to sher fixed smples (32% power, 9 rounds of 10 sec soniction). Then the shered chromtin were liquoted into seven tues with 50 μl ech nd incuted with H3 ntiody (4 μl), H3K27me2 ntiody (8 μl), H3K27me3 ntiody (8 μl), Ezh2 ntiody (5 μl) nd nti-igg ntiody (4 μl, used for ckground) overnight t 4 C with shking. On the next dy, Protein G Agrose Beds were used to retrieve ntiodies-chromtin complex followed y elution t 65 C overnight with shking. On the third dy, ech liquots were centrifuged t 6,000 rpm for 10min nd the superntnt were trnsferred into Proteinse K tue for 65 C overnight incution. On the fourth dy, chromtin were trnsferred into purifiction column tue s provided in the kit. All smples were then sujected to qpcr using Bdnf gene primer trgeting promoter, trnsltionl strt site nd 3'-UTR (Tle 1). The dt were first deducted y ckground signl nd then normlized to n input control tht consisted of qpcrs from 1% crosslinked chromtin efore immunoprecipittion nd finl results were reported y Ct methods. Sttisticl nlysis JMP 9.0 (SAS Institute Inc, Cry, NC, USA) ws used for ll dt nlysis. For experiment 1, unpired t-test were used to compre control nd tretment group t ech single time point. For experiment 2, control, nlysis of vrince (ANOVA) were used to compre the difference mong nive, D10-fsting nd D3/10-fsing, which used the generl liner modeling procedure (GLM, SAS Institute Inc.). Significnt differences were considered s P < 0.05 unless otherwise mentioned. 65

76 3.3. Experimentl design Experiment 1: test the fesiility of in vivo sirna injection to knock down gene Eighteen one-dy-old chicks were used to test the effectiveness of single injection of sirna trgeting GAPDH mrna. They were rndomly divided into three groups, which hd six chicks per group nd mrked ACSF injection, sirna negtive control nd sirna trgeting GAPDH. All chicks hd ccess to food nd wter d liitum. On experimentl dy, oth negtive nd positive control sirna trgeting GAPDH were diluted into nm in Invivofectmine 2.0 followed mnufcturer s protocol. Briefly, first, 6 μl DEPC wter ws dded into the sirna liquot following centrifuge t 12,000 rpm for 10 minutes. Then, 6 μl complextion uffer ws dded into the sirna liquot. After rief vortexing, 12 μl Invivofectmine 2.0 ws dded into the sirna liquot nd followed y nother rief vortexing. After incuting t 50 C for 30 minutes, sirna ws redy to e used when cooled down to 37 C. Then 6 μl of ACSF or sirnas were injected into the lterl ventricle of D3 chicks. On D4, ll chicks were scrificed instntly y decpittion. The forerin (FB) ws collected. Experiment 2: test the selectiveness nd effectiveness of EZH2 nti-sense oligo. EZH2 ntisense oligonucleotides preprtion The sequences of EZH2 ntisense nd sense phosphorothiote oligonucleotides were dopted from previous studies [7], which were 5 -ACATTCCCCGGACTCTCAA-3 nd 5 -TTGAGAGTCCGGGGAATG-3, respectively. On rrivl, they were diluted into rtificil Cererospinl fluid (ACSF) (124 mm NCl, 2.5 mm KCl, 2.0 mm MgSO 4, 1.25 mm KH2PO 4, 26 mm NHCO 3, 2.5 mm CCl 2, 10 mm Glucose nd 4 mm sucrose) solution t finl concentrtion of 1 μg/μl or 0.5 μg/μl nd kept t 4 C. On the dy of injection, they were slowly wrmed up to 37 C.. test the effectiveness of EZH2 nti-sense Eighteen one-dy-old chicks were used to test the effectiveness of single EZH2 nti-sense injection. They were rndomly divided into three groups, which hd six chicks per group nd mrked ACSF, EZH2 nti-sense (0.5 μg/μl) nd EZH2- nti-sense (1 μg/μl). All chicks hd ccess to food nd wter d liitum. On D2, EZH2 nti-sense were diluted to 0.5 μg/μl nd 1 μg/μl in ACSF nd kept t 4 C. On D3, ll injection solutions were slowly wrmed up to 37 C. Then 5.5 μl of ACSF nd EZH2 nti-sense (t oth 0.5 μg/μl nd 1 μg/μl) were injected into the lterl ventricle. On D4, ll chicks were scrificed instntly y decpittion. The hypothlmus nd forerin (FB) were collected. c. test the selectiveness of EZH2 nti-sense Eighteen one-dy-old chicks were used to test the effectiveness of single EZH2 nti-sense injection. They were rndomly divided into three groups, which hd six chicks per group nd mrked ACSF, EZH2 sense (1 μg/μl) nd EZH2-nti- 66

77 sense (1 μg/μl). All chicks hve ccess to food nd wter d liitum. Then 5.5 μl of ACSF, EZH2 sense nd EZH2 nti-sense were injected into the lterl ventricle. On D4, ll chicks were scrificed instntly y decpittion. The hypothlmus nd FB were collected. Experiment 3: single fsting t three dy-of-ge (D3) fter EZH2 nti-sense injection Sixty-six one-dy-old chicks were rndomly picked nd evenly divided into three groups (eighteen for nive, twenty for control nd twenty for the tretment group). All of them were supplied with food nd wter ville d liitum. The nive group received 5.5 ul ACSF on D3 into the lterl ventricle. They were never restrined from food or wter. On D4 nd D11, nine chicks were rndomly picked from nive group nd scrificed instntly y decpittion. In the men time, on D3, 5.5 ul of ACSF nd EZH2 nti-sense (1 ug/ul) were injected into the lterl ventricle of the control nd tretment groups, respectively. Right fter injection, food ws withdrwn from oth for 24 hr. Twelve chicks from ech group were rndomly picked nd scrificed instntly y decpittion on D4. For the other twelve in ech group, food ws supplied d liitum. On D11, ll chicks left in ech group were scrificed y decpittion. PVN nd forerin (FB) were collected from ll chicks. Experiment 4: doule fsting on oth D3 nd ten dys-of-ge (D10) fter EZH2 nti-sense injection Thirty-six chicks were rndomly picked nd divided into two groups (eighteen per group). All of them were supplied with food nd wter ville d liitum. On D3, one group received ACSF 5.5ul injection into the lterl ventricle while the other group received EZH2 nti-sense (1 ug/ul) insted. After injection, in ACSF injected group, eighteen chicks were rndomly divided into control, D10-fsted nd D3-nd-D10 repetitive fsted (D3/10-fsted) group. For the control group, chicks were never fsted. For D10-fsted group, chicks were fsted only on D10 for 24 hr. For D3/10-fsted group, food ws withdrwn from them for 24 hr on D3 initilly nd ck to norml supply on D4. On D10, previous D3-fsted chicks were fsted gin for 24hr. All three groups of chicks were scrificed on D11 nd PVN nd FB smples were collected. For EZH2 nti-sense injected group, the tretments were exctly the sme s tht of the ACSF injected group. 67

78 3.4. Results 1. sirna in vivo injection hd no effect in protein regultion In the FB, compred to the ACSF injection group, negtive control sirna did not inhiit either GAPDH protein or mrna levels (P > 0.05) (Fig. 13). The positive control sirna lso did not inhiit GAPDH protein nd mrna, either (P > 0.05). 2. EZH2 ntisense oligonucleotides exhiited high effectiveness nd efficiency in EZH2 protein knockdown. EZH2 ntisense oligonucleotides effectively inhiited EZH2 protein nd mrna Becuse sirna did not work vi in vivo injection, we sustituted it with phosphorothiote oligonucleotides, which were more stle thn sirna under physiologicl conditions. After the EZH2 ntisense injection, in oth the FB nd PVN, EZH2 ntisense up to 1 μg/μl did not inhiit EZH2 mrna levels t 12 hr post injection (Fig. 14 A & C). However t 24 hr post-injection, oth ntisense concentrtions significntly inhiited EZH2 mrna in oth tissues (Fig. 14 B & D). To confirm the effectiveness of EZH2 ntisense, we lso quntified the reltive EZH2 protein mount in the FB nd PVN t 24 hr post injection. Although significnt mrna inhiition ws oserved with 0.5 μg/μl ntisense, EZH2 protein ws only inhiited y EZH2 ntisense t 1 μg/μl in oth the FB nd PVN. H3K27me2, which is the product of EZH2, lso showed n pprent inhiition y EZH2 ntisense t 1 μg/μl in the FB. However, in the hypothlmus, H3K27me2 ecme significntly elevted compred to the ACSF injected group. Therefore, EZH2 ntisense t 1ug/ul, ut not 0.5 μg/μl, effectively inhiited EZH2 protein t 24 hr post injection in oth the FB nd PVN.. EZH2 ntisense oligonucleotides selectively inhiit EZH2 protein Following EZH2 sense injection t 1 ug/ul, EZH2 protein ws not different from tht of the ACSF injected group (Fig 16). On the other hnd, EZH2 ntisense t 1 μg/μl inhiited EZH2 protein compred to theacsf injection. Thus, we confirmed tht EZH2 ntisense oligonucleotides otined high efficiency in EZH2 protein inhiition. 3. EZH2 ntisense impired the response to erly fsting in the PVN In the previous study, D3 fsting ws shown to significntly inhiit BDNF protein nd mrna expression in oth the FB nd PVN. This chnge ws due to the regultion of methyltion t H3K27 vi the PRC2. As the only suunit contining methyltrnsferse cpcity in the PRC2, EZH2 plyed n importnt role during the entire regultion process. Thus, interruption of EZH2 protein should significntly chnge the response of the ody to fsting stress. As we hd evluted the effectiveness nd efficiency of EZH2 ntisense oligonucleotides fter in vivo injection, we were interested to monitor the effect of chnges in 68

79 the PVN fter the sme fsting stress ut with disrupted EZH2 protein expression. First, we compred the D3-fsting to the nive group in ACSF injected group.. ACSF-injection overll hd no effect on glol methyltion t H3K27 expression in FB nd PVN First, we mesured the ody weight chnges of chicks red with different tretment (Tle 4). For the nive group, the ody weight ws mg on D4 nd mg on D11. After D3-fsting tretment, significnt decrese of ody weight ws oserved, which ws 20 mg nd 21 mg lower on D4 nd D11, respectively. Additionl EZH2 ntisense injection on D3-fsting did not chnge this phenomenon. Next, we mesured the chnges of epigenetic mrkers chnge fter D3-fsting. The previous study confirmed tht fter D3-fsting, the FB did not show significnt chnges in methyltion t H3K27 nd PRC2 complex key fctors' level. In the current "D3-fsting + ACSF injection" group, results were overll consistent with the previous results. This suggested tht ACSF injection overll hd only minimum effect in the rin (Fig. 17 E, F, G & H). On oth one nd eight dys post "D3-fsting + ACSF injection", no chnges of H3K27me2 were detected. On the other hnd, oth totl H3 nd H3K27me3 were elevted on D4 fter "D3-fsting + ACSF injection" ut returned to nive levels on D11. We lso tested the cetyltion level t the H3K27 site. H3K27c exhiited constnt inhiition fter "D3-fsting + ACSF injection" on oth D4 nd D11 in FB. To confirm the glol cetyltion nd methyltion level t H3K27, the mrna levels of key fctors in the PRC2 nd TRX complexes in FB. No significnt difference ws detect in SUZ, EZH2, EED nd CBP etween the nive group nd "D3- fsting + ACSF injection" group (Fig. 18 E, F, G & H). Agin, these dt confirmed tht FB ws not the mjor feeding regultion center nd did not respond to feeding stress. Next, we tested the glol methyltion sttus t H3K27 in the PVN from the sme nimls. In the previous finding, the PVN exhiited significnt chnges fter D3-fsting stress lone. In the current study, fter "D3-fsting + ACSF injection", we found tht the PVN reveled similr chnges in H3K27 methyltion ptterns (Fig. 17 A, B, C & D) s well s key fctors in PRC2 nd TRX complex (Fig. 18 A, B, C& D) compred to D3-fsting without ACSF injection. This indicted tht ACSF overll hd little effect in the PVN, s well. After "D3-fsting + ACSF injection", n instnt surge of totl H3, H3K27me2 nd H3K27me3 level ws oserved on D4 nd lsted until D11 (Fig. 17 A, B, C & D). This dt greed with previous dt from D3- fsting only chicks. On the other hnd, the cetyltion level t H3K27 site ws significntly inhiited fter "D3-fsting + ACSF injection" on oth D4 nd D11. Agin, we tested the mrna levels of key fctors in the PRC2 nd TRX to mtch the modifiction t H3K27 site (Fig. 18 A, B, C & D). After "D3-fsting + ACSF injection", EED, SUZ nd EZH2 hd mrna surge on D4. On D11, while EED remined significntly elevted, SUZ ecme indifferent while EZH2 ws inhiited compred to control. This incresed PRC2 69

80 components greed with the incresed methyltion ptterns t H3K27 from the protein nlysis. On the other hnd, CBP lso exhiited n incresed trend, which ws insignificnt, fter "D3-fsting + ACSF injection".. BDNF protein nd mrna exhiited inhiited levels fter "ACSF injection + D3-fsting" BDNF ws used s trget gene to mtch with the glol H3K27 methyltion chnges fter fsting stress. In the FB, fter "D3-fsting + ACSF injection", we found tht the BDNF protein ws constntly inhiited on oth D4 nd D11 (Fig. 19 B & D). These chnges mtched the dt in the FB from previous "D3-fsting only" tretment. Similrly, BDNF mrna showed n pprent decrese on D4, ut ws reversed on D11. Furthermore, we used CHIP ssy to detect the distriution of H3K27me2, H3K27me3 nd totl H3 long the Bdnf gene in the FB (Fig. 21 A-I). According to the BDNF mrna, it ws first inhiited on D4 nd then elevted on D11. Thus, there should e n increse nd drop of methyltion t H3K27 on D4 nd D11, respectively. Our results mtched this hypothesis. We found tht fter "D3-fsting + ACSF injection", there ws constnt surge of H3K27me2 t oth the Bdnf promoter re nd eginning of coding region on D4 ut not 3'-UTR (Fig. 21 B, E & H). On D11, H3K27me2 level returned ck to nive level. On the other hnd, H3K27me3 ws inhiited y "D3-fsting + ACSF injection" on D4 long Bdnf gene nd ecme more depressed on D11 (Fig. 21 C, F & I). Comine the chnges from oth H3K27me2 nd H3K27me3, it mtched the chnges of Bdnf mrna fter the tretment. We further monitored the totl H3 level long the Bdnf gene. A constnt increse in H3 ws found long Bdnf gene, which lsted until D11 (Fig. 21 A, D & G). Becuse the H3K27 methyltion ws minly crried y EZH2 in the PRC2, we lso tested the EZH2 distriution long the Bdnf gene. An pprent pek of EZH2 ws monitored long the Bdnf gene 24 hr fter "D3-fsting + ACSF injection" (Fig. 22 D, E & F). On D11, while the pek remined t Bdnf coding eginning region, the promoter nd 3'-UTR were no longer different from the nive. The dynmic chnges of EZH2 distriution overll mtched tht of H3K27me2 ut not H3K27me3. Next, the expression of the BDNF protein/mrna ws evluted in the PVN from the sme nimls. Other thn trditionl neurotrophic fctor, BDNF lso served s n norexigenic fctor in the PVN, which inhiited feeding. Thus, fsting should inhiit the expression of the BDNF protein. As expected, BDNF protein ws inhiited fter "D3-fsting + ACSF injection" tretment on D4 in the PVN (Fig. 19 A). This chnge ws comptile with the noticele inhiition of BDNF mrna level on D4 (Fig. 19 C). However, on D11, no difference ws detected on either BDNF protein or mrna etween the nive nd "D3-fsting + ACSF injection" groups (Fig. 19 A & C). To further study the mechnism under BDNF mrna chnges, we then monitored the distriution of methylted H3K27 long the Bdnf gene in the PVN. We found tht right fter "D3-fsting + ACSF injection", oth H3K27me2 nd H3K27me3 instntly peked t oth the Bdnf promoter nd coding eginning regions on D4 (Fig. 20 B, C, E & F). This peked level remined until D11. However, no pprent chnge ws detected t the Bdnf 3'-UTR region until D11 (except incresed 70

81 H3K27me3 t Bdnf 3'-UTR on D4) (Fig. 20 H & I). Totl H3 showed similr dynmic deposition long the Bdnf gene. It did not respond to the tretment on D4 ut ecme noticely incresed on D11 (except inhiited level on D4 t Bdnf coding eginning region) (Fig. 20 A, D & G). All these chnges mtched with the significnt drop of BDNF mrna on D4, ut not on D11. Furthermore, we tested the level of EZH2 to mtch up the methyltion sttus t H3K27. After "D3-fsting + ACSF injection", EZH2 exhiited pprent elevtion on D4 long Bdnf gene ut ws significntly inhiited on D11 (except EZh2 t 3'- UTR, which ecme sme s nive level) (Fig. 22 A, B & C). 4. EZH2 ntisense injection chnged the response towrds D3-fsting stress. EZH2 ntisense eliminted the surge of methyltion t H3K27 fter D3-fsting in PVN Becuse EZH2 is the only methyltrnsferse in the PRC2, knockdown the EZH2 protein should eliminte methyltion t H3K27, nd thus impir the ody's response towrds D3-fsting stress. Thus, we injected chicks with EZH2 ntisense nd monitored the chnges inside the PVN nd FB fter D3-fsting tretment nd compred them to "D3-fsting + ACSF injection". In FB, we found tht fter "EZH2 ntisense injection + D3-fsting", the EZh2 ntisense pprently eliminted the methyltion t the H3K27 site. Both H3K27me2 nd H3K27me3 exhiited significntly lowered levels compred to oth nive nd "D3-fsting + ACSF" groups (Fig. 17 F & G). Totl H3 responded the sme s tht of H3K27me2 nd H3K27me3 (Fig. E). At the sme time, EZh2 sved the noticely inhiited H3K27c cused y "D3-fsting + ACSF" on oth D4 nd D11. In ddition, H3K27c ecme even more elevted thn nive levels (Fig. 17 H). The significnt drop of oth di- nd tri-methyltion t H3K27, led to the hypothesis tht the mrna levels of PRC2 fctors would drop fter the EZH2 ntisense injection. PCR results greed with this hypothesis: EZH2 ntisense oviously inhiited EZH2, which vi unknown mechnism lso inhiited SUZ nd EED (Fig. 18 E, F & G). This result mtched the chnges of methyltion sttus t H3K27. In ddition, CBP mrna ws lso incresed fter EZH2 ntisense injection (Fig. 18 H). In the PVN, we previously found tht D3-fsting hd tremendous effect on the methyltion sttus t the H3K27 site, which resulted from chnges of the PRC2. After EZH2 ntisense injection, the PVN overll hd significnt loss of methyltion t the H3K27 site. On D4, compred to "ACSF injection + D3-fsting" tretment ", "EZH2 ntisense + D3-fsting" ws le to pprently inhiit the surge of H3K27me3 nd H3K27me2 on D4 (Fig. 17 B & C). On D11, while the inhiition of H3K27me2 remined, H3K27me3 ecme even more elevted in the "EZH2 ntisense + D3-fsting" group. Similrly, the H3 level ws constntly inhiited fter EZH2 ntisense injection, which lsted until D11 (Fig. 17 A). However, compred to the "ACSF injection + D3-fsting" tretment, EZH2 ntisense did not eliminte the significnt drop of cetyltion levels t H3K27 compred to the nive. On D4, H3K27c remined the sme etween the ACSF nd EZH2 injected groups fter D3-fsting (Fig. 17 D). On D11, this inhiition of H3K27c remined in the PVN compred to nive nd "ACSF injection + D3-fsting" chicks. H3K27c ecme further inhiited in the EZh2 injected group. 71

82 The mrna level of PRC2 nd TRX key fctors were lso tested in the PVN. CBP mrna ws not ffected y EZH2 ntisense s it showed no difference to tht of ACSF injection (Fig. 18 D). However, fctors in PRC2 chnged drmticlly. First, EZH2 ntisense drmticlly inhiited EZH2 mrna levels strting 24 hr post injection nd lsted 8 dys post injection (Fig. 18 B). For EED, "EZH2 ntisense injection + D3-fsting" effectively eliminted its surge led y "ACSF injection + D3- fsting" on oth D4 nd D11 (Fig. 18 C). For SUZ, EZH2 ntisense did not chnged its mrna level from tht of the ACSF injection on D4. However, SUZ mrna ws significntly lower on D11 fter "EZH2 ntisense + D3-fsting" when compred to tht of ACSF injection (Fig. 18 A).. BDNF protein nd mrna expression level confirmed the fct tht EZH2 ntisense eliminted ody's response towrds D3-fsting As the significnt glol inhiition of oth methyltion nd cetyltion hppened t H3K27 in the FB, we suspected tht BDNF protein nd mrna levels should e ffected fter "EZH2 ntisense injection + D3-fsting". We found tht on D4, EZh2 ntisense not only eliminted BDNF protein inhiition led y "ACSF injection + D3-fsting" in the FB, ut lso rought the BDNF protein to much higher level compred to the nive group (Fig. 19 B). BDNF mrna on D4 ws very similr to its protein chnges. On D11, while "ACSF injection + D3-fsting" significntly incresed BDNF mrna compred to the nive group, "EZH2 ntisense injection + D3-fsting" rought BDNF mrna ck to nive group (Fig. 19 D). We then used CHIP ssys to test the distriution of methylted H3K27 long Bdnf gene in FB. Compred to "ACSF injection + D3-fsting", EZH2 ntisense eliminted the surge of H3K27me2 led y D3-fsting long the Bdnf gene on D4 (Fig. 21 B, E & H). However, more significnt pek of H3K27me2 ws monitored on D11 in "EZH2 ntisense injection + D3- fsting". On the other hnd, EZH2 ntisense restored H3K27me3 to nive group level compred to the inhiition cused y "ACSF injection + D3-fsting" long the Bdnf gene on oth D4 nd D11 (except t 3'-UTR on D4) (Fig. 21 D, F & I). These chnges overll mtched the chnges of BDNF mrna. On the other hnd, totl H3 either styed t the sme level s the nive or ws more elevted fter "EZH2 ntisense injection + D3-fsting" compred to the "ACSF injection + D3-fsting" on D4 long Bdnf gene (Fig. 21 A, D & G). This pttern of H3 lsted until D11. Next, we tested the BDNF response towrds D3-fsting fter EZH2 ntisense injection in PVN. We hypothesized tht BDNF protein nd mrna should e spred from D3-fsting stress fter EZH2 fsting. We found tht EZH2 ntisense spred the BDNF inhiition from "ACSF + D3-fsting" nd rought oth BDNF protein nd mrna ck to nive levels (Fig. 19 A & C). On D11, while BDNF protein ws elevted fter "EZH2 ntisense injection + D3-fsting" when compred to the "ACSF injection + D3-fsting", no chnges ws found for BDNF mrna. We further tested the dynmic deposition of methylted H3K27 on the Bdnf gene in the PVN. As expected, the "rescue" pttern ws detected t multiple loci of Bdnf. On D4 nd D11, while "ACSF injection + D3-fsting" incresed the 72

83 deposition of oth H2K27me2 nd H3K27me3 t the Bdnf promoter nd coding eginning region, EZH2 ntisense eliminted this surge nd rought them closer to the nive group (Fig. 20B, C, E & F). However, on D11, not only the pek of H3K27me2 nd H3K27me3 disppered, they ecme further inhiited thn the nive group. At the sme time, the chnges t Bdnf 3'-UTR were miniml. These chnges overll mtched the chnges of BDNF mrna levels. In ddition, on D4, EZH2 ntisense elevted the H3 deposition long Bdnf gene on D4 injection nd significntly inhiited on D11 (Fig 20 A, D & G). Furthermore, EZh2 ws lso monitored to mtch the chnges of H3K27me2 nd H3K27me3 chnges. We found tht on D4, t the Bdnf promoter nd coding eginning regions, the EZH2 pek resulting from the D3-fsting ws eliminted nd ws even more inhiited thn the nive group (Fig. 22 A, B & C). Additionlly, the EZH2 ntisense inhiited EZH2 deposition on D11 long the Bdnf gene s well. 5. EZH2 eliminted the "moleculr memory" towrds repetitive fsting stress in PVN. EZH2 ntisense impired "moleculr memory" PVN We gin mesured the chnges of ody weight following either D10- or D3/10-fsting tretment (Tle 5). Compred to the nive group, D10-fsting significntly decresed chicks ody weight. An dditionl D10-fsting on the top of D3-fsting further decresed ody weight. On the other hnd, EZH2 ntisense injection overll hd no effect on ody weight of chicks in nive nd D10-fsting group. However, Compred to "D3/10-fsting + ACSF injection" group, the ody weight of the "D3/10- fsting + EZH2 ntisense" group ws less ttenuted. As we previously showed, fsting t D3 could protect the ody y minimizing the response to the sme stress in lter dys. We clled it "moleculr memory". This protection mechnism is minly vi regulting the methyltion sttus t H3K27. As n importnt methyltrnsferse, EZH2 plys key role in "moleculr memory". Thus, interruption of the PRC2 should impir the H3K27 methyltion, which cuses dysfunction of this memory towrds repetitive stress. To test this hypothesis, we injected EZH2 ntisense into chicks nd put them under repetitive fsting stress conditions. First, we tested the FB's response towrd repetitive stress. As we showed erlier, the FB hd little function in feeding regultion nd did not exhiit ny moleculr memory towrds the fsting stress. In the current study, we confirmed this hypothesis gin y showing tht the FB did not exhiit ny memory towrds fsting stress. After single D10-fsting, there ws overll no chnge in H3K27 methyltion or cetyltion nd totl H3 levels compred to the nive group (Fig 23 E, F, G & H). Repetitive D3/10-fsting only significntly inhiited H3K27me3 levels. However, fter EZH2 ntisense injection, EZh2 ntisense non-specificlly inhiited H3K27me2 nd incresed H3K27c in either type of fsting stress. Totl H3 nd H3K27me3 did not exhiit ny chnges in either fsting treted groups. We mesured the mrna levels of the PRC2 to confirm the chnges of methyltion t H3K27 in the FB. In the ACSF injected group, fter single D10-fsting stress, noticele increse nd decrese of SUZ nd EED were monitored, 73

84 respectively (Fig. 24 E & G). However, fter D3/10-fsting, while SUZ decresed nd ecme indifferent from the nive, EED remined noticely inhiited. Both EZH2 nd CBP mrna levels, lthough elevted, were not significnt compred to the nive (Fig. 24 F & H). After D3/10-fsting tretment, EZH2 ecme significntly elevted. CBP, on the other hnd, remin silenced to the repetitive fsting. In the EZH2 ntisense injected group, EZH2 mrna ws inhiited fter either type of fsting stress (Fig. 24 F). After D3/10-fsting, SUZ nd CBP were inhiited while EED remined unchnged (Fig. 24 E, G & H). Next, we pplied the sme test towrds the PVN smple. In the ACSF injected group, the PVN reveled "moleculr memory" pttern: while single fsting on D10 significntly elevted/inhiited protein level from the nive group, D3 fsting prior to D10 fsting llevited this elevted/inhiited level. In Western Blot testing, we found tht H3K27me2, H3K27me3 nd H3K27c followed this pttern (Fig. 23 B, C &D). For H3, it exhiited constntly incresed levels fter either D10-fsting or D3/10-fsting (Fig. 23 A). However, fter EZH2 ntisense injection, "moleculr memory" ptterns disppered. First, EZH2 ntisense injection constntly eliminted the pek of H3K27me2 led y single fsting on D10 or repetitive fsting on D3/10 (Fig. 23 B). For H3K27me3 nd H3K27c, EZH2 ntisense injection cused constnt elevtion compred to "ACSF injection + D10- fsting" tretment (Fig. 23 C & D). In ddition, EZH2 ntisense cused constnt inhiition of H3 in oth D10 nd D3/10-fsting group (Fig. 23 A). The mrna levels of the PRC2 key fctors greed with the methyltion sttus in the PVN. Compred to the ACSF injection group, EZH2 ntisense constntly inhiited not only EZH2, ut lso SUZ nd EED mrna levels fter single fsting on D10 (Fig. 24 A, B & C). After D3/10-fsting tretment, EZH2 ntisense remined effective nd significntly inhiited SUZ nd EZH2 while elevted the level of EED. CBP from the TRX overll ws not ffected y EZH2 ntisense compred to tht of ACSF injection (Fig. 24 D).. BDNF expression supports the sttement tht EZH2 ntisense impired "moleculr memory" We then studied BDNF protein nd mrna chnges in the FB. After D10-fsting, significnt inhiition of BDNF protein ws found (Fig. 25 B). After D3/10-fsting fsting, BNDF reveled prtilly recovered protein level from D10-fsting, which however remined lower thn the nive. BDNF mrna levels in the FB shown no chnges fter D10-fsting ut much inhiited levels fter repetitive fsting in the ACSF injected group (Fig. 25 D). We further used CHIP ssys to study these chnges in BDNF mrna levels. Without EZH2 ntisense, D3-fsting ws le to elicit surge of totl H3 nd H3K27me2 levels long the Bdnf gene (Fig. 27 A, B, D, E, G & H). After D3/10-fsting, the surge of H3 ecme even igger while H3K27me2 remined the sme level s tht of D3-fsting. On the other hnd, H3K27me3 ws inhiited y D3-fsting nd exhiited more significnt inhiition fter D3/10-fsting (Fig. 27 C, F & I). However, fter EZH2 ntisense injection, H3 incresed its deposition fter D10-fsting long Bdnf ut decresed its deposition fter D3/10- fsting (Fig. 27 A, D & G). At the men time, EZH2 ntisense ws le to decresed H3K27me2 deposition long Bdnf in oth 74

85 single nd repetitive fsting tretments (except t the Bdnf promoter nd 3'-UTR regions) (Fig. 27 B,E & H). On the contrry, H3K27me3 deposition were constntly incresed long the Bdnf gene fter EZH2 ntisense injection, in oth single nd repetitive fsting tretments (Fig. 27 C, F & I). EZH2 itself showed unchnged, decresed nd incresed levels t Bdnf promoter, coding eginning region nd 3'-UTR regions, respectively, fter D3-fsting (Fig 28 D, E & F). EZH2 ntisense lso inhiited EZH2 level long Bdnf gene fter D3/10-fsting tretment (Fig 28 D, E & F). Next, we tested the chnges of BDNF in the PVN. Compred to the FB, PVN exhiited much more meningful chnges fter fsting stress. In the ACSF injected group, oth BDNF protein nd mrna exhiited the "moleculr memory" ptterns fter repetitive fsting stress (Fig. 25 A & C). EZH2 ntisense impired this pttern of oth BDNF protein nd mrna. After single fsting on D10, BDNF protein nd mrna exhiited significnt elevted levels fter EZH2 ntisense injection (Fig. 25 A & C). After repetitive fsting, oth BDNF protein nd mrna expression levels showed the reversl. The results from CHIP ssys greed with tht of BDNF mrna chnges in PVN. First, fter ACSF injection t Bdnf promoter nd coding eginning regions, oth H3K27me2 nd H3K27me3 exhiited the "moleculr memory" pttern: while single fsting D10 significntly shifted its level from the nive, repetitive fsting on oth D3 nd D10 rought its level ck or close to the nive group (Fig. B, C, E & F). Following EZH2 ntisense injection, D10-fsting significntly inhiited H3K27me2 nd H3K27me3 level t oth Bdnf promoter nd coding eginning regions (Fig. B, C, E & F). However, this pttern ws reversed fter D3/10-fsting (except H3K27me3 ws not different etween EZH2 ntisense nd ACSF group). 3'-UTR region remined unchnged regrdless of injection nd fsting(fig. 26 H & I). On the other hnd, totl H3 levels tended to increse upon D3-fsting nd ecme more incresed upon D3/10-fsting without EZH2 ntisense (Fig. 26 A, D & G). However, EZH2 ntisense significntly incresed totl H3 fter single D10 fsting t the Bdnf promoter nd coding eginning region. This pttern reversed fter D3/10-fsting tretment. EZH2 deposition fter ACSF injection exhiited "moleculr memory" only t Bdnf promoter region ut not others (Fig28 A, B & C). After EZH2 ntisense injection, EZH2 deposition ws inhiited t Bdnf promoter region fter single fsting on D3. In ddition, EZH2 ntisense lso elevted EZH2 level long Bdnf gene (Fig28 A, B & C). 75

86 3.5. Discussion To the uthor's knowledge, only limited numer of pulictions reported the chnges of histone [6] nd histone modifiction enzymes [349] in hypothlmus fter unfvorle nutritionl stress. Thus, we re interested in knowing the histone covlent modifiction regultion fter nutritionl stress. In previous study, we showed tht fter 24 hr fsting on D3, significnt incresed methyltion ws detected t H3K27 in PVN. In ddition, PVN is le to estlish so clled "moleculr memory", which protects the ody from future sme stress. In the present study, we first confirmed tht ACSF injection in lterl ventricle overll did not ffect the ody's response towrd D3-fsting. Additionlly, we lso confirmed the existence of "moleculr memory" inside the PVN fter ACSF injection into the lterl ventricle. However, fter EZH2 ntisense ppliction, the PVN lost its correct responses fter single24 hr fsting on D3. In ddition, the PVN filed to develop "moleculr memory" towrds the fsting stress fter EZH2 ntisense injection. Feeding regultion is determined y the neurl-networking in the five different su-hypothlmic nuclei [51]. Among them, PVN is one of most importnt nuclei nd plys pivotl role in feeding nd energy homeostsis regultion. A histologicl study showed tht the PVN received innervtion fiers from lmost ll res of the hypothlmus [105]. Nerly ll known orexigenic nd norexigenic fctors incresed nd decresed feeding respectively, when micro-injected into the PVN [50]. Multiple numerous neuro-peptides, including BDNF [79], ws lso detected in the PVN. In ddition, some recent studies find tht fsting stress tends to chnge histone methyltion sttus inside PVN [6]. In the current study, we showed tht fter single 24 hr fsting on D3, there ws significnt incresed methyltion nd decresed cetyltion t H3K27 in PVN. This chnged pttern of H3K27 covlent modifiction lsted until 8 dys post fsting. These dt suggest tht histone modifiction is highly responsive rther thn sttic towrds nutritionl stresses. As previously reported y other groups tht nutritionl stress could leve covlent modifiction on oth histones [6,349] nd nucleotides [350,351]. Our dt provided evidence to support this hypothesis. In ddition, 8 dys post injection, elevtion of methyltion nd cetyltion t H3K27 remined exist. These results mtched our previous finding tht 24 hr fsting on D3 could keep inducing H3K27 methyltion nd methyltion mchinery, i.e. PRC2, up to t lest 38 dys. This suggests tht single nutritionl stress during erly rin development could induce long term covlent modifiction on histones. This durle histone modifiction cn led to chnging of oth chromtin structure nd gene trnscription nd, thus, incresed risk towrds some diseses such s oesity. Some recent epidemiologicl studies climed tht nutritionl stress in erly life period is le to cuse long term disturnce in energy nd metolic homeostsis in humns [376]. Thus, our dt supported this hypothesis y showing tht long term histone modifiction deposition on chromosome could e the potentil mechnism. We were lso interested in investigting the mening of this long term modifiction on histone. Previous studies showed tht nimls ecme more resistnt towrds het chllenge lter in life if exposed to het stress during thermlcontrol estlishment period [377,378]. It is minly due to dption of histones modifiction in the preoptic nd 76

87 nterior hypothlmus (PO/AH) [7]. We suspected the histone modifiction in PVN fter fsting hd the sme the function s tht of PO/AH fter het stress. Thus, we compred repetitive 24 hr fsting stress (on oth D3 nd D10) to the single 24 hr fsting (on D3) in chicks. Similrly, we found tht the ody exhiits less disrupted methyltion s well s cetyltion t H3K27 fter repetitive fsting stress compred to single fsting stress. We clled this chnge of pttern s the "moleculr memory". We used BDNF protein, BDNF mrna nd H3K27me2/3 lying on Bdnf gene to support this finding s well. This indictes tht the long term histone covlent modifiction serves s protective mechnism for the ody towrd future sme stress type. Importntly, this is the first study series reporting tht repetitive fsting during feeding-control estlishment develops "moleculr memory" nd protects the ody from future sme fsting stress. In ddition, this "moleculr memory" utilizes histone covlent modifiction vi PRC2. However, it remins elusive how neurons in the PVN could develop "moleculr memory" in the first plce. Additionlly, it is not cler whether the ody could uild "cross moleculr-memory" towrds different types of stress. It is very surprising to see tht not ll fctors inside PRC2 otined "moleculr memory": in the PVN while repetitive fsting induce the memory for EED nd SUZ, EZH2 kept incresing in oth single nd repetitive fsting sitution. PRC2 is huge H3K27 methyltion mchinery [213]. In its functionl core, three key fctors hve een discovered, which re EZH2, SUZ nd extr-sex com EED [213]. As EZH2 is the only methyltrnsferse inside PRC2, its expression level incresing suggested the sme trend for H3K27 methyltion [242]. However, H3K27me2 /me3 exhiited the similr "moleculr memory" pttern s well fter repetitive fsting stress. Literture review suggested tht EZH2 hd rely no function if presented lone without other non-ctlytic suunits in PRC2 [248]. This is ecuse EZH2 does not hve DNA inding domin, which prevent EZH2 from histone inding [259,355]. So fr EZH2 hd een shown to hve direct connection with EED in [251,259,357] nd SUZ[252,259,358], which forms triplet model tht EZH2 sits in the middle[259]. EED inds to histones [255,356] while SUZ connects proteins on chromosome, such s heterochromtin protein 1lph [252]. These linkges insure the EZH2 to hve full contct with histones nd proceed methyltion. Muttion or knockout SUZ [259,357,358] nd EED [259,277,359] depleted the function of PRC2. These dt suggest tht oth ctlytic (EZH2) nd non-ctlytic suunit (EED nd SUZ) re pivotl to PRC2. In the current study, we reported tht "moleculr memory" exhiited in EED nd SUZ, ut not EZH2. This indictes tht lthough ody responds to the fsting stress through EZH2-methylting-H3K27, it is indeed vi regulting of EED nd SUZ. Thus, vi regulting EED nd SUZ only could determine level of H3K27 methyltion. However, we re not cler how the ody instills memory towrd SUZ nd EED nd why EZH2 does not. BDNF is initilly known for its function in neuron prolifertion, differentition, growth nd interction [125]. Recently, it is lso ctegorized s n norexigenic fctor [75,159]. Both cute injection [159,160] (including t the PVN [161]) or long term intrcereroventricμlr injection [162] of BDNF inhiited the feeding nd incresed energy expenditure in tested 77

88 nimls. While fsting inhiited BDNF level inside in the ventromedil hypothlmus [75] nd dorsl vgl complex of hinder rin [159], re-fed led to recovered BDNF [159]. In femle humns with either ulimi nervos (BN) or norexi nervos (AN), significntly lowered serum BDNF level ws oserved [163]. All these dt highlighted the importnce of BDNF in feeding regultion. In our current study, we identified drop in oth the BDNF protein nd mrna levels nd long term BDNF mrna inhiition fter single 24 hr fsting on D3. This finding greed with the findings from previous studies. But more importntly, we were le to mtch the BDNF mrna chnges with the H3K27me2/3 distriution long the Bdnf gene. It is widely ccepted tht H3K27 methyltion compcts nucleosomes nd thus locks trnscriptionl elongtion on gene templtes [361,362]. Thus, incresed H3K27me2/3 fter fsting indicted mssive gene trnscription inhiition in the PVN. However, it remined out of clue tht when nd where of H3K27me2/3 deposition will led to induce specific gene inhiition. We picked three res on Bdnf gene, which re promoter (-869 to -801 p), coding strting region (+91 to +190 p ) nd 3'-UTR side (+1623 to p). Sme s the glol chnges in PVN, instnt incresed level of H3K27 me2/me3 ws deposited minly hppened t Bdnf promoter nd coding eginning region (ut sometimes Bdnf 3'-UTR s well). EZH2, on the other hnd, ws lso significntly chnges ut long Bdnf gene. This indictes tht fsting inhiits Bdnf gene trnscription minly vi regulting depositing PRC2 long Bdnf gene. The PRC2 is known for its cpcity to di- nd tri methylte H3K27 site [242,243,247,248]. Becuse chicks' PVN showed incresed H3K27 methyltion levels fter fsting stress, we were interested to investigte wht hppened if PRC2 ws impired. We picked EZH2 s our trget ecuse it is the only methyltrnsferses inside PRC2 [242]. An erly study reported tht n EZH2 null muttion is lethl towrds nimls [379]. Thus, EZH2 could only e knockdown vi sirna nd oligodeoxyrionucleotide. First, we tried to test the effectiveness of sirna knockdown on the trget mrna. We picked sirna trgeting GAPDH s our test nd performed in vivo intrcereroventriculr injection. However, even t nm (finl concentrtion in chick CSF, suggested concentrtion > 5 nm, chick t D3 hd 4u μl CSF) nd with the help of Invivofectmine 2.0 Regent, which ws more thn the 5400 fold higher thn suggested concentrtion, no effect on GAPDH protein nd mrna levels in the chick FB ws oserved 24 hr fter 6 μl sirna exhiited. Literture review did find some successful studies to induce gene knockdown in the CNS vi in vivo sirna intrcereroventriculr injection. But sirna ws either injected topiclly [380,381,382] or newly remodeled which is more resistnt to degrdtion [383]. In our cse, we suspected the filure ws due to the comintion of following fctors: 1) premture ending of sirna rection in FB (suggested time y mnufcture is 48 hr) [384,385], 2) nked signl quick degrdtion in CSF [386] nd 3) difficult penetrtion into deeper rin tissue [385] even with the presence of Invivofectmine 2.0 Regent. Next we dopted the method from n erlier study, using oligodeoxyrionucleotides trgeting the EZH2 to knockdown the EZH2 protein [7]. Compred to norml oligonucleotides, the oligodeoxyrionucleotide hd nonridging oxygen replced y sulfur tom t every single phosphorus, which mkes them highly resistnt towrds endonucleses. In ddition, the doule- 78

89 phosphorothioltion t oth the 3'-end nd 5'-end of the deoxyrionucleotide further protects it from exonucleses. Thus, EZH2 oligodeoxyrionucleotides ntisense should e very stle fter in vivo intrcereroventriculr injection. As expected, EZH2 oligodeoxyrionucleotide ntisense t 1 μg/μl effectively nd selectively inhiited EZH2 protein nd mrna expression fter 24 hr post injection. So fr, two mechnisms hve introduced to explin how oligodeoxyrionucleotide ntisense work inside cell. These mechnisms re the RNse H pthwy [369,370,371], locking trnscription inititor progression [372] nd inhiiting mrna splicing [368]. The RNse H pthwy utilizes RNse H enzyme to hydrolyze the RNA/DNA complex, which significntly inhiits mrna of trget gene [368]. However, other two pthwys re expected to inhiit mrna trnsltion rther thn hydrolyze nd decrese mrna [372]. Thus, in our cse, EZH2 oligodeoxyrionucleotides ntisense should utilize RNse H pthwy s we oserved significnt inhiition of EZH2 mrna. We evluted the effect of EZH2-ntisense inhiition y monitoring the methyltion sttus t H3K27 in the PVN. Wheres the D3-fsting significntly incresed the EZH2 mrna on D4, EZH2 ntisense eliminted this response. In ddition to the inhiition of EZH2 mrna, the pek of H3K27me2/3 protein ws lso eliminted. This indictes tht EZH2 knockdown depletes the function of PRC2 nd thus impirs the chnges of H3K27 methyltion. In other words, the ody's response towrds D3-fsting ws impired. On D11, EZH2 mrna remined inhiited in "EZH2 ntisense + D3-fsting" group compred to "ACSF + D3-fsting" group. This oservtion indicted tht EZH2 ntisense remined effective even 8 dys fter injection. The EZH2 ntisense we used is phosphorothioted oligonucleotides, which re firly resistnt to the intrcellulr endonucleses nd exonucleses [368]. This explined constnt inhiition of EZH2 mrna in PVN neurons, which led to constnt inhiition of methyltion t H3K27. Interestingly, SUZ nd EED mrna lso shown chnged level fter EZH2 ntisense injection. As we hve lsted the EZH2 ntisense sequence, it could not ind to SUZ nd EED mrna. Literture review found tht phosphorothioted oligonucleotides re cple of cusing non-specific mrna inhiition [368]. However, individul cell hve intct iologicl environment where interruption of one molecule could ffect the level of the others, especilly they hve collortion. Thus, this phenomenon could either e the response of PVN to EZH2 ntisense or non-specific inhiition cused y EZH2 ntisense. Monitoring the chnges of BDNF protein mrna nd deposition on Bdnf gene greed with the effect of EZH2 ntisense t glol level. Furthermore, EZH2 ntisense lso eliminted D3/10-fsting induced long-term H3K27 methyltion modifiction. We showed tht this long-term modifiction ws eneficil for the ody nd protect ody from future fsting stress. As H3K27 specific methyltrnsferse, EZH2 plyed pivotl role during this process, lthough EZH2 itself does not otin "moleculr memory". To provide more evidence, we gin pplied EZH2 ntisense towrds the repetitive fsting test. We found tht fter the injection, EZH2 mrna ws constntly inhiited compred to non-ezh2 injected group. In ddition, the "moleculr memory" pttern ws totlly eliminted for EED, ut not SUZ. As key fctor for PRC2 functioning, lost of "moleculr memory" for EED ws expected to eliminte "moleculr memory" pttern for products of PRC2. Accordingly, H3K27me2/me3 79

90 no longer exhiited ny "moleculr memory" s well. These dt indicte tht interrupting EZH2 expression impired neurons in the PVN to otin memory towrd fsting, which eliminted protection in the PVN. This hypothesis ws lter supported y the results of the BDNF protein, mrna nd H3K27me2/3 distriution on the Bdnf gene s well. We lso monitored some discrepncies in our dt. For exmple, in "ACSF injected + D3-fsting" group, lthough EZH2 deposition mtched perfectly of H3K27me2/me3 level in PVN on D4, it did not mtched perfectly well of H3K27me2/me3 on D11. We do not understnd why this hppened. It could e due to the co-effect of other PcG complex [245], chromosome proteins, nucleotide covlent modifiction or histone covlent modifiction t 3'-UTR tht locks the function of PRC2 t this re. In ddition, H3K27me3 nd H3K27c were found to hve the exctly sme ptterns in repetitive fsting study, ut no fter single D3-fsting. H3K27me3 nd H3K27c re the products of PRC2 [245] nd TRX [387], respectively. They ntgonize ech other to regulte gene trnscription level under different circumstnce [295,363]. Most of the time, H3K27me3 nd H3K27c re reversely prllel to ech other [226,388]. In our sitution, ecuse ll chicks hd een fsting efore, this prllel H3K27me3 nd H3K27c effect could e fsting specific. It is lso possile tht the significnt incresed H3K27me3 nd H3K27c could e deposited into different re of the chromosome, ffecting different genes. Overll, our dt showed tht the PVN is cple of responding to fsting stress quickly nd methyltes H3K27 vi regulting of PRC2. These chnges lst for firly long time ecuse it protected the ody from interrupted inner environment fter future sme fsting stress. Upon impirment of the PRC2 vi knockdown of EZH2, the PVN is no longer cple of responding to fsting correctly, thus it cnnot estlish ny memory to protect homeostsis from future fsting. 80

91 D4 D11 D3-fsting ntisense Body weight SEM Tle 5 Chnges in ody weight of chicks following fsting. At three dys of ge, eighteen irds were injected with ACSF into lterl ventricle of rin nd fed d liitum (nïve). Eighteen irds were injected with ACSF nd nother eighteen irds were injected with EZH2 ntisense oligonucleotides. Right fter injection, these irds were fsted for 24 hours (D3-fsted). Dt from the sme scrifice dy red different letters indicte significnt different. Nïve controls nd fsted chicks (from oth ACSF nd ntisense injected group) were scrificed on dy 4 (D4) nd dy 11 (D11). Before scrificing, ody weight ws mesured. There ws significnt decrese in the ody weight etween nïve nd "D3-fsting + ACSF" chicks on 24 hrs nd 8 dys post fst. This pttern of ody weight remined sme on D11. 81

92 D11 D3-fsting D10-fsting ntisense Body weight * SEM Tle 6 Chnges in ody weight of chicks following D10-fsting or D3/10-fsting. At three dys of ge, eighteen irds were injected with ACSF into the lterl ventricle nd then rndomly nd evenly divided into three groups. One group ws rndomly picked nd fed d liitum until D11 (nïve). For D3- nd D10- doule fsting group (clled D3/10-fsting in the following prgrph), food ws withdrwn for 24 hr on D3. On D4, food ws ck to norml supply to the fsted group. On D10, D3/10- fsting groups were fsted gin for 24hr. For single fsting on D10 group (clled D10-fsting), food ws only withdrwn on D10 for 24 hr. All three groups of chicks were scrificed on D11 nd PVN nd FB smples were collected. In EZH2 ntisense injected group, ll tretments remined the sme s tht of the ACSF injected group, except ACSF ws replced with EZH2 ntisense. Dt red sterisk mrk indictes significnt difference etween ACSF nd EZH2 ntisense injected group. Dt red Greek letter indictes significnt difference mong group which receives the sme injection chemicls.. In ACSF injected group, D10-fsing significntly inhiited chick ody weight. If pplied fsting on D3 prior to D10 fsting, chick ody weight ws significntly decresed compred to oth the nive group nd 10-fsting group. EZH2 ntisense injected group showed similr pttern compred to tht of ACSF group: there ws n overll down regultion of ody weight with incresed fsting time. In etween EZH2 ntisense injected group, while EZH2 ntisense hd no effect on the nive nd D10-fsting group, it significntly incresed ody weight of D3/10-fsting group compred to ACSF injected group. 82

93 R e l tiv e p r o te in q u n tity o f G A P D H / - c tin r e l tiv e m R N A e x p r e ssio n G A P D H / - c tin A A C S F n e g tiv e c o n tro l p o s itiv e c o n tro l B G A P D H - c tin * A C S F in je c tio n n e g tiv e c o n tro l s ir N A p o s itiv e c o n tro l s ir N A 0.0 A C S F in je c tio n n e g tiv e c o n tro l s ir N A s ir N A t rg e tin g G A P D H 83

94 Figure 11. Chnges in protein nd mrna levels of GAPDH in the chick forerin (FB) (A & B) following nti-gapdh sirna injection. At three dys of ge, irds were injected with ACSF, GAPDH sense nd GAPDH ntisense sirna into 3 rd ventricle of rin. 24 hr fter injection, ll chicks were scrificed nd FB ws extrcted. Totl protein ws isolted nd evluted y Western lot using the ntiody ginst GAPDH. Blot density were compred with -ctin. Ech vlue is the men stndrd error of the men of 6 individul chicks. GAPDH protein shown no difference mong GAPDH sense nd GAPDH ntisense sirna injection group (A). Totl mrna ws isolted nd evluted y rel-time polymerse chin rection (PCR) using the Syr green method with Gpdh gene-specific primers (see Tle 1). Gene expression levels were compred with -ctin. Ech vlue is the men stndrd error of the men of 6 individul chicks. GAPDH mrna shown no difference mong GAPDH sense nd GAPDH ntisense sirna injection group, either (B). 84

95 r e l tiv e m R N A e x p r e ssio n E Z H 2 / - c tin r e l tiv e m R N A e x p r e ssio n E Z H 2 / - c tin A B h o u r p o s t in je c tio n 2 4 h o u r p o s t in je c tio n 1 2 h o u r p o s t in je c tio n 2 4 h o u r p o s t in je c tio n A C S F E Z H 2 n tise n 0.5 u g /u l E Z H 2 n tise n 1 u g /u l A C S F E Z H 2 n tise n 0.5 u g /u l E Z H 2 n tise n 1 u g /u l 85

96 Figure 12. Chnges in mrna levels of EZH2 in the chick's FB (A) nd hypothlmus (B) following EZH2 ntisense injection. At three dys of ge, irds were injected with ACSF nd EZh2 ntisense oligonucleotides t 0.5 ug/ul nd 1 ug/ul into lterl ventricle of the rin. Chicks were scrificed t 12 hr or 24 hr post injection. FB nd hypothlmic tissues were collected. Totl mrna ws isolted nd evluted y PCR using the Syr green method with Ezh2 gene-specific primers (see Tle 1). Gene expression levels were compred with -ctin. Ech vlue is the men stndrd error of the men of 6 individul chicks. EZH2 mrna showed no difference fter EZH2 ntisense t 12 hr post injection in the FB (A) nd hypothlmus t oth 0.5 ug/ul nd 1ug/ul. However, t 24 hr post injection, EZH2 mrna ws significntly inhiited t oth 0.5 ug/ul nd 1ug/ul in oth FB (B) nd hypothlmus (D) histogrm with sme letter indicted insignificnt difference (P > 0.05) while histogrm with different letter indicted significnt difference ( P< 0.05). 86

97 R e l tiv e p r o te in q u n tity o f E Z H 2 / - c tin R e l tiv e p r o te in q u n tity o f H 3 K 2 7 m e 2 / - c tin R e l tiv e p r o te in q u n tity o f E Z H 2 / - c tin R e l tiv e p r o te in q u n tity o f H 3 K 2 7 m e 2 / - c tin A B F B F B A C S F E Z H 2 n tis e n s 0.5 u g /u l E Z H 2 n tis e n s 1 u g /u l 0.0 A C S F E Z H 2 n tis e n s 0.5 u g /u l E Z H 2 n tis e n s 1 u g /u l C A C S F E Z H 2 n tise n s 0.5 ug /ul E Z H 2 n tise n s 1 u g /u l D A C S F E Z H 2 n tise n s 0.5 ug /ul E Z H 2 n tise n s 1 u g /u l E Z H 2 H 3 K 2 7 m e 2 - c tin - c tin h y p o th l m u s h y p o th l m u s A C S F E Z H 2 n tis e n s 0.5 u g /u l E Z H 2 n tis e n s 1 u g /u l 0 A C S F E Z H 2 n tis e n s 0.5 u g /u l E Z H 2 n tis e n s 1 u g /u l 87

98 Figure 13. Chnges in protein nd mrna levels of EZH2 in the chick's FB (A & B) nd PVN (C & D) following EZH2 ntisense injection. At three dys of ge, irds were injected with ACSF nd EZh2 ntisense oligonucleotides t 0.5 ug/ul nd 1 ug/ul into lterl ventricle of rin. 24 hr fter injection, ll chicks were scrificed nd FB nd hypothlmus tissues were collected. Totl protein ws isolted nd evluted y Western Blot. Protein expression levels were compred with -ctin. Ech vlue is the men stndrd error of the men of 6 individul chicks. In FB, lthough EZH2 mrna ws inhiited y oth 0.5 nd 1 ug/ul EZH2 ntisens fter 24 hours s shown previously, EZH2 protein ws only inhiited y EZH2 ntisense t 1 ug/ul (A). The product of EZH2, H3K27me2, followed the sme trend s EZH2 in FB (B). In the hypothlmus, EZH2 protein ws inhiited y EZH2 ntisense t 1 ug/ul ut not 0.5 ug/ul (C). However, H3K27me2 reveled significnt incresed level fter 1ug/ul EZH2 ntisense injection while 0.5 ugl/ul did not ffect its level (D). Histogrm with sme letter indicted insignificnt difference (P > 0.05) while histogrm with different letter indicted significnt difference ( P< 0.05). 88

99 R e l tiv e p r o te in q u n tity o f E Z H 2 / - c tin A C S F E Z H 2 se n s 1 u g /u l E Z H 2 n tise n 1 u g /u l E Z H 2 - c tin h y p o th l m u s A C S F E Z H 2 s e n s 1 u g /u l E Z H 2 n tis e n s 1 u g /u l Figure 14. Chnges in protein levels of EZH2 in the chick's PVN (C & D) following ACSF, EZH2 sense (1 ug/ul) nd EZH2 ntisense (1 ug/ul) injection. At three dys of ge, irds were injected with ACSF, EZh2 sense nd EZH2 ntisense oligonucleotides into lterl ventricle of rin. 24 hr fter injection, ll chicks were scrificed nd hypothlmus ws extrcted. Totl protein ws isolted nd evluted y Western Blot. Gene expression levels were compred with -ctin. Ech vlue is the men stndrd error of the men of 6 individul chicks. EZH2 protein showed no difference etween ACSF nd EZH2 sense injected group t 24 hr post injection. However, EZH2 ntisense significntly inhiited EZH2 protein compred to ACSF injected group. Histogrm with sme letter indicted insignificnt difference (P > 0.05) while histogrm with different letter indicted significnt difference ( P< 0.05). 89