Received 25 Jn 216 Accepted 19 Sep 216 Pulished 31 Oct 216 DOI: 1.138/ncomms13283 OPEN Monitoring peripherl nerve degenertion in ALS y lel-free stimulted Rmn scttering imging Feng Tin 1,2,3,, Wenlong Yng 4,, Dniel A. Mordes 1,2,3,5, Jin-Yun Wng 1,2,3, Johnny S. Slmeh 6, Jonie Mok 1,2,3, Jennie Chew 7, Arti Shrm 8, Ester Leno-Durn 1,2, Stomi Suzuki-Uemtsu 1,2,3,w, Noki Suzuki 1,2,3,w, Steve S. Hn 1,2,3,w, F-Ke Lu 4,9, Minio Ji 4,w, Rosnn Zhng 1,2,3, Yue Liu 1,2,3,w, Jck Strominger 1,2, Neil A. Shneider 8, Leonrd Petrucelli 7, X. Sunney Xie 4 & Kevin Eggn 1,2,3 The study of myotrophic lterl sclerosis (ALS) nd potentil interventions would e fcilitted if motor xon degenertion could e more redily visulized. Here we demonstrte tht stimulted Rmn scttering (SRS) microscopy could e used to sensitively monitor peripherl nerve degenertion in ALS mouse models nd ALS utopsy mterils. Three-dimensionl imging of pre-symptomtic SOD1 mouse models nd dt processing y correltion-sed lgorithm reveled tht significnt degenertion of peripherl nerves could e detected coincidentlly with the erliest detectle signs of muscle denervtion nd preceded physiologiclly mesurle motor function decline. We lso found tht peripherl degenertion ws n erly event in FUS s well s C9ORF72 repet expnsion models of ALS, nd tht seril imging llowed long-term oservtion of disese progression nd drug effects in living nimls. Our study demonstrtes tht SRS imging is sensitive nd quntittive mens of mesuring disese progression, gretly fcilitting future studies of disese mechnisms nd cndidte therpeutics. 1 Deprtment of Stem Cell nd Regenertive Biology, Hrvrd Stem Cell Institute, Hrvrd University, Cmridge, Msschusetts 2138, USA. 2 Deprtment of Moleculr nd Cellulr Biology, Hrvrd Stem Cell Institute, Hrvrd University, Cmridge, Msschusetts 2138, USA. 3 Stnley Center for Psychitric Reserch, Brod Institute of MIT nd Hrvrd, Cmridge, Msschusetts 2142, USA. 4 Deprtment of Chemistry nd Chemicl Biology, Hrvrd University, Cmridge, Msschusetts 2138, USA. 5 Deprtment of Pthology, Msschusetts Generl Hospitl, Boston, Msschusetts 2114, USA. 6 Deprtment of Neurology, University of Msschusetts Memoril Medicl Center, Worcester, Msschusetts 1655, USA. 7 Deprtment of Neuroscience, Myo Clinic, 45 Sn Plo Rod, Jcksonville, Florid 32224, USA. 8 Deprtment of Neurology, Columi University Medicl Center, New York, New York 132, USA. 9 Deprtment of Neurosurgery, Brighm nd Women s Hospitl, Hrvrd Medicl School, Boston, Msschusetts 2215, USA. These uthors contriuted eqully to this work. w Present ddresses: Deprtment of Trnsplnttion, Reconstruction nd Endoscopic Surgery, Division of Advnced Surgicl Science nd Technology, Tohuku University, Sendi 98-8572, Jpn (S.S.-U.); Deprtment of Neurology, Tohoku University School of Medicine, Sendi 98-8574, Jpn (N.S.); Deprtment of Neurology, Msschusetts Generl Hospitl, Boston, Msschusetts 2114, USA (S.S.H.); Deprtment of Physics, Fudn University, Shnghi 2433, Chin (M.J.); School of Life Sciences, Tsinghu University, Beijing 184, Chin (Y.L.). Correspondence nd requests for mterils should e ddressed to X.S.X. (emil: xie@chemistry.hrvrd.edu) or to K.E. (emil:eggn@mc.hrvrd.edu). NATURE COMMUNICATIONS 7:13283 DOI: 1.138/ncomms13283 www.nture.com/nturecommunictions 1
NATURE COMMUNICATIONS DOI: 1.138/ncomms13283 Amyotrophic lterl sclerosis (ALS) ptients suffer from spreding prlysis nd terminl decline often within only 3 yers of dignosis 1,2. Mouse strins crrying humn disese-relted muttions hve emerged s importnt models for understnding the moleculr nd cellulr events tht underlie ALS 1,3 nd phenocopy mny processes oserved in ALS ptients including protein ggregtion nd motor neuron degenertion 3 6. Improved methods for monitoring motor neuron degenertion re much needed, in prt due to the high-degree of vriility of disese progression in rodent models 2,7 nd ALS ptients 7 1. End point nlyses currently used to monitor disese progression in mouse models re lso either lorious histologicl mesurements or ehviourl ssys tht cn e impcted y opertor-specific vrince. Unised imging methods could provide more precise mesures of disese onset nd progression, s well s reduce the length nd miguity currently inherent to trils of cndidte interventions in these models. Stimulted Rmn scttering (SRS) microscopy is n emerging chemicl imging technique tht cn mp the distriution of molecules including lipids, proteins nd nucleic cids in living cells nd tissues sed on their intrinsic moleculr virtion 11 17. SRS nd its precursor coherent nti-stokes Rmn scttering hve een demonstrted s powerful tool for myelin imging 18 26. Although there is growing interest in the importnce of demyelintion in ALS, systemtic studies of peripherl nerve degenertion hve not yet een crried out with SRS 27. Here we descrie the use of SRS imging to visulize peripherl degenertion in severl mouse models of ALS nd humn postmortem tissue. We found tht peripherl nerve degenertion s monitored y SRS imging ws one of the erliest detectle pthologicl events in ALS mouse models nd tht disese progression could e followed relily over time in living nimls through seril imging. We lso demonstrte tht SRS imging could e employed to evlute cndidte therpeutics, confirming tht the compound minocycline significntly slows peripherl nerve degenertion in the mouse. To demonstrte the potentil clinicl utility of our pproch, we showed tht motor nerve degenertion cn similrly e monitored in postmortem tissue from ALS ptients. Results SRS imging of the scitic nerve. In SRS imging, two focused ems of synchronized picosecond pulsed lsers (pump nd Stokes) re overlpped in spce nd time on the smple to produce SRS signl 11 13. The energy difference of the photons of the two ems cn e tuned to mtch the virtionl frequencies of molecules to e imged. As peripherl nerves re chrcterized y myelin sheth with high lipid content, we optimized prmeters to llow imging of lipids in fixed, isolted scitic nerves (Methods section, Fig. 1, Supplementry Fig. 1) 11,19 26. Collection of pump light, fter the lsers SRS interction with the smple in focus, ws chieved y the use of high numericl perture condenser nd nd-pss filter efore detection with photodiode (Methods section nd Supplementry Fig. 1). After signl processing, the myelin sheth nd nodes of Rnvier were clerly visulized in the nerve of wild-type (WT) nimls (Fig. 1d,e) t n improved resolution nd ese of execution reltive to stndrd histologicl methods (Supplementry Fig. 2). To explore whether SRS imging might e useful for studying peripherl nerve degenertion in ALS mouse models, we imged the scitic nerve from the ventrl root to the tiil nerve rnch in 16-week-old mice exhiiting end-stge prlysis (Fig. 1). We found myelin sheth discontinuity nd the ppernce of ovl structures in these nimls tht seemed to e derived from degenerting nerve fiers (Fig. 1e). These ovoid structures were uiquitous throughout the regions imged nd were similr to those oserved in models of nerve injury nd Wllerin degenertion y histology nd electron microscopy (Fig. 1,e nd Supplementry Figs 3,4) 28 31. The three-dimensionl (3D) reconstruction of z-stck SRS imges demonstrted tht these ovoids were primrily irregulr ellipsoids (Fig. 1f j). To determine how erly degenertion in the mouse model could e detected, we performed SRS imging on dissected scitic nerves t 4, 8, 12 nd 16 weeks in trnsgenic mles nd their WT littermtes (Fig. 1). This timecourse suggested erly chnges in the peripherl myelin of mice, s erly s 4 weeks, numerous structurl chnges y 8 weeks nd continued degenertion s nimls ged (Fig. 1d g, Supplementry Figs 3 nd 5). Similr chnges were oserved in smller cohort of 12-week-old trsngenic femle mice (Supplementry Fig. 3). Ovoid structures re primrily lipids. To further understnd the ovoid structures ppering in nerves, we performed whole-mount immunohistochemistry with ntiodies specific to neurofilment (n xonl protein), myelin sic protein (MBP), n ntigen produced y Schwnn cells, nd CD45, which mrks infiltrting mcrophges on nerve injury 32. We then ttempted to visulize ovoid structures y SRS followed y imging of immunofluorescence y confocl microscopy (Methods section). We found tht numer of ovoids in the SRS chnnel were lelled y nti-mbp ntiodies, while there ws negligile overlp etween SRS signl nd neurofilment or CD45 stining (Fig. 2, Supplementry Fig. 6). To further chrcterize the chemicl composition of ovoids, we conducted detiled SRS spectrl scn in oth WT nd scitic nerve. Myelin is minly composed of cholesterol, lipid, protein nd wter 33 35. The lipid components of myelin include glctocereroside, sphingomyelin, phosphtidylcholine nd phosphtidylethnolmine 35. We used the lest squre method to dissocite the spectr of myelin nd degenerted myelin to the seven chemicls mentioned ove (Methods section). We found tht ovoids hd chemicl composition similr to myelin, ut with trend towrds n incresed undnce of lipids nd significnt decrese in wter nd cholesterol (Fig. 2). We, therefore, concluded ovoid structures were likely derived from myelinting cells nd due to their high lipid content we termed them lipid ovoids. Quntifiction of myelin degenertion. We next estlished computtionl method for quntifying lipid ovoids s mens of monitoring peripherl nerve degenertion in longitudinl opticl sections. We selected three distinct lipid ovoids with typicl sizes nd shpes to serve s templtes (Supplementry Fig. 7). These templtes were then used to generte sptil correltion mps for ech of the SRS imges to e quntified. Ares showing strong correltion to the templtes were identified s regions of interest (ROI; Supplementry Fig. 7 nd Methods section). ROIs tht were found to hve n re within pre-estlished rnge (size lrger thn 7.4, 14.8 nd 13. mm 2, respectively) nd tht were sufficiently round (circulrity thresholds for the three templtes were.75,.8 nd.55, respectively) were scored s lipid ovoids (Supplementry Fig. 7). Quntifiction ws repeted for ech of the three templtes in the 2 opticl slices in given z-stck (Supplementry Fig. 7). By summing ll counts in given z-stck, we could clculte the totl numer of ovoids present in given loction. Using this pproch, we quntified myelin degenertion in mice nd WT littermtes etween 4 nd 16 weeks of ge. Consistent with the qulittive chnges we oserved y mnul inspection, there ws trend for n increse in the numer of lipid ovoids in 2 NATURE COMMUNICATIONS 7:13283 DOI: 1.138/ncomms13283 www.nture.com/nturecommunictions
ARTICLE NATURE COMMUNICATIONS DOI: 1.138/ncomms13283 c Birth Symptomtic Endstge 4-week 8-week 12-week 16-week Dorsl root gnglion Scitic nerve L4L4 L5L5 L6L6 PN Site of imging TN SN Peronel nerve d Spinl cord 25 N=5 2 15 1 5 NS 4-week 8-week 12-week 16-week e 4-week Wild-type, 16-week WT Surl nerve Tiil nerve Wild-type Numer of lipid oviods per 5, µm2 per opticl section mouse 16-week Wild-type, 16-week h i g f j Figure 1 Age-point ex vivo SRS imging of scitic nerves identified erly lipid ovoid deposition in mouse ALS model. () Experimentl design of ge-point ex vivo SRS imging. () Scitic nerve dissection from perfused trnsgenic mice or WT littermtes nd the site of SRS imging. (c) Lipid ovoid quntifiction result during disese progression time window of mice. Po.5, Po.1, NS is not significnt. N ¼ 5 for ech strin or ech ge point. Dt re presented s men±s.e.m., nd error rs show s.e.m. (d,e) Ex vivo scitic nerve SRS imges from trnsgenic versus WT non-trnsgenic mice. Scle r, 5 mm. (f j) 3D reconstruction of lipid ovoids t different cross sections. Structures likely to e the wrpping vesicles cn e visulized in the WT 3D reconstruction (the rrow in f). These ovoids hd vriety of sizes, shpes nd structures inside. The width of these ovoids ws elow 15 mm, which ws just smller thn tht of nery myelin sheth. The length of ovoids vries drmticlly. Lrger lipid ovoids cn e round 4 mm long (the rrow in g), while smller ones were lmost round in shpe. Some of the lipid ovoids were empty inside (the rrow in i), some were solid (the rrow in j), while others were hlf empty (the rrow in h) seemingly with unidentified structures inside. Scle r, 2 mm. 4-week-old mice (9±2 ovoids per 5, mm2 per opticl section for versus 5± ovoids per 5, mm2 per opticl section for WT, N ¼ 5 for ech strin, P ¼.13; Fig. 1c) tht ecme significnt t 8 weeks of ge (64±22 ovoids per 5, mm2 per opticl section for versus 5±1 ovoids per 5, mm2 per opticl section, N ¼ 5 for ech strin, P ¼.8; Fig. 1c). As suggested y mnul inspection, the extent of myelin degenertion ws greter nd more significnt in mice t 12 weeks (13±16 ovoids per 5, mm2 per opticl section for versus 12±5 ovoids per NATURE COMMUNICATIONS 7:13283 DOI: 1.138/ncomms13283 www.nture.com/nturecommunictions 3
NATURE COMMUNICATIONS DOI: 1.138/ncomms13283, 16-week CD45 SRS Merge 2 µm 5 4 myelin lipid ovoids WT myelin Aritrry units 3 2 1 C GC SPH PC&PE BSA Wter Figure 2 Chrcteriztion of lipid ovoids suggests tht they were composed of myelin lipids. () SRS lipid imging nd two-photon excited fluorescence imging of CD45-Alex 488 stined scitic nerve from 16-week end-stge mouse. Scle r, 2 mm. () Chemicl spectrl nlysis of lipid ovoids. Po.5, Po.1, Po.1. Anlysed chemicls hve een revited s cholesterol (C), glctocereroside (GC), sphingomyelin (SPH), phosphtidylcholine (PC), phosphtidylethnolmine (PE) nd ovine serum lumin (BSA). Dt re presented s men±s.e.m., nd error rs show s.e.m. 5, mm 2 per opticl section for WT, N ¼ 5 for ech strin, P ¼.15; Fig. 1c) nd 16 weeks (189±52 ovoids per 5, mm 2 per opticl section for versus 2±6 ovoids per 5, mm 2 per opticl section for WT, N ¼ 5 for ech strin, P ¼.12; Fig. 1c). As second mens of quntifiction, we ttempted to mimic clssicl pproch for monitoring peripherl myelin degenertion, the counting of intct nerve fiers in cross section 36. To egin, we reconstructed cross sectionl imges of scitic nerves from z-stck imges (Fig. 1f, Supplementry Fig. 7c). Using SRS imging, ndpss Fourier filter nd contrst enhncement to improve the reconstructed cross sectionl imges, we found tht we could reconstruct nerve fiers present ner the surfce of the scitic nerve (Fig. 1f, Supplementry Fig. 7c). A numer of representtive norml nd locked nerve fiers were selected s templtes nd we pplied method similr to tht used for lipid ovoid quntifiction to count the numer of locked nd intct nerve fiers. This quntifiction method could detect difference etween WT nd nimls. However, significnce ws not reched until 16 weeks of ge (39.4±13.% for versus 5.7±3.% for WT, N ¼ 5 for ech strin, P ¼.3; Supplementry Fig. 8). Although these cross-sectionl mesurements were less sensitive thn the longitudinl mesurement of ovoids, they were consistent with historicl counts, which detected ltertions in myelinted fiers t similr time point 36. We next sked whether degenertion occurred with ny sptil preference long scitic nerves. mice t 8 postntl weeks nd their WT littermtes were studied to cpture erly chnges in myelin (Fig. 1c). All microscopic fields of the freshly perfused scitic nerves were cquired in the lipid chnnel nd integrted y n imge-processing lgorithm (Methods section). The resulting imges provided high-resolution glol view of myelin sheths where ny detiled regions long the nerve could e individully mgnified nd exmined. (Supplementry Fig. 7). Automted quntifiction ws then employed to count the plnr lipid ovoid numers of eight ssigned su-sections from the proximl to the distl end of the nerve. We found tht sites of degenertion were distriuted throughout the 8-week scitic nerve (Supplementry Fig. 9iv vi). In contrst, the WT scitic nerve showed miniml signs of dmge (Supplementry Fig. 9i iii). Both the proximl nd the distl regions of scitic nerve mnifested significnt increse of lipid ovoid deposition over the WT scitic nerve (N ¼ 4, P ¼.84 for proximl region, P ¼.18 for distl region; Supplementry Fig. 9). We did oserve trend for increse in lipid ovoids in distl regions of the nerve, lthough it ws not significnt (N ¼ 4, P ¼.2; Supplementry Fig. 9) nd ws mirrored y similr trend for n incresed numer of distl ovoids in controls. SRS imging in vivo. We next explored the fesiility of seril imging in ALS mouse models y ttempting to imge the scitic nerve in vivo using modified opticl pth (Supplementry Fig. 1). As n initil end-point imging study, mice t 8 weeks, 12 weeks nd 16 weeks, respectively were nesthetized nd immoilized on the microscope stge (Fig. 3 nd Methods 4 NATURE COMMUNICATIONS 7:13283 DOI: 1.138/ncomms13283 www.nture.com/nturecommunictions
NATURE COMMUNICATIONS DOI: 1.138/ncomms13283 ARTICLE mouse Disese onset Endstge Birth 5-week 8-week 11-week 13-week 16-week c SRS in vivo imging d Wild-type 5-week e Wild-type 8-week f Wild-type 11-week g Numer of lipid ovoids per 5, µm 2 per opticl section 25 WT (N =5) (N =5) (Low copy numer) 2 15 1 5 4 5 6 7 8 9 1 11 12 Age of imged mice (weeks) Figure 3 Progressive lipid ovoid deposition ws visulized y long-term seril in vivo SRS imging. () Experimentl design of long-term seril in vivo SRS imging. () Stge lignment for in vivo SRS imging of scitic nerves. (c) Incision size of representtive imged mouse. Scle r, 1 cm. (d f) versus WT long-term seril in vivo SRS imges t different ges during disese progression. Scle r, 5 mm. (g) Quntifiction of lipid ovoids for long-term seril in vivo SRS imging throughout disese progression. Po.5, Po.1. N ¼ 5 for ech strin t ech time point. For 5-week-old mice, P ¼.43. For 8-week-old mice, P ¼.12. For 11-week-old mice, P ¼.31. Dt re presented s men±s.e.m., nd error rs show s.e.m. section). A smll incision ws mde on the hind lim to expose the scitic nerve nd imging ws crried out voiding muscle dmge (Fig. 3c). To remove motion rtifcts from respiring niml, we replced ny movement-distorted portion of imges with the verge of the previous nd the susequent opticl slices (Supplementry Fig. 11). We gin found significnt difference etween nd WT mice t lest s erly s 8 weeks (74±23 ovoids per 5, mm 2 per opticl section, N ¼ 5 for versus 4±1 ovoids per 5, mm 2 per opticl section, N ¼ 4 for WT, P ¼.31; Supplementry Fig. 12). Following our initil success with in vivo imging, we ttempted to monitor the rte of myelin degenertion in NATURE COMMUNICATIONS 7:13283 DOI: 1.138/ncomms13283 www.nture.com/nturecommunictions 5
NATURE COMMUNICATIONS DOI: 1.138/ncomms13283 individul nimls y performing seril SRS imging. We selected severl time points for long-term survivl imging in ttempt to cpture the erliest onset of lipid ovoid deposition (Fig. 3). We found tht repeted imging ws well tolerted (Supplementry Fig. 1, Supplementry Movie 1) nd tht imging did not dmge the nerve of control nimls over the time investigted (Fig. 3d g). In contrst, the progression of nerve degenertion could e redily monitored in mice (Fig. 3d f). Lipid ovoid numer ws significntly incresed in mice s group t 5 weeks postntl (12±4 ovoids per 5, mm 2 per opticl section for versus 3±1 ovoids per 5, mm 2 per opticl section for WT, N ¼ 5 for ech strin, P ¼.43; Fig. 3d,g). By 8 weeks postntl, ech of the individul nimls exhiited significntly incresed numer of lipid ovoids reltive to the verge of WT nimls (67±9 ovoids per 5, mm 2 per opticl section for versus 5±1 ovoids per 5, mm 2 per opticl section for WT, N ¼ 5 for ech strin, P ¼.12; Fig. 3e,g). This significnt increse in myelin degenertion ws even greter t 11 weeks of ge (175±28 ovoids per 5, mm 2 per opticl section for versus 4±1 ovoids per 5, mm 2 per opticl section, N ¼ 5 for ech strin, P ¼.31; Fig. 3g). Interestingly, one niml exhiited much slower disese progression rte (Fig. 3g). This outlier ws lter confirmed y quntittive PCR to crry lower copy numer of the humn trnsgene 37 (Supplementry Fig. 13), which ws consistent with previous oservtions tht low copy-numer mice showed delyed nd less-severe phenotypes 37. Ordering events in nerve degenertion. To further understnd the timing of peripherl myelin degenertion reltive to other events occurring during disese progression in ALS mouse models, we performed oth SRS imging nd electromyogrphy (EMG) in 5-week-old nimls 38. Consistent with previous studies, quntittive mesures of motor neuron physiology including compound motor ction potentil (CMAP), motor unit size nd motor unit numer estimtion (MUNE) did not show significnt differences etween 5-week-old nd WT mice (Fig. 4 c nd Supplementry Tle 1) 38. In contrst, five of six mice imged y SRS exhiited n incresed numer of lipid ovoids reltive to the verge of their WT littermtes. Furthermore, the overll comprison etween the nd WT cohorts showed significnt increse in lipid ovoid numer within ALS nimls (P ¼.34; Fig. 4d). By needle EMG, three out of six mice within the sme cohort displyed ctive denervtion with firilltion potentils nd positive shrp wves, while ll WT littermtes were negtive (Fig. 4e). These results confirmed tht firilltion potentils nd positive shrp wves in needle EMG re one of the erliest detected minimlly invsive iomrkers in the mouse model 2,38. Using SRS imging, degenertive myelin morphology ws oserved in 4 out of 6 mice, which gin ws not oserved in ny WT controls (Fig. 4f). Thus our results indicted tht peripherl nerve degenertion s mesured y SRS imging reched significnce t time comprle to the initil oservtion of muscle denervtion y EMG nd weeks efore the first detectle decline in motor nerve function. This erly onset of nerve degenertion ws confirmed in lrger group of control nd nimls t 6 weeks of ge (Supplementry Fig. 14). Evlution of minocycline in mice. The ntiiotic minocycline hs previously een reported to significntly extend the lifespn of mice 39, though it lter filed to show efficcy in clinicl tril 4. We wondered whether SRS imging ws sufficiently quntittive nd sensitive to e useful in revisiting the efficcy of this compound in mouse model. As we hd reproducily oserved peripherl nerve degenertion t 5 weeks of ge y SRS, we initited minocycline tretment t this ge. nimls were rndomly ssigned to either minocycline (N ¼ 15) or vehicle tretment (N ¼ 13) groups, with dily dministrtion vi intrperitonel injection (Methods section). In ddition to control SRS imging t 5 weeks, just efore initil drug dministrtion, we performed two lter survivl SRS imging sessions on ech niml in the study t 8 nd 11 weeks of ge (Fig. 5). Before the onset of drug dministrtion (5 weeks), we found no difference etween the overll numer of lipid ovoids in trnsgenic nimls rndomized to the minocycline nd vehicle tretment groups (11±3 ovoids per 5, mm 2 per opticl section for minocycline treted versus 1±2 ovoids per 5, mm 2 per opticl section for vehicle treted, P ¼.82; Fig. 5c nd Supplementry Fig. 15), while oth surpssed tht of WT nimls of the sme ge (3±1 ovoids per 5, mm 2 per opticl section). In contrst, following 3 weeks of minocycline tretment nimls receiving minocycline showed significntly lower urden of lipid ovoids compred with the vehicle controls (3±9 ovoids per 5, mm 2 per opticl section for minocycline versus 62±12 ovoids per 5, mm 2 per opticl section for vehicle, P ¼.36; Fig. 5,c), though oth groups were elevted over WT control nimls (5±1 ovoids per 5, mm 2 per opticl section). This effect of minocycline ws sustined nd ecme more significnt t the 11-week imging time point (64±1 ovoids per 5, mm 2 per opticl section for minocycline versus 96±7 ovoids per 5, mm 2 per opticl section for vehicle, P ¼.12; Fig. 5c nd Supplementry Fig. 15), lthough the lipid ovoid deposition in these nimls ws not completely rescued s demonstrted y comprison with WT nimls (4±1 ovoids per 5, mm 2 per opticl section for WT t 11 weeks of ge; Fig. 3g). To rule out the possiility tht mice with lower copy numer of the trnsgene might hve een indvertently rndomized to the minocycline group, we determined the reltive copy numer of the trnsgene in ech of the study nimls. Although we did oserved vrition in copy numer within the overll cohort, we found tht this vrition ws equitly distriuted etween the minocycline nd vehicle controls, with no significnt difference in copy numer etween the groups (Supplementry Fig. 15c). We next ttempted to relte trnsgene copy numer, drug effects nd the extent of peripherl nerve degenertion s mesured y lipid ovoid ccumultion (Fig. 5d). Before initil drug tretment t 5 weeks, we found there ws positive correltion etween trnsgene copy numer nd lipid ovoid ccumultion tht ws similr in the two seprte groups of nimls. When SOD1 copy numer ws incorported into nlysis of nimls t 8 weeks, the positive correltion etween SOD1 copy numer nd ovoid numer remined, however, the overll frequency of lipid ovoids ws significntly reduced cross nimls of ll copy numers in the minocycline group (Fig. 5e). A similr pttern ws oserved t 11 weeks (Fig. 5f). Thus SRS imging fcilitted the finding tht minocycline ws le to modestly ut significntly reduce the urden of peripherl nerve injury regrdless of trnsgene copy numer. Imging SOD1G37R mice. To determine whether the erly degenertive phenotype we oserved were conserved in dditionl ALS models, we imged dissected nerves from SOD1G37R trnsgenic mice 4. These mice exhiit lter disese onset nd slower rte of phenotypic progression thn oserved in 6 NATURE COMMUNICATIONS 7:13283 DOI: 1.138/ncomms13283 www.nture.com/nturecommunictions
NATURE COMMUNICATIONS DOI: 1.138/ncomms13283 ARTICLE 6 P =.78 11 P =.76 Compound motor ction potentil (mv) 5 4 3 2 Motor unit size (µv) 1 9 8 7 c 1 6 6 WT WT d P =.28 P =.3 25 Motor unit numer estimtion 5 4 3 2 Numer of lipid ovoids per 5, µm 2 per opticl section 2 15 1 5 1 WT WT e Rtio of mice showing positive shrp wves in needle EMG 1.8.6.4.2 N = 4 WT Norml Anorml 1+ Anorml 2+ f N = 6 N = 4 1 Rtio of mice showing myelin degenertion sign in SRS.8.6.4.2 N = 6 WT Asent Mild Ovious Figure 4 The sensitivity of SRS imging ws comprle to EMG for 5-week versus WT mice. ( c) Quntittive prmeters mesured y EMG for 5 weeks versus WT mice. (d) Quntifiction of lipid ovoids y SRS imging for 5 weeks versus WT mice. (e) Rtio of mice showing qulittive positive shrp wves y EMG for 5-week versus WT mice. (f) Rtio of mice showing myelin degenertion signs y SRS imging for 5 weeks versus WT mice. N ¼ 4 for versus N ¼ 6 for WT. Dt re presented s men±s.e.m., nd error rs show s.e.m. mice, mking it preferred model for some studies. Intriguingly, t 1 weeks of ge, the erliest time point exmined, we lredy oserved significnt increse in degenertion reltive to controls (31±12 ovoids per 5, mm 2 per opticl section for SOD1G37R versus 4±1 ovoids per 5, mm 2 per opticl section for WT, N ¼ 4 for ech strin, P ¼.43; Supplementry Fig. 16d). This gretly preceded the erliest chnges thus fr reported y histologicl nlysis (fter 25 weeks) 4,27. Further imging t symptom onset (44 weeks) nd end-stge (52 weeks) showed further significnt increse in ovoid numers (Supplementry Fig. 16). We oserved numer of lipid ovoids in end-stge SOD1G37R nimls tht ws comprle to tht we found in mice (239±17 ovoids per 5, mm 2 per opticl section for SOD1G37R 189±52 ovoids per 5, mm 2 per opticl section for ; Fig. 1c nd Supplementry Fig. 16d). Non-SOD1 mouse models of ALS. We sought to employ SRS imging in dditionl ALS models to expnd its utility. The next model we exmined ws creted y deno-ssocited virus (AAV)-medited expression of the C9ORF72 hexonucleotide repet expnsion throughout the centrl nervous system 5 (Fig. 6). This C9ORF72 model hs een reported to exhiit neuronl RNA foci, dipeptide repet protein inclusions, TDP-43 pthology, corticl nd Purkinje cell loss s well s locomotive defects y 24 weeks 5. By imging ex vivo scitic nerve smples of AAV-C9ORF72 mice crrying pthogenic (G 4 C 2 ) 66 repets (66R) (N ¼ 12) nd their littermte controls crrying (G 4 C 2 ) 2 repets (2R) (N ¼ 13) etween 26 weeks nd 28 weeks, we oserved lipid ovoid deposition in the mjority of 66R mice (Fig. 6). Quntifiction demonstrted significntly incresed numer of lipid ovoids in 66R nimls (29±6 ovoids per 5, mm 2 per opticl section for 66R versus 13±3 ovoids per 5, mm 2 per opticl section for 2R, P ¼.24; Fig. 6c). We next exmined mutnt FUS trnsgenic model in which either humn WT or P525L mutnt FUS is expressed from the mouse MAPT (tu) promoter (Fig. 6d). In this mouse model, It ws een shown B3% of motor neurons were lost nd B4% of tiilis nterior neuromusculr junctions were denervted t 52 weeks postntl 41. By exmining scitic nerves from hfusp525l mice, we found significnt lipid ovoid deposition ws lredy NATURE COMMUNICATIONS 7:13283 DOI: 1.138/ncomms13283 www.nture.com/nturecommunictions 7
NATURE COMMUNICATIONS DOI: 1.138/ncomms13283 Vehicle 8-week, Minocycline Endstge mouse Birth 5-week Minocycline tretment 8-week 11-week 16-week SRS in vivo imging c Numer of lipid ovoids per 5, µm 2 per opticl section P =.36 18 Minocycline P =.12 16 Vehicle 14 12 1 8 6 P =.82 4 2 4 5 6 7 8 9 1 11 12 Age of mouse (weeks) d e f Numer of lipid ovoids per 5, µm 2 per opticl section 4 35 3 25 2 15 1 5.8.85.9.95 5-week R 2 =.2279 R 2 =.838 1 1.5 1.1 1.15 1.2 Reltive enrichment of Numer of lipid ovoids per 5, µm 2 per opticl section 18 16 14 12 1 8 6 4 2 8-week.8.85.9.95 R 2 =.1334 1 R 2 =.193 1.5 1.1 1.15 1.2 Reltive enrichment of Numer of lipid ovoids per 5, µm 2 per opticl section 18 16 14 12 1 8 6 4 2 11-week.8.85.9.95 R 2 =.97 1 R 2 =.291 1.5 1.1 1.15 1.2 Reltive enrichment of Figure 5 SRS imging of under minocycline tretment. () Experimentl design of long-term seril in vivo SRS imging for under minocycline tretment. () versus WT long-term seril in vivo SRS imges t 8 weeks postntl. Scle r, 5 mm. (c) Quntifiction of lipid ovoids for long-term seril in vivo SRS imging throughout disese progression. (d f) Correltion nd regression nlysis etween copy numer nd lipid ovoid deposition t 5 weeks (d), 8 weeks (e) nd 11 weeks (f) of postntl ge. present in these mice t 12 weeks of ge (43±6 ovoids per 5, mm 2 per opticl section for hfusp525l versus 15±2 ovoids per 5, mm 2 per opticl section for hfus WT, P ¼.8; Fig. 6e,f). Overll, these results vlidte SRS imging s powerful resource to study peripherl nerve degenertion in emerging ALS mouse models. Imging Scitic nerve crush nd EAE models. We resoned tht if the lipid ovoids we oserved in ALS models were resulting from injuring of the underlying neuron, then similr chnges in myelin morphology might e oserved distlly to the site of nerve crush, s result of Wllerin degenertion. We therefore next tested the utility of SRS imging in the context of mouse model of xonl injury y scitic nerve crush 22,23,26,42. The scitic nerve of one hind lim ws crushed in WT nd nimls then imged distlly of the injury t oth 1 week nd 3 weeks fter initil injury (Fig. 7). We found WT nimls exhiited significnt peripherl nerve degenertion t 1 week (662±287 ovoids per 5, mm 2 per opticl section in crush versus 4±3 ovoids per 5, mm 2 per opticl section in the uncrushed, N ¼ 4 for ech group, P ¼.38; Fig. 7-d). At 3 weeks fter injury, significnt recovery ws oserved (222±9 ovoids per 5, mm 2 per opticl section for 8 NATURE COMMUNICATIONS 7:13283 DOI: 1.138/ncomms13283 www.nture.com/nturecommunictions
NATURE COMMUNICATIONS DOI: 1.138/ncomms13283 ARTICLE AAV-C9RF72 (n)ggggcc RNA foci, inclusion odies, MN loss, Lipid ovoids motor defects detected y SRS d FUS P525L trnsgenic mouse Lipid ovoids detected y SRS Severe MN loss nd denervtion in TA Birth 24-week 26-week 28-week Birth 4-week 12-week 5-week AAV-C9RF72 repets, 26-week to 28-week 2 GGGGCC 66 GGGGCC e 12-week Tu::FUS WT Tu::FUS P525L c f AAV-C9RF72 FUS trnsgenic N = 13 N = 12 N = 7 N = 6 P =.24 P =.8 9 7 Numer of lipid ovoids per 5, µm 2 per opticl section 75 6 45 3 15 Numer of lipid ovoids per 5, µm 2 per opticl section 6 5 4 3 2 1 2 GGGGCC 66 GGGGCC Tu::FUS WT Tu::FUS P525L Figure 6 SRS imging of non-sod1 mouse models demonstrted the generl utility in studying ALS. () Disese progression of AAV-C9ORF72 repet expnsion over-expression mouse model of ALS. () Representtive SRS imges of AAV-C9ORF72 mouse model of ALS showing helthy nerve fier morphology for 2 repets (2R) nd lipid ovoid deposition (66R). Scle r, 5 mm. (c) Lipid ovoid quntifiction result of AA-C9ORF72 mouse model of ALS. Dt re presented s men±s.e.m., nd error rs show s.e.m. (d) Disese-relted phenotype progression in FUSP525L mouse ALS model. (e) FUSP525L versus FUS WT ex vivo SRS imges t 12 weeks of ge. Scle r, 5 mm. (f) Quntifiction of lipid ovoids nd sttisticl significnce nlysis for FUSP525L mouse model of ALS. Dt re presented s men±s.e.m., nd error rs show s.e.m. 3 weeks fter crush, N ¼ 4 for ech group, P ¼.38 versus 1 week of recovery; Fig. 7,d). We found tht the morphology of degenerting fiers fter nerve crush ws very similr to tht oserved in SOD1 mouse models, nd when whole mount immunofluorescence ws compred with SRS signl it gin suggested tht they were not likely to e infiltrting immune cells (Supplementry Fig. 17). We lso exmined the response to nerve crush in mice nd found tht, with the exception of n elevted seline of lipid ovoids reltive to controls, these mice initilly responded similrly to injury (28±1 ovoids per 5, mm 2 per opticl section for uncrushed versus 59±126 ovoids per 5, mm 2 per opticl section for crushed t 1 week, N ¼ 4 for ech group, P ¼.38; Supplementry Fig. 18). However, peripherl recovery in mice ws muted nd did not rech significnce (59±126 ovoids per 5, mm 2 per opticl section for 1 week fter crush versus 295±135 ovoids per 5, mm 2 per opticl section for 3 weeks fter crush, N ¼ 4 for ech group, P ¼.62; Supplementry Fig. 18c). We then compred the nture of peripherl nerve degenertion we oserved in ALS nd nerve crush models with tht resulting from the dmge of primry myelinting glil cells in the Autoimmune encephlomyelitis (EAE) model of multiple sclerosis (Fig. 7e) 43,44. In lte stge EAE nimls, we clerly NATURE COMMUNICATIONS 7:13283 DOI: 1.138/ncomms13283 www.nture.com/nturecommunictions 9
NATURE COMMUNICATIONS DOI: 1.138/ncomms13283 Wildtype mouse Birth Crush Imging 4-week 5-week Imging 8-week Numer of lipid ovoids per 5, µm 2 per opticl section 1, 8 6 4 2 1-week recovery Uncrushed Distl of crush N = 4 3-week recovery c Wild-type, 1 week fter recovery d Wild-type, 3 weeks fter recovery Uncrushed Distl of crush Uncrushed Distl of crush e Immuniztion Endstge SJL mouse Erly stge Lte stge f Birth EAE score: 1 Helthy Limp til Untreted EAE erly stge EAE lte stge Score: Score: 1.5 Score: 5 2 3 4 5 Wek git Hindlim Totl Moriund prlysis prlysis g EAE MS model P =.16 35 P =.89 P =.22 Numer of lipid oviods per 5, µm 2 per opticl section 3 25 2 15 1 5 Untreted EAE erly stge EAE lte stge Figure 7 Lipid ovoid deposition could result from initil xon degenertion. () Experimentl design of SRS imging for scitic nerve crush model. () Lipid ovoid quntifiction result of scitic nerve crush experiment. Po.5. N ¼ 4 for ech strin or ech ge point. Dt re presented s men±s.e.m., nd error rs show s.e.m. (c,d) Scitic nerve SRS imges of WTmice fter crush nd recovery. Scle r, 5 mm. (e) MS like phenotypes nd disese progression of EAE model. (f) Scitic nerve SRS imges of EAE model. The left rrow shows demyelintion morphology, nd the right rrow shows remyelintion morphology. Scle r, 5 mm. (g) Lipid ovoid quntifiction result of EAE model SRS imging. 1 NATURE COMMUNICATIONS 7:13283 DOI: 1.138/ncomms13283 www.nture.com/nturecommunictions
NATURE COMMUNICATIONS DOI: 1.138/ncomms13283 ARTICLE Humn spinl cord Dorsl Ventrl ALS ptients Controls Postmortem dissection from ptients Fixtion in formlin SRS imging Numer of lipid ovoids per 5, µm 2 per opticl section 8 7 6 5 4 3 2 1 Ventrl P =.6 P =.88 Dorsl Control ALS N = 4 c Ventrl roots d Dorsl roots Control ALS Control ALS e mouse model Methods Pthology mnifesttion Qulittive myopthy nd neuropthy Needle EMG Lipid ovoid deposition SRS imging NMJ denervtion IHC, genetic leling Axonl trnsport defect TEM Motor unit physiology chnges EMG MN morphology chnges IHC nd cell counts Peripherl immunity ctivtion, weight loss, symptom onset IHC nd count, ehviorl tests Astrocyte infiltrtion IHC nd cell counts Lim prlysis Behviorl tests Endstge Behviorl tests Mssive MN cell ody deth IHC nd cell counts 5 6 7 8 9 1 12 14 15 16 18 Reference: 2,46 47,48 2,49 38 5 3,51,52 53,54 2,55 3 36 Age (weeks) Figure 8 SRS imging identified lipid ovoid deposition in humn ALS ptient smples. () Experimentl design of SRS imging for humn ALS ptient smples. () Lipid ovoid quntifiction of ventrl (motor fier enriched) nd dorsl (sensory fier enriched) root smples from either humn ALS ptients or controls. Po.5. N ¼ 4 for ech group. Dt re presented s men±s.e.m., nd error rs show s.e.m. (c) SRS imging of ventrl root nerve fiers from humn ALS ptients versus controls. Scle r, 5 mm. (d) SRS imging of dorsl root nerve fiers from humn ALS ptients versus controls. Scle r, 5 mm. (e) Timeline of ALS ssocited pthologicl chnges nd dignostic methods. EMG, electromyogrphy; IHC, immunohistochemistry; MN, motor neuron; NMJ, neurl musculr junction; SRS, stimulted Rmn scttering; TEM, trnsmission electron microscope. identified severl sites where the morphology of myelin ws ltered (Fig. 7f). Following cross sectionl reconstruction of these regions from opticl slices, n pprent onion ul morphology could e visulized (Supplementry Fig. 19). Our nlysis indicted tht reltive to controls EAE nerves displyed n incresed proportion of under-myelinted s well s over-myelinted nerve fiers (Supplementry Fig. 19,c). In contrst, we oserved only modest increse in lipid ovoid numer even in moriund EAE nimls (14±4 ovoids per 5, mm 2 per opticl section for untreted versus 25±5 ovoids per 5, mm 2 per opticl section for EAE lte stge, N ¼ 4 for ech group, P ¼.22; Fig. 7g nd Supplementry Fig. 17). This chnge ws not comprle with the drmtic increse of lipid ovoids etween SOD1 end-stge nd non-crrier nimls. Our study demonstrted tht ALS nd MS models showed distinct peripherl myelin phenotypes. SRS imging of humn nerve smples. We finlly sought to vlidte the utility of SRS imging in ALS ptient smples 1.To this end, we imged post-mortem lumr nerve smples from spinl cord ventrl (motor) nd dorsl (sensory) roots of spordic ALS ptients nd controls (Fig. 8 nd Methods section). The quntifiction of z-stck SRS imges demonstrted there hd een sustntil degenertion within the motor nerve fiers NATURE COMMUNICATIONS 7:13283 DOI: 1.138/ncomms13283 www.nture.com/nturecommunictions 11
NATURE COMMUNICATIONS DOI: 1.138/ncomms13283 from ALS ptients (51±19 ovoids per 5, mm 2 per opticl section for ALS ventrl versus 2±2 ovoids per 5, mm 2 opticl section for control ventrl, N ¼ 4 for ech group, P ¼.4; Fig. 8,c). In contrst, no significnt difference ws oserved etween dorsl roots from ALS ptients nd those from controls (7±6 ovoids per 5, mm 2 opticl section for ALS dorsl versus 8±5 ovoids per 5, mm 2 opticl section for control dorsl, N ¼ 4 for ech group, P ¼.88; Fig. 8,d). Discussion Here we show tht SRS imging cn used to visulize peripherl nerve degenertion in ALS mouse models nd post mortem mterils. The sensitivity of SRS imging ws comprle to EMG for detecting erly signs of neuronl degenertion, ut lso provided richness of structurl informtion out the peripherl nerve tht EMG does not 45. Strikingly, SRS imging identified lipid ovoids in some mice tht showed no ctive denervtion y EMG, suggesting tht SRS lone or in comintion with EMG my llow for the erliest in vivo detection of peripherl nervous degenertion in mouse models yet reported (Fig. 8e) 2,3,36,38,46 55. In study of minocycline, we found tht SRS imging enled detection of the slowing of peripherl nerve degenertion induced y this compound. In the future, SRS could e used to potentilly shorten the time frme for the evlution of cndidte ALS therpeutics s well s provide more quntittive view of drug effects. Our SRS imging studies suggest tht lipid ovoids ccumulting in the peripherl nerves of pre-symptomtic ALS nimls were most likely lipids derived from degenerting myelinting cells. We lso visulized nd quntified the peripherl demyelinting/remyelinting processes in the EAE model of MS, which were sustntilly distinct from those oserved in ALS models 56. The EAE model did not disply ny significnt lipid ovoid deposition in peripherl nerves. Insted, we found tht the pttern of degenertion in ALS mouse models nd ptient smples ws more similr to tht which occurs during Wllerin degenertion following xon injury 31,57. Notly, it hs een shown tht introducing the Wllerin degenertion slow (Wld(s)) muttion into the ckground slowed disese progression 58. Similrly, forced expression of ATF3 (ctivting trnscription fctor 3), which is normlly induced fter scitic nerve injury, lso delyed muscle trophy nd extend the lifespn of mice 57. More intriguingly ut less well sustntited, SARM1, whose ctivtion hs een oserved fter Wllerin degenertion, flls within region of linkge disequilirium ssocited with ALS 59. The imging studies reported here would seem to support the notion tht signlling pthwys involved in Wllerin degenertion wrrnt further ttention in ALS. With the development of fier lser SRS nd SRS sed histologicl methods for studying rin tumours, the possiility of moving SRS microscope into surgery room or clinic for rpid lel-free histology is eing explored 6 63. Our studies hve shown tht SRS imging might lso eventully hve utility for pthologists in rpid nlysis of peripherl postmortem smples from ALS ptients. However, given the current stte of SRS technologies the gretest immedite utility of SRS imging will likely e in the study of pre-clinicl niml models. Our nturl history study of peripherl nerve injury in vriety of ALS models y SRS imging further indictes tht sustntil nerve degenertion hs occurred efore the disese-relted chnges in the outwrd ehviourl mesures tht re often used to determine disese onset nd the commencement of drug dministrtion studies in SOD1 models. As result, the therpeutic window remining t the lter time when mny studies normlly egin my e smll, which could explin why detection of drug effects in these models hs often een difficult. Incorporting SRS imging s mens of determining the erliest onset of neurl degenertion nd then proceeding with initition of tretment my llow for improved sensitivity in detecting the effects of experimentl interventions. Methods Mouse models of ALS. The mouse ALS models used in this study were the strin crrying humn ALS-ssocited trnsgene (B6SJL-Tg(SOD1G93A) 1Gur/J, JAX Lortory, No. 2726) or SOD1G37R (B6. Cg-Tg(SOD1G37R) 1Dwc/J, JAX Lortory, No. 16149). To cquire nimls with nd without trnsgenic in the sme litter s the optiml comprison, we crossed mle mice to non-trnsgenic genetic ckground strin C57BL/6J (JAX Lortory, No. 664). Pups were genotyped on the trnsgene using primers specified y JAX Lortory (oimr113, oimr114, oimr7338 nd oimr7339) efore wening on dy 21 postntl. After genotyping, mice crrying nd their non-trnsgenic littermtes were fed with specil cre pproved for ALS disese study. Only mles (unless specified) were selected nd rised for the SRS imging experiments unless prticulrly mentioned. Mice with trnsgene were weighed every 3 dys fter 8 weeks postntl to monitor the onset of muscle trophy. Normlly, mice strt to lose weight vrying from 12 to 14 weeks. ALS ehviourl stging monitor ws performed strting from 12 weeks postntl, with regulr diet nd hydrogel provided in the cges. mice under ALS level 2 were monitored every 3 dys, while mice over ALS level 3 were monitored dily until scrifice for imging or histology. Unless mentioned specificlly, mle nimls were sujected to ex vivo nd in vivo SRS imging for this study to reduce the vriility resulted from popultion of mixed genders. Copy numer quntittive PCR tests were performed sed on JAX Lortory protocols. The FUSP525L nd AAV-C9ORF72 mouse model of ALS hs een descried previously 5,41. All experimentl protocols nd procedures were pproved y the Institutionl Animl Cre nd Use Committee of Hrvrd University. These procedures were consulted nd supervised y the veterinrin of the Office of Animl Resources. Ex vivo SRS imging. Mice crrying nd ge-mtched non-trnsgenic WT littermtes were imged t the ge of 4, 8, 12 nd 16 weeks postntl. Mice were perfused with 4% prformldehyde, nd their scitic nerves were dissected from oth legs. The scitic nerves were rinsed in PBS nd mounted on glss slides covered y coverslip for SRS imging. As descried in the min text, we minly imged the midpoint of the dissected scitic nerve (from the ventrl root nd ending t the rnching point of tiil nerve). The imging ws performed on n Olympus BX61WI upright microscope with FV3 scnning unit. The lser source employed ws pico EMERALD from A.P.E., which provided comined em of two 7-ps lsers, whose wvelengths were 1,63 nm for Stokes em nd tunle output for pump em respectively. In this study, we focus on the CH high wvenumer region (2,83 3,2 cm 1 ) for imging nd hyperspectrl imging. For morphology imging, the energy difference of the photons from these two wvelengths ws round 2,85 cm 1 which corresponded to the CH 2 symmetric stretching virtion in lipids (Supplementry Fig. 4). The 1,63 nm lser ws modulted y n electro-opticl modultor t 1 MHz. The lsers were trnsmitted to scnning microscope nd scnned y x y glvo scnners nd focused on the smple y high NA ojective. The ojective we pplied in the experiment ws the IR version Olympus UPLANAPO 6X wter immersion ojective (NA 1.2). The trnsmitted light through the smple ws collected y Nikon oil immersion condenser (NA 1.4) nd filtered y nd pss filter (Chrom, coherent nti-stokes Rmn scttering 89/22) to remove the 1,64 nm lser component. The pump lser ws susequently illuminted on lrge re silicon photodiode detector (OSI Optoelectronics, S-1CL). The detected photocurrent ws mplified y filtered trns-impednce mplifier (Supplementry Fig. 2) nd sent to the lock-in mplifier for demodultion. The demodulted signl ws then sent to the dt cquisition unit of the microscope nd ws eventully sent to personl computer to show the imges of the smple tissue. The opticl power used on the smple is out 9 mw of pump nd 8 mw of IR. The signl we get from myelin is out 11 mv with photodiode connected to 5-ohm resistor. Our premplifier will mplify this signl y out 23 times. However, the signl level depends on the thickness of smple, so it is vries depending on the smples preprtion process. The imging speed ws 1 s/frme or 1.5 s/frme in few occsions, nd the imge size ws 512 y 512 pixels. The 3D reconstruction ws generted y imgej plug-in clled volume viewer with ll the three dimension scles corresponds to rel life sizes. Whole-mount stining. The whole mount stining ws performed sed on previously pulished procedure 32. After eing fixed in 4% prformldehyde t 4 C overnight, scitic nerves were wshed in 1% TritonX-1/PBS for 15 min. The nerve smples were then incuted in 1% donkey serum/1% TritonX-1/PBS t 12 NATURE COMMUNICATIONS 7:13283 DOI: 1.138/ncomms13283 www.nture.com/nturecommunictions
NATURE COMMUNICATIONS DOI: 1.138/ncomms13283 ARTICLE 4 C overnight. On the next dy, the nerves were incuted in nti-mbp (ctlog numer: MAB386MI, dilution fctor: 1 5)/nti-Neurofilment (ctlog numer: SMI-32P, dilution fctor: 1 5)/1% donkey serum/1% TritonX-1/PBS for 72 h t 4 C. After incution, the nerves were wshed with 1% TritonX-1/PBS for 6 times (1 h ech) t room temperture. The smples were then trnsferred into fluorescence-lelled secondry ntiodies in 1% donkey serum/1% TritonX-1/ PBS. After 48 h t 4 C, the nerves were wshed with 1% TritonX-1/PBS for 6 times (1 h ech) t room temperture. For immunostining of CD45, nerves were incuted in the fluorescence-lelled nti-cd45 (ctlog numer: BioLegend #13127 for 647 nm excittion nd ctlog numer: BioLegend #13121 for 488 nm excittion, dilution fctor: 1 2) in 1% donkey serum/1% TritonX-1/PBS for nother 48 h nd wshed with 1% TritonX-1/PBS for 6 times (1 h ech) t room temperture. High resolution imges were then cquired sed on LSM 88 confocl system t Hrvrd Center for Biologicl Imging nd our multiple chnnel SRS imging seprtely. The SRS lipid imging nd two-photon excited epifluorescence imging of MBP nd neurofilment stining re performed t the sme loction y locking different lsers. The MBP, which ws excited t 543 nm, ws imged y two-photon excittion of 1,63 nm lser only under SRS microscope. The neurofilment, which ws tgged t 488 nm, ws imged y two-photon excittion of pump lser t round 816 nm only under SRS microscope. SRS lipid imging ws otined when oth light ems were on the smple. The opticl power used on the smple ws out 9 mw of pump nd 8 mw of IR. The imging speed ws 1 s/frme, nd the imge size ws 512 y 512 pixels. Chemicl composition nlysis of lipid ovoids. The nerve smple ws imged t 35 wvenumers from 2,83 to 3,2 cm 1. The first nd lst spectrl point were not used for lter spectrl nlysis. Seven chemicls were used for spectrl retrievl. They re cholesterol(c), BSA, wter nd glctocereroside, sphingomyelin, phosphtidylcholine, phosphtidylethnolmine. The lest squre method ws used to retrieve the component of ech chemicl. We dded the concentrtion of phosphtidylethnolmine nd phosphtidylcholine ecuse they hve very similr spectrum. The lipid ovoids were mnully selected nd myelin sheth ws selected from histogrm from pek to 5% in the histogrm. After point of interest ws identified, the men ws then clculted for ech chemicl component. We imged five loctions on mice nd WT mice. The error r ws the stndrd devition of men in processed chemicl content of five imging loctions. We used the one-til unpired student t-test to clculte the P vlue. The imging prmeter were the sme s those used for ex vivo imging experiment. The opticl power used on the smple ws out 9 mw of pump in the middle of scnning rnge, nd the power ws lower t oth ends of spectrl scnning rnge. The IR power ws round 8 mw. The imging speed ws 1 s per frme, which mens the imge for ech colour tkes 1 s to tke, nd the imge size ws 512 y 512 pixels. Quntifiction of SRS imges. The commercilized imge processing progrm ImgeJ ws used s computtionl tool for SRS imges quntifiction. The plug-in cvmtch_templte ws employed to clculte the correltion etween SRS imge nd templtes with the method clled correltion coefficient. The correltion pictures were then trnsformed to threshold picture with n empiriclly chosen threshold vlue for ech templte. The res with correltion coefficients higher thn the threshold vlue would e designted s ROIs, which were emphsized y the red colour. Then the function of Anlyze Prticles ws used to select the ROIs stisfying defined roundness nd size requirements. The circulrity ws ssigned to e.1 smller thn the circulrity otined y templte on the imge from where the templte hd een selected. The size of the ROI ws empiriclly selected, which ws normlly round 2% smller thn the size of originl templtes. The results of the Anlyze Prticles function for ech templte were sved to spredsheet, nd the totl numer ws quntified. If the positions of the red res counted y Anlyze Prticles function from two templtes overlpped, it would e counted for only once. 2 slices of z-stck imges were nlysed with this method, nd the totl numer of lipid ovoids ws clculted s the numer of lipid ovoids of this z-stck imge set. All scitic nerve SRS imges of ALS mouse models were quntified sed on the sme lgorithm nd templte. Different lipid ovoids templtes re used in humn ptients nd mouse model quntifictions. Quntifiction of intct nd degenerted xon numers were performed from the 3D reconstruction of the z-stck imges of scitic nerves. The principle is the sme s the one used for lipid ovoid counting. The difference ws tht we hd 14 templtes for intct myelinted nerve fiers nd 4 templtes for locked nerve fiers. The rtio of locked nerve fiers versus totl myelinted nerve fiers (intct þ locked) ws used to evlute the progression of the disese. We quntified the thickness of myelin sheth with longitudinl SRS imge of the myelin sheth. We normlly chose the imge which sectioned the center of mjority of myelin sheth. We utilized the plug-in tueness in FIJI to highlight the oundry of myelin sheth nd clirted the inside of myelin sheth nd outside s zero intensity. Then we used the line profile of the imge in the direction perpendiculr the myelin sheth to clculte the thickness of ech wll of myelin sheth. The thickness ws defined y the width t the ottom of ech myelin wll in the line profile. The position of line ws scnned through whole imge nd the thickness informtion of ll myelin sheth t ll loctions in the imge ws output to csv file for sttisticl nlysis. The quntifiction of humn ALS versus control smples ws principlly the sme s descried y the ex vivo nd in vivo SRS imging of ALS mouse models sed on ImgeJ. But there re severl differences. First of ll, the templtes were different from templte used in mouse model. The templtes for quntifying SRS imges of humn smple were selected from SRS imges of nerves from ALS ptients. They were tested to hve good distinguishing ility etween tissues from ALS ptients nd control. Second, there ws very ovious loss of nerve fiers in the ventrl nerves of ALS ptients. Therefore, we lso clculted the verge signl intensity of ech imge s quntifiction of verged lipid enrichment. The quntifiction results from lipid ovoids counting were normlized y this verged lipid signl intensity. So the finl quntifiction hd the unit of numer of lipid ovoids of defined verge lipid enrichment. Entire scitic nerve scn. The sme opticl pthwy s the ex vivo SRS imging ws employed for entire scitic nerve scn. Homemde softwre ws used to utomticlly move the stge nd tke SRS imges. The stge ws moved y step of 18 mm to imge the whole scitic nerve. This corresponded to 2% overlp etween neighor imges, which ws necessry for good stitching qulity. The dt processing initited with compensting the light fll-off t imge corners. The imges were then stitched together with the FIJI plug-in Grid/collection stitching 64. The opticl power used on the smple ws out 9 mw of pump nd 8 mw of IR. The imging speed ws 1 s per frme, nd the imge size ws 512 y 512 pixels. In vivo end-point SRS imging. Mice crrying nd ge-mtched non-trnsgenic littermtes were imged t the ge of 6, 7, 8, 12 nd 16 weeks postntl. The imged mice were nesthetized y 4% isoflurne (VetOne, ctlog numer: 5217) under n ir flowing pprtus nd kept live during the entire imging process. A smll incision ws mde on the right leg of the imged mouse to expose the scitic nerve without removing ny muscle. The mouse ws mounted on specilly mnufctured mouse rck (Fig. 3), nd the scitic nerve ws gently covered nd protected y coverslip on metl holder (Fig. 3c). The in vivo imging setup ws similr to the ex vivo SRS imging setup descried ove except tht the ck scttered light ws collected. To effectively collect the reflected light, we instlled qurter wve plte nd polrized em splitter right efore the microscope ojective, so tht the reflected lsers fter pssing the qurter wve plte second time were switched to polriztion orthogonl to the input light nd reflected y the polrized em splitter. The light collection in the epi-mode ws less efficient compred with trnsmission mode. Therefore, higher lser power ws employed for in vivo experiment. To e specific, the opticl power used on the smple ws out 12 mw of pump nd 8 mw of IR. Mice in this section would e euthnized immeditely fter imging, thus termed s in vivo end-point SRS imging. Imging speed ws 1 s per frme, nd the imge size ws 512 y 512 pixels. Long-term seril in vivo SRS imging. Mice crrying nd ge-mtched non-trnsgenic littermtes were imged using the in vivo end-point SRS imging system descried previously. However, ll the imged mice would e wound clipped on their incision immeditely fter imging. On Dy 7 fter ny surgery, the wound clips would e removed crefully during mouse nesthetiztion. Long-term seril in vivo SRS imging ws initited t the ge of 5 weeks postntl, with susequent in vivo SRS imging on the sme mouse performed every 3 weeks until 11 weeks postntl. Further ehviourl monitoring nd stging s descried previously were performed to evlute the dignostic cpility of SRS imging. The opticl power used on the smple ws out 12 mw of pump nd 8 mw of IR. The imging speed ws 1 s per frme, nd the imge size ws 512 y 512 pixels. EMG. To minimize electricl rtifcts y the wound clips pplied to the mice during SRS imging, we conducted EMG first nd sujected the mice for recovery of 12 h efore in vivo SRS imging. Genotyped 5-week old mice were nesthetized y 1.25% (v/v) vertin (Sigm T4842). Animls were plced immeditely on heting pd to mintin their core temperture t 37 C. Mesurements were performed using Crdinl Synergy EMG mchine. A ground self-dhesive gelled surfce electrode ws plced over the til. Potentils were recorded from severl sites of the hind-lim muscles with concentric needle electrode (3 G) using gin of 5 mv per division nd nd pss filter with low nd high cut-off frequency settings of 2 nd 1, Hz, respectively. The entire recording process took 15 2 min per mouse 46. Minocycline tretment nd seril in vivo SRS imging of mouse model of ALS. To minimize the vrition of trnsgene copy numer, we purchsed 3 mice t the ge of 4 weeks from JAX Lortory (No. 2726), where oth low copy numer nd high copy numer nimls hve een removed. These nimls were supposed to crry 21 3 copies trnsgene. An equl numer of mice (N ¼ 15) were rndomly sujected to NATURE COMMUNICATIONS 7:13283 DOI: 1.138/ncomms13283 www.nture.com/nturecommunictions 13