Reversible DNA methylation regulates seasonalphotoperiodic time measurementTyler J. Stevensona,1 and Brian J. Prendergasta,b

aInstitute for Mind and Biology and bDepartment of Psychology, University of Chicago, Chicago, IL 60637

Edited by Bruce S. McEwen, The Rockefeller University, New York, NY, and approved September 5, 2013 (received for review June 5, 2013)

In seasonally breeding vertebrates, changes in day length inducecategorically distinct behavioral and reproductive phenotypes viathyroid hormone-dependent mechanisms. Winter photoperiodsinhibit reproductive neuroendocrine function but cannot sustainthis inhibition beyond 6 mo, ensuring vernal reproductive re-crudescence. This genomic plasticity suggests a role for epigeneticsin the establishment of seasonal reproductive phenotypes. Here,we report that DNA methylation of the proximal promoter for thetype III deiodinase (dio3) gene in the hamster hypothalamus isreversible and critical for photoperiodic time measurement. Shortphotoperiods and winter-like melatonin inhibited hypothalamicDNA methyltransferase expression and reduced dio3 promoterDNA methylation, which up-regulated dio3 expression and inducedgonadal regression. Hypermethylation attenuated reproductiveresponses to short photoperiods. Vernal refractoriness to shortphotoperiods reestablished summer-like methylation of the dio3promoter, dio3 expression, and reproductive competence, reveal-ing a dynamic and reversible mechanism of DNA methylation inthe mammalian brain that plays a central role in physiological ori-entation in time.

biological rhythms | photoperiodism

Precise seasonal timing of reproduction is ubiquitous in nature,and seasonal time measurement demands behavioral, neural,

hormonal, and genomic plasticity. The annual change in daylength (photoperiod) is the major proximate environmental cuefor accurate seasonal timing of reproduction. In many vertebrates,changes in photoperiod drive the development of distinct seasonalreproductive phenotypes (e.g., refs. 1–3). The light/dark cycleentrains an endogenous circadian rhythm in nocturnal pinealmelatonin (MEL) secretion that, in turn, regulates the re-productive neuroendocrine system (4, 5). In many long-daybreeding mammals, the decreasing day lengths of late summerand autumn yield longer duration MEL signals that induce go-nadal involution via signaling at thalamic, hypothalamic, andpituitary targets (6–8). In these species, short day (SD) reproduc-tive inhibition is constrained by an endogenous seasonal timingmechanism that renders the neuroendocrine system refractory toMEL after ∼20 wk of exposure to SD, triggering “spontaneous”reproductive development in anticipation of spring (9, 10). InSD breeding species, winter day lengths have the opposite effectand are inductive. The neural and genomic plasticity responsiblefor MEL-mediated seasonal time measurement and the devel-opment of refractoriness to MEL remain unspecified.Recent work has identified a critical role for hypothalamic

thyroid hormone (T4) signaling in the transduction of photope-riod information into the reproductive neuroendocrine system.Although thyroid secretion of the prohormone T4 does notchange seasonally (11), hypothalamic expression of deiodinaseenzymes that catabolize T4 into the receptor-active triiodothyronine(T3) [deiodinase type II (DIO2)] or the receptor-inactive en-antiomer [deiodinase type III (DIO3)] are strikingly regulatedby changes in photoperiod (12–16), providing a seasonal gat-ing mechanism for thyroid hormone receptor signaling. Win-ter photoperiods elevate dio3 expression, quench T3 signaling,and inhibit gonadotrophin secretion, whereas spring/summerphotoperiods elevate dio2 expression, enhance T3 signaling,

and stimulate gonadotrophin release (12–16). Pineal MEL isnecessary for the induction of dio3 mRNA expression by photo-period (15), and MEL-driven dio3 expression acts as a molecularswitch for the seasonal control of reproduction (17). Identifyinghow photoperiod and MEL control dio3 expression is funda-mental to an understanding of how time information is repre-sented in the CNS.Epigenetic mechanisms figure prominently in the control of

gene expression (18, 19), but whether epigenetic modificationsmediate seasonal changes in gene expression in the seasonaltiming system is currently unknown. DNA methylation consists ofthe addition of a methyl group at CpG dinucleotide residues in themammalian DNA template (18); high levels of DNA methylationnear promoter regions inhibit transcription (19). Methylation is animportant epigenetic mechanism in the control of behavior andphysiology (20–23), because it affords rapid (24) reversible (25)and cyclical (26) regulation of gene expression. DNA methyl-transferases (DNMTs) that mediate DNA methylation are cata-lyzed either de novo (DNMT3a and DNMT3b) or maintainmethylation of DNA hemimethylated sites (DNMT1) (19).The present experiments tested the hypothesis that DNA

methylation mediates effects of photoperiod and MEL on re-productive physiology by examining methylation of the dio3proximal promoter in the seasonally breeding Siberian hamster(Phodopus sungorus), a common animal model for investigationsof seasonal timekeeping. We examined whether photoperiod-and MEL-driven changes in reproduction and hypothalamic dio3expression are governed by methylation in the dio3 proximalpromoter, and whether neuroendocrine refractoriness in mid-winter is mediated by a reversal of DNA methylation patterns.

Significance

This work examined whether epigenetic mechanisms partici-pate in the regulation of seasonal reproduction. In long-day(summer) breeding hamsters, exposure to inhibitory winterphotoperiods, or winter-like patterns of melatonin, alteredDNA methyltransferase expression; decreased DNA methyla-tion in the proximal promoter region of deiodinase type III(dio3) in the hypothalamus; and, in turn, increased hypotha-lamic dio3 expression. Pharmacological blockade of photope-riod-driven demethylation attenuated reproductive responsesto winter photoperiods. Winter demethylation was reversed inanticipation of spring: spontaneous reproductive developmentwas accompanied by remethylation of the dio3 promoter anddecreases in dio3 mRNA. Methylation dynamics in the adultbrain are reversible and may constitute an important compo-nent of the mechanism by which seasonal time is representedin the nervous system.

Author contributions: T.J.S. and B.J.P. designed research; T.J.S. performed research; T.J.S.and B.J.P. analyzed data; and T.J.S. and B.J.P. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.

Data deposition: The sequences reported in this paper have been deposited in the Gen-Bank database (accession nos. KC153107–KC153111, EU812319, and EU812320).1To whom correspondence should be addressed. E-mail: tyler.stevenson@abdn.ac.uk.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1310643110/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1310643110 PNAS | October 8, 2013 | vol. 110 | no. 41 | 16651–16656

PHYS

IOLO

GY

Dow

nloa

ded

by g

uest

on

Janu

ary

6, 2

020

Dow

nloa

ded

by g

uest

on

Janu

ary

6, 2

020

Dow

nloa

ded

by g

uest

on

Janu

ary

6, 2

020

Dow

nloa

ded

by g

uest

on

Janu

ary

6, 2

020

Dow

nloa

ded

by g

uest

on

Janu

ary

6, 2

020

The results identify that DNA methylation in the adult brain isreversible, photoperiodically regulated, and plays a central rolein neuroendocrine genomic plasticity.

ResultsIsolation and Characterization of the dio3 Proximal Promoter. Toevaluate DNA methylation in the dio3 transcriptional regulatoryregion, we sequenced the Siberian hamster dio3 proximal pro-moter (Fig. S1), defined here as the 916 bp preceding the dio3start codon. The hamster dio3 promoter exhibited a high pro-portion of guanine and cytosine nucleotides (34% and 32%,respectively) and contained a total of 63 CpG sites. This CpGfrequency ([Observed/Expected] × [Promoter Sequence Length] =0.625) is considered high, and exceeded the threshold for a “CpGisland” (27). Adenine and thymine comprised only 18% and 16%,respectively, of the sequenced promoter region. Putative tran-scription binding sites in the promoter region were statisticallyevaluated (Fig. S2 and Table S1).

Short Photoperiods Decrease dio3 Proximal Promoter Methylation.To examine whether changes in day length alter DNA methyl-ation in the dio3 proximal promoter, male Siberian hamsterswere exposed to SD photoperiods, which induced gonadal re-gression; control hamsters remained in long day (LD) photoperiods(Fig. 1A). As expected, hypothalamic dio3 mRNA expression wassignificantly increased in SD relative to LD (t = 2.09, P < 0.05; Fig.1B), absent any changes in dio2 (t = 1.16, P = 0.26). The increase indio3 mRNA is consistent with numerous reports in this species andresults in marked decreases in hypothalamic T3 concentrationsunder SD (15).To assess whether changes in hypothalamic dnmt expression

participate in the photoperiodic stimulation of dio3 mRNA inSD, we measured expression of mRNAs for dnmt1, dnmt3a, and

dnmt3b. SD treatments sufficient to up-regulate dio3 mRNAsignificantly decreased the expression of dnmt3b (t = 2.12, P <0.005; Fig. 1C) and dnmt1 mRNA (t = 2.12, P < 0.05; Fig. 1C).Indeed, dnmt3b expression and dnmt1 expression were eachapproximately fourfold higher in LD hamsters than in SDhamsters, suggesting the potential for increases in methylation atmultiple DNA residues and throughout the hypothalamus underLD photoperiods. Using immunocytochemical methods, we in-vestigated whether DNMT3b expression was present in hypotha-lamic regions that also express dio3. In LD hamsters, DNMT3bimmunoreactivity (DNMT3b-ir) was evident throughout the hypo-thalamus. Importantly, DNMT3b-ir was robust in the ependymalcell layer surrounding the third ventricle (Fig. 1D), a region ofthe hypothalamus in which MEL binds and dio3 mRNA isexpressed in SD photoperiods (12–16). Diffuse hypothalamicexpression of DNMT3b in LD suggests that increases in methyla-tion may be occurring in multiple hypothalamic nuclei; however,photoperiodic differences in DNMT3b-ir were not quantified in thepresent work.To determine if omnibus increases in dnmt3b expression in LD

are associated with increased methylation of the dio3 promoter,the amount of methylated CpGs in the dio3 proximal promoterwas quantified using a methylation-sensitive restriction enzyme(MSRE) assay [digestion with restriction enzymes targeting CpGsites (restriction endonucleases BstUl and Hpall), followed byamplification with primers specific to the dio3 proximal pro-moter]. SD significantly decreased dio3 proximal promoter DNAmethylation: methylation levels were two- to threefold higher inLD relative to SD (t = 2.14, P < 0.05; Fig. 1E). Next, to confirmthe MSRE assay independently and to identify specific CpGresidues that are methylated in response to changes in daylength, hypothalamic DNA from LD (n = 7) and SD (n = 7)hamsters was treated with sodium bisulfite, and a region of DNA(280 bp, located within the MSRE amplicon) was amplified,

Week0 4 8

mc(e

mulovsits eT

3 )

0.0

0.1

0.2

0.3

0.4

0.5

LD SD

*** ***

A

dio2 dio3

AN

Rm

evital eR

0

2

4

6

8

10

12LDSD

*

B

dnmt1 dnmt3a dnmt3b

AN

Rm

evit aleR

0.0

0.5

1.0

1.5

2.0

2.5

3.0

*

C

*

LD SD

noitalyhtemlabol

G%

0

20

40

60

80

*

E

D

F

EC

III

LDSD

LDSD

0255075

100

11 10 9 8 7 6 5 4 3 2 1

SDLD

% C

pG M

ethy

latio

n

dio3 promoter CpG site

-490 -210

Fig. 1. Short photoperiods inhibit reproduction and activate hypothalamic mRNA expression via epigenetic mechanisms. Acute transfer from LD to SDphotoperiods caused gonadal regression (A), increased hypothalamic dio3 mRNA expression (B), and decreased hypothalamic dnmt1 and dnmt3b mRNAexpression (C) after 8 wk. (D) Immunocytochemical localization of DNMT3b (DNMT3b-ir) in the hamster mediobasal hypothalamus (MBH). DNMT3b-ir wasevident throughout the MBH and in the ependymal cell (EC) layer along the third ventricle (III). (E) Transfer from LD to SD reduced DNA methylation in thedio3 proximal promoter, as measured using an MSRE assay. (F) Proportion of LD and SD hamsters in which methylation was present in each of 11 CpG sites inthe dio3 proximal promoter, as measured by direct sequencing of sodium bisulfite-treated DNA. The abscissa (not to scale) depicts the 11 CpG sites in the −490to −210-bp region of the dio3 proximal promoter. All data are mean ± SEM. *P < 0.05; ***P < 0.005 vs. LD value.

16652 | www.pnas.org/cgi/doi/10.1073/pnas.1310643110 Stevenson and Prendergast

Dow

nloa

ded

by g

uest

on

Janu

ary

6, 2

020

sequenced, and analyzed. Three CpG sites in the proximal pro-moter (sites 5, 4, and 2) exhibited fewer methylation marks in SDrelative to LD, with a complete absence of methylation at sites 5and 4 in SD hamsters (Fig. 1F). Together, these data indicate thatexposure to short photoperiods is associated with reduced dnmtmRNA expression, decreases in the number of methyl groups onspecific motifs in the dio3 proximal promoter, and increases indio3 mRNA.

MEL Is Sufficient for Reducing dio3 Promoter DNA Methylation.Long-duration MEL signals are necessary and sufficient for re-productive neuroendocrine responses to winter day lengths (3).To test the hypothesis that long-duration MEL signals are suf-ficient to decrease methylation of the dio3 promoter, LD-housedhamsters were treated with afternoon injections of MEL (50 μgadministered s.c. 4 h before light offset), effectively mimickingthe expansion of nocturnal MEL duration in SD by summatingwith the endogenous MEL signal (28). MEL was administeredfor 2 consecutive days, or for 1 or 4 wk, which allowed specifi-cation of the temporal dynamics of MEL-mediated changes inhypothalamic gene expression and promoter methylation. MELtreatments for 1 or 4 wk initiated gonadal regression (P < 0.001),but 2 d of MEL did not (P > 0.05; Fig. 2A). MEL triggered short-latency decreases in dio2 expression (F = 5.55, P < 0.005; Fig. 2B),with maximal inhibition of dio2 expression evident after only 2 d(P < 0.005, all comparisons; Fig. 2B). Decreases in dio2 expressionfollowing MEL injections were not evident following ∼10 wk ofSD photoperiod treatments. The significance of this discrepancyis unclear and may reflect differences in time course or pharma-cological effects of relatively high doses of MEL. MEL graduallyup-regulated dio3 mRNA (F = 2.98, P < 0.05; Fig. 2B) over 4 wk(P < 0.005; Fig. 2B), establishing that long-duration MEL signalsare sufficient to increase hypothalamic dio3 expression.Because dnmt3b is implicated in de novo acquisition of DNA

methylation patterns at unmethylated CpG sites (19), we exam-ined whether MEL signals are sufficient to inhibit dnmt3b ex-pression in a manner consistent with the pattern of disinhibiteddio3 mRNA. Indeed, MEL-driven changes in hypothalamicdnmt3b expression preceded changes in dio3 expression (F =2.82, P < 0.05; Fig. 2B): Two consecutive days of MEL treatmentdid not alter dnmt3b expression, but 1 wk of daily MEL injectionreduced dnmt3b expression by ∼50% (P < 0.05); dnmt3b in-hibition was sustained by MEL thereafter (P < 0.05). Lastly,MEL also decreased DNA methylation in the dio3 proximalpromoter region (F = 2.84, P < 0.05; Fig. 2C), temporally mir-roring the inhibition of dnmt3b expression: dio3 promoter meth-ylation was reduced after 1 wk of MEL treatment (P < 0.05) andremained significantly lower thereafter (P < 0.05; Fig. 2C). Thesedata establish a temporal pattern in which MEL initially inhibits

dnmt3b expression and dio3 promoter DNA methylation (after∼1 wk), culminating in increased dio3 mRNA (after 4 wk).

Hypermethylation Attenuates Reproductive Responses to InhibitoryPhotoperiods. To determine if decreases in hypothalamic meth-ylation are necessary for reproductive responses to SD, the nextexperiment challenged hamsters with enhanced DNA methyla-tion [via chronic treatment with 3-aminobenzimide (3AB;40 μg/d administered s.c.), as described in SI Materials andMethods] during exposure to LD and SD. 3AB is an inhibitorof poly(ADP ribosyl)ation (PARP activity) and promotes methylbinding to DNA, leading to enhanced DNA methylation (29).3AB attenuated but did not abolish gonadal responses in SD (F =21.53, P < 0.001; Fig. S3A). 3AB-treated hamsters had largertestes relative to SD controls after 2 wk in SD (P < 0.001).Substantial gonadal regression occurred during the next 2 wk inboth SD groups, but testes of 3AB-treated hamsters remainedsignificantly larger than those of saline-injected controls after4 wk in SD (P < 0.05). Importantly, 3AB treatments did not directlystimulate gonadal growth in LD hamsters (F = 1.01, P = 0.37; Fig.S3A), indicating that 3AB-induced hypermethylation is not byitself gonadostimulatory. Because 3AB modifies methylationlevels via its inhibition of PARP, the present data cannot excludea contribution of reduced PARP activity in the effects of 3AB ongonadal regression. However, to establish that 3AB increasedhypothalamic DNA methylation, we measured 5-methyl-2′deoxy-cytidine, an indicator of methylated DNA, and confirmed that3AB treatments were effective in globally up-regulating hypo-thalamic DNA methylation compared with saline-treated SDhamsters (t = 2.07, P < 0.05; Fig. S3B). The 3AB injections likelyinduced omnibus methylation on numerous genes, and addi-tional convergent evidence is required to establish the suffi-ciency of hypermethylation to block photoperiod-induced gonadalresponses. Nevertheless, the data in Fig. S3 provide a preliminaryfunctional link between DNA methylation and the expression ofseasonal phenotypic responses.

Photorefractoriness Is Associated with Remethylation of the dio3Proximal Promoter. Hamster seasonal reproductive inhibitionrequires a direct response to SD, but restimulation of re-productive physiology in late winter/early spring is mediatedby an endogenous timing mechanism that renders the hypo-thalamus unresponsive (refractory) to inhibitory MEL signalsafter ∼5 mo (9, 10). Similar timing mechanisms are implicatedin most vertebrate seasonal rhythms and reflect an evolutionarilyconserved process for measuring seasonal time (30, 31). In lightof the central role of reduced dio3 promoter methylation in theinduction of the winter reproductive phenotype, we examinedwhether remethylation of the same region of DNA mediates

0

20

40

60

80

100C

* *

LD MEL2 days

MEL1 week

MEL4 weeks

% M

ethy

latio

n

Week

0.2

0.3

0.4

0.5

0.6

0.7

****

A

***

0 2 4

LD MEL 2 daysMEL 1 week MEL 4 weeks

Test

is v

olum

e (c

m3 )

0.0

0.5

1.0

1.5

2.0

2.5LD MEL 2 days MEL 1 week MEL 4 weeks

B

* *

*

*********

Rel

ativ

e m

RN

A

dio2 dio3 dnmt3b

Fig. 2. MEL mediates SD-induced epigenetic events. Hamsters housed in an LD photoperiod were treated with MEL (50 μg administered s.c.) in the lateafternoon (for 2 d, 1 wk, or 4 wk) to extend the endogenous plasma MEL profile. (A) MEL treatments mimicked effects of SD on the reproductive system. MELtreatments up-regulated hypothalamic dio3 mRNA expression and decreased hypothalamic dnmt3b mRNA expression (B) and decreased methylation of thedio3 proximal promoter (C). All data are mean ± SEM. *P < 0.05; ***P < 0.005 vs. LD value.

Stevenson and Prendergast PNAS | October 8, 2013 | vol. 110 | no. 41 | 16653

PHYS

IOLO

GY

Dow

nloa

ded

by g

uest

on

Janu

ary

6, 2

020

development of neuroendocrine photorefractoriness and gonadalrecrudescence. Male and female hamsters were exposed to SDfor 42 wk, and age-matched controls were housed in LD or SDfor 10 wk. Reproductive organ masses decreased after short-term (10 wk) exposure to SD (P < 0.001), but extended SDtreatment caused gonadal recrudescence, indicative of neuroen-docrine photorefractoriness to SD (9, 10) (Fig. 3A). Fur molt, amorphological trait that changes seasonally and reflects pituitarylactotroph activity, confirmed the SD-induced changes in physi-ology and subsequent development of photorefractoriness (χ2 =8.6, P < 0.005; not illustrated). Hypothalamic dio2 mRNA ex-pression was comparable across all groups (F = 0.33, P > 0.70;Fig. 3B), but dio3 expression varied markedly across states ofphotosensitivity (F = 7.88, P < 0.01; Fig. 3B). As expected, dio3mRNA was elevated following acute exposure to SD (P < 0.005),but after 42 wk in SD, dio3mRNA had returned to LD-like levels(P < 0.005 vs. SD, P > 0.80 vs. LD), indicating that SD-inducedincreases in dio3 expression are reversed in the photorefractoryreