Involvement of DNA methylation in memory processing in the honey bee Gabrielle A. Lockett, Paul Helliwell and Ryszard Maleszka

Transcription

1 812 Learning and memory Involvement of DNA methylation in memory processing in the honey bee Gabrielle A. Lockett, Paul Helliwell and Ryszard Maleszka DNA methylation, an important and evolutionarily conserved epigenetic mechanism, is implicated in learning and memory processes in vertebrates, but its role in behaviour in invertebrates is unknown. We examined the role of DNA methylation in memory in the honey bee using an appetitive Pavlovian olfactory discrimination task, and by assessing theexpressionofdna methyltransferase3, akeydriver of epigenetic reprogramming. Here we report that DNA methyltransferase inhibition reduces acquisition retention and alters the extinction depending on treatment time, and DNA methyltransferase3 is upregulated after training. Our findings add to the understanding of epigenetic mechanisms in learning and memory, extending known roles of DNA methylation to appetitive and extinction memory, and for the first time implicate DNA methylation in memory in invertebrates. NeuroReport 21: c 21 Wolters Kluwer Health Lippincott Williams & Wilkins. NeuroReport 21, 21: Keywords: Apis mellifera, associative learning, DNA methylation, epigenetics, extinction, memory Research School of Biology, The Australian National University, Canberra, Australia Correspondence to Professor Ryszard Maleszka, Research School of Biology, The Australian National University, Canberra 2, Australia Tel: ; fax: ; maleszka@rsbs.anu.edu.au Received 4 May 21 accepted 6 June 21 Introduction Long-term memories are widely thought to be underpinned by enduring structural changes in complex neuronal circuitry, and can persist for decades despite molecular turnover. Although the molecular mechanisms underlying these persistent changes remain to be fully elucidated, important contributions are made by epigenetic mechanisms: processes such as post-translational histone modifications and DNA methylation that have the capacity to alter transcription but not the DNA sequence itself. Epigenetic mechanisms have been extensively implicated in learning and long-term memory processing in vertebrates: epigenetic marks and their key mediating enzymes are affected by fear conditioning [1 5], a dearth of these enzymes, whether by mutation or inhibition, affects learning and memory [1 4,6 1], and epigenetic processes have been implicated in synaptic plasticity, a likely cellular correlate of learning and memory [2,1,11]. The honey bee possesses complete and functional DNA methylation enzymology similar to that found in vertebrates [12,13]. DNA methylation has been co-opted for a multitude of functions throughout evolution, and in the honey bee it appears to modulate the expression of a set of ubiquitously expressed genes [14]. Its complete DNA methylation system and complex, sophisticated behavioural repertoire make the honey bee an excellent candidate in which to assess the role of DNA methylation in learning and memory in invertebrates. We examined the effects of the potent DNA methyltransferase (DNMT) inhibitor (DNMTi) zebularine on learning and memory, and training-associated DNA methyltransferase3 (Dnmt3) transcription changes in the mushroom bodies (MBs): highly plastic signal processing centres essential for olfactory learning and memory processing in insects [15]. The honey bee is a well-characterized model animal for learning and memory; assays typically use the Pavlovian olfactory conditioning of the proboscis extension response (PER) [16 18]. We assessed the role of DNA methylation in both acquisition and extinction memory processing. Extinction is the reduction in a conditioned response when the associated cue is presented repeatedly without reinforcement. The retrieval of previously consolidated memories reactivates them and initiates both the reconsolidation of the original memory and the formation of a new context-dependent extinction memory. The trace dominance hypothesis posits that these two coexist and compete, with the dominant process subsequently determining behaviour [19]. Histone acetylation is known to be involved in the extinction of memory [5,7,8], though the role of DNA methylation has not been assessed. The ability to reduce conditioned responses when they are no longer needed is an evolutionarily critical part of learning, so a complete understanding of the basis for extinction is essential if we are to understand the molecular mechanisms underlying learning and memory, and the nature of human disorders of extinction such as phobias and posttraumatic stress disorder. Materials and methods Proboscis extension response assays We assessed learning and memory using a PER olfactory association paradigm in which bees are required to discriminate between rewarding and aversive odours [18]. For each experiment three to five independent biological replicates were performed as described [18]. Briefly, bees were collected and mounted at 6 days old then fed at the end of the day. At 7 days bees were given c 21 Wolters Kluwer Health Lippincott Williams & Wilkins DOI: 1.197/WNR.b13e32833ce5be

2 Memory in honey bees requires DNA methylation Lockett et al. 813 acquisition training and again fed at the end of the day. Exactly 24 h after acquisition training bees were given extinction training: five conditioned stimulus (CS)-only trials, 2 min apart. Five hours after extinction training bees were tested once more (CS-only) for the correct CSunconditioned stimulus (US) association. In each experiment bees were treated topically on the thorax once (24 h before acquisition training, 1 h before acquisition training, 1 min after acquisition training, 1 h after acquisition training or 2 h before extinction training) with 1 ml of 2 mm zebularine (Tocris Bioscience, Bristol, UK) in dimethylformamide; bees treated with 1 ml dimethylformamide served as vehicle-only controls. Topical thoracic treatment is an efficient method for delivery to the brain [2]. Quantitative real-time PCR Three independent replicates of this experiment were performed. Bees were obtained as described above and at 7 days divided into two groups: the first ( naïve ) were immediately snap frozen, the second were given acquisition training as described above. All bees not giving the correct response in the last training trial were discarded. Thirty minutes after training further bees were snap frozen (the trained sample). The remaining bees were fed at the end of the day, and 24 h after training tested for their retention of the CS US association. We used this test to verify a typical learning rate (5 8%) for each replicate. For each sample 1 brains (pooled) were dissected dividing the central protocerebrum from the remainder of the brain. The central protocerebrum is heavily predominated by the MBs so we describe this dissected piece as the MBs, though it does include other regions. Total RNAs were extracted and reverse-transcribed into cdna as detailed earlier [21]. Quantitative real-time PCRs were performed in 2 ml volumesusingarotorgene3 cycler (Corbett Research, Qiagen, Doncaster, Australia) as described [21], but with a 621C annealing temperature. Regions were amplified from each cdna in triplicate (each with 3 4 technical replicates) for each biological replicate. Amplified regions span exon boundaries ensuring amplification of cdna only. Primer sequences are Dnmt3_f: CAGCGATGACCTGCGATCGGCGATA, Dnmt3_r: TACAGGG TTTATATCGTTCCGAAC, ribosomal protein S8: Rps8_f: ACGAGGTGCGAAACTGACTGA, and Rps8_r: GCACTGTC CAGGTCTACTCGA. Takeoff and amplification efficiency values determined by RotorGene software (version 6.1 build 81; Corbett Research Qiagen) were used to calculate relative expression ratios (R) and ratio standard error values [21]. Results DNMT inhibition reduces the acquisition memory retention We examined the requirement for DNMT function in learning and memory by administering DNMTi treatment at various times relative to a learning task, using the PER assay. All DNMTi treatment times tested produce a decrease in acquisition retention (Fig. 1a). This is not statistically significant for any single treatment time, however, the analysis across treatment times indicates a highly significant effect of DNMTi on acquisition retention (P =.8, paired Student s t-test, 1-tailed). DNMTi treatment has no significant effect on learning during acquisition training ( within-session acquisition ; Fig. 1b and c). Within-session extinction and extinction retention require DNMT function at different times We examined within-session extinction (performance during extinction training) and extinction retention in vehicle-treated and DNMTi-treated bees, at a variety of treatment times. Our results show that DNMTi can affect both formation of the extinction memory during extinction training and extinction retention, depending on the treatment time (Fig. 2). Within-session extinction is altered by DNMTi treatment 24 h before or 1 min after acquisition training, or 2 h before extinction training (Fig. 2a c), whereas extinction retention is reduced by treatment 1 h before or 1 h after acquisition training (Fig. 2d and e). These differences are significant within each treatment time (w 2 test). Bidirectional modulation of within-session extinction depends on when DNMT is inhibited We show that DNMTi at three treatment times modulates within-session extinction, increasing or decreasing PER relative to controls depending on treatment time. When DNMTi treatment is given 24 h before or 1 min after acquisition training within-session extinction is increased relative to controls (Fig. 2a and b), conversely treatment 2 h before extinction training dramatically reduces within-session extinction compared to controls (Fig. 2c). Dnmt3 is upregulated in a learning-associated area of the brain after training We assessed levels of Dnmt3 mrna 3 min after training compared to levels in naïve bees, in both the MBs and the remainder of the brain. Training increases levels of Dnmt3 mrna in the MBs, whereas no significant change occurs in the rest of the brain (Fig. 3). Discussion Our behavioural assays reveal a consistent reduction in acquisition retention with DNMTi (Fig. 1a), significant across treatment times. Memory formation is not significantly affected by DNMTi (Fig. 1b and c) showing a role for DNA methylation in acquisition memory consolidation but not formation. Our results in the honey bee are consistent with those in the rat [1,4], revealing DNA methylation to be a phylogenetically conserved

3 814 NeuroReport 21, Vol 21 No 12 Fig. 1 (a) 1 (b) 1 (c) 1 9 zeb veh 9 zeb veh 9 zeb veh Treatment time A1 A2 A3 A1 A2 A3 PER% Effects of DNA methyltransferase inhibitor (DNMTi) on acquisition retention and within-session acquisition. (a) Retention 24 h after acquisition training is consistently reduced by the DNMT inhibitor zebularine (zeb) relative to vehicle controls (veh) at all treatment times. This is not statistically significant at any single treatment time (w 2 test), though analysis across treatment times reveals a highly significant effect of DNMTi (P =.8, paired Student s t-test). From left (n veh, n zeb ; P from w 2 test): treatment 24 h before training (n = 66, 69; P =.39), 1 h before training (n = 86, 75; P =.14), 1 min after training (n = 55, 71; P =.2), 1 h after training (n = 59, 63; P =.9), and 2 h before testing (n = 55, 68; P =.12). Acquisition did not differ during acquisition training (A1 A3) when treatment was (b) 24 h before acquisition training (n veh = 87 94, n zeb = 9 95; P =.49,.25,.35; w 2 test), or (c) 1 h before acquisition training (n veh = 86 1, n zeb = ; P =.93,.36,.81; w 2 test). Acquisition PER% expresses the number of bees giving the correct response out of bees giving any response. Fig. 2 (a) 1 (b) (c) (d) (e) PER% zeb (44) veh (49) zeb (4) veh (37) zeb (38) veh (37) zeb (33) veh (51) zeb (27) veh (39) 5 Test Test Test Test Test Effects of DNA methyltransferase inhibitor (DNMTi) on within-session extinction and extinction retention. Within-session extinction: treatment with the DNMTi zebularine (a) 24 h before acquisition training or (b) 1 min after acquisition training, or (c) 2 h before extinction training significantly alters proboscis extension response (PER)% in at least one extinction training trial (labelled 1 5; w 2 test: *P <.5, **P <.1). Extinction retention: DNMTi treatment 1 h before acquisition training (d) or 1 h after acquisition training (e) significantly reduced retention of the extinction memory, tested 5 h after extinction training ( test ; w 2 test: *P <.5, **P <.1). Number of bees tested in each group is indicated in the legend. Extinction PER% expresses the number of bees giving the conditioned response out of all bees who gave the conditioned response 24 h after acquisition training (scale identical between panels). mechanism in memory consolidation. Together with the implication of Dnmt3 in memory processing, these data indicate for the first time a role for DNA methylation in memory in an invertebrate, similar to findings in vertebrate models for learning and memory. Insects without vertebratelike DNA methylation such as Drosophila melanogaster are typically short-lived and so unlike the honey bee have no need for long-term memory lasting several months: it may be that DNA methylation is required for or facilitates dramatically increased long-term learning and memory capabilities. The present experiments provide the first assessment of DNA methylation involvement in extinction memory processing in any paradigm. We show a considerable requirement for DNMT function in extinction memory (Fig. 2) which is consistent with the established role for histone acetylation in extinction in rodent fear conditioning [5,7,9] and cocaine-induced conditioned place preference [8] paradigms, given the interplay of different epigenetic mechanisms in memory [6]. Our results show that epigenetic control of extinction extends to olfactory associative appetitive memories. Specifically we find

4 Memory in honey bees requires DNA methylation Lockett et al. 815 Fig. 3 R DNA methyltransferase3 (Dnmt3) expression in the brain after training. Training increases the amount of Dnmt3 mrna in the mushroom bodies (left) but not the remainder of the brain (right). R indicates relative mrna level in the trained sample relative to the naïve sample, normalized against any change in the reference gene (Rps8). Error bars indicate ratio standard error. within-session extinction and extinction retention to have different temporal requirements for DNMT function, in line with these two distinct processes corresponding to short-term and long-term extinction memory. Extinction retention can be made DNMT-dependent by inhibiting DNMTs around acquisition training time, without affecting the ability to learn extinction in the interim (Fig. 2d and e). This implies that the alteration of the acquisition memory by DNMTi changes the molecular mechanisms to be involved in future extinction retention. In the crab for example, strong but not weak training recruits histone acetylation [22]. Furthermore, given the critical recent discovery that an increased US duration during acquisition training makes extinction retention protein synthesis-dependent [23], we hypothesize that specific qualities of the acquisition memory set wheels in motion which determine the input of various molecular mechanisms in future extinction retention. We show DNMTi affects extinction memory (Fig. 2) more strongly than acquisition (Fig. 1a). Acquisition and extinction require some of the same cellular and molecular mechanisms, though they each also engage unique processes [24]. In the honey bee extinction but not acquisition is reduced by sleep deprivation [25], furthermore the acquisition and extinction of fear each induce different H3 and H4 histone acetylation patterns in the mouse [5]. Our results suggest DNA methylation too may be involved differently in acquisition and extinction memory. Where acquisition involves only a single memory, the presence of two competing memories in extinction may introduce an extra level of complexity, altering the nature of DNA methylation involvement. Our results (Fig. 2) are the first to show that DNA methylation affects extinction; furthermore as epigenetic treatments have only been reported to have monodirectional or no effect on within-session extinction [5,7,9], our demonstration of bidirectional epigenetic modulation of within-session extinction is also unique. Our interpretation of the bidirectional modulation is that the two treatment times resulting in increased within-session extinction weaken the acquisition memory, thus when the acquisition and extinction memories compete the extinction memory will predominate. Conversely, the treatment time decreasing within-session extinction would be explained by a weakened extinction memory allowing the acquisition memory to predominate. This hypothesis is in agreement with the trace dominance theory [19]. If competition between acquisition and extinction traces modulates within-session extinction, then it follows that our within-session extinction data provide further evidence of the effect on acquisition memory processing: the acquisition memory is still formed and still consolidated, albeit in a qualitatively weaker form less able to compete with the extinction memory during withinsession extinction. Furthermore, that treatment long before acquisition training (Fig. 2a) weakens the acquisition trace (manifested as altered within-session extinction) in a similar manner to treatment 1 min after training (Fig. 2b) further supports our conclusion that DNMTi specifically impedes consolidation. We also found consistent upregulation of the de novo Dnmt3 after training, in the MBs only (Fig. 3). Our findings strongly resemble those in vertebrates: transcription of de novo DNMTs is increased in a learning-associated brain area after fear conditioning [1]. Upregulation of Dnmt3 suggests the honey bee may exhibit altered methylation at specific CpGs in response to training, as has been shown in the rat [3,4]. Our results imply that memory formation in the honey bee, as in vertebrates, increases de novo DNMT requirements. In conclusion, this study uncovers a critical role for DNA methylation in learning and memory in the honey bee, analogous to findings in vertebrates. We find increased expression of Dnmt3 in a learning-associated area of the brain after training, and behaviourally show that DNMT inhibition affects both acquisition and extinction memory processing. We also find within-session extinction rates to be bidirectionally modulated, consistent with the trace dominance hypothesis. Further research needs to address whether the role of DNA methylation in extinction extends to other learning paradigms and to vertebrate models of learning and memory.

5 816 NeuroReport 21, Vol 21 No 12 Acknowledgements The authors thank Joanna Maleszka for assistance with and training in brain dissection, and Fiona Wilkes for assistance in editing the manuscript. This research was supported by the NHMRC (grant ). References 1 Miller CA, Sweatt JD. Covalent modification of DNA regulates memory formation. Neuron 27; 53: Fischer A, Sananbenesi F, Wang X, Dobbin M, Tsai L. Recovery of learning and memory is associated with chromatin remodelling. Nature 27; 447: Gupta S, Kim SY, Artis S, Molfese DL, Schumacher A, Sweatt JD, et al. Histone methylation regulates memory formation. J Neurosci 21; 3: Lubin FD, Roth TL, Sweatt JD. Epigenetic regulation of bdnf gene transcription in the consolidation of fear memory. J Neurosci 28; 28: Bredy TW, Wu H, Crego C, Zellhoefer J, Sun YE, Barad M. Histone modifications around individual BDNF gene promoters in prefrontal cortex are associated with extinction of conditioned fear. Learn Mem 27; 14: Miller CA, Campbell SL, Sweatt JD. DNA methylation and histone acetylation work in concert to regulate memory formation and synaptic plasticity. Neurobiol Learn Mem 28; 89: Bredy TW, Barad M. The histone deacetylase inhibitor valproic acid enhances acquisition, extinction, and reconsolidation of conditioned fear. Learn Mem 28; 15: Malvaez M, Sanchis-Segura C, Vo D, Lattal KM, Wood MA. Modulation of chromatin modification facilitates extinction of cocaine-induced conditioned place preference. Biol Psychiatry 21; 67: Lattal KM, Barrett RM, Wood MA. Systemic or intrahippocampal delivery of histone deacetylase inhibitors facilitates fear extinction. Behav Neurosci 27; 121: Feng J, Zhou Y, Campbell SL, Le T, Li E, Sweatt JD, et al. Dnmt1 and Dnmt3a maintain DNA methylation and regulate synaptic function in adult forebrain neurons. Nat Neurosci 21; 13: Levenson JM, Roth TL, Lubin FD, Miller CA, Huang I. Evidence that DNA (cytosine-5) methyltransferase regulates synaptic plasticity in the hippocampus. J Biol Chem 26; 281: Wang Y, Jorda M, Jones PL, Maleszka R, Ling X, Robertson HM, et al. Functional CpG methylation system in a social insect. Science 26; 314: The Honeybee Genome Sequencing Consortium. Insights into social insects from the genome of the honeybee Apis mellifera. Nature 26; 444: Foret S, Kucharski R, Pittelkow Y, Lockett GA, Maleszka R. Epigenetic regulation of the honey bee transcriptome: unravelling the nature of methylated genes. BMC Genomics 29; 1: Fahrbach SE. Structure of the mushroom bodies of the insect brain. Annu Rev Entomol 26; 51: Kuwabara M. Bildung des bedingten Reflexes von Pavlovs Typus bei der Honigbiene, Apis mellifica. J Fac Sci Hokkaido Univ Ser VI Zool 1957; 13: Bitterman ME, Menzel R, Fietz A, Schäfer S. Classical conditioning of proboscis extension in honeybees (Apis mellifera). J Comp Psychol 1983; 97: Maleszka R, Helliwell P, Kucharski R. Pharmacological interference with glutamate re-uptake impairs long-term memory in the honeybee, Apis mellifera. Behav Brain Res 2; 115: Nader K. Memory traces unbound. Trends Neurosci 23; 26: Barron AB, Maleszka J, Vander Meer RK, Robinson GE, Maleszka R. Comparing injection, feeding and topical application methods for treatment of honeybees with octopamine. J Insect Physiol 27; 53: Kucharski R, Maleszka R. Microarray and real-time PCR analyses of gene expression in the honeybee brain following caffeine treatment. J Mol Neurosci 25; 27: Federman N, Fustiñana MS, Romano A. Histone acetylation is recruited in consolidation as a molecular feature of stronger memories. Learn Mem 29; 16: Stollhoff N, Eisenhardt D. Consolidation of an extinction memory depends on the unconditioned stimulus magnitude previously experienced during training. J Neurosci 29; 29: Lin C-H, Yeh S-H, Lu H-Y, Gean P-W. The similarities and diversities of signal pathways leading to consolidation of conditioning and consolidation of extinction of fear memory. J Neurosci 23; 23: Hussaini SA, Bogusch L, Landgraf T, Menzel R. Sleep deprivation affects extinction but not acquisition memory in honeybees. Learn Mem 29; 16: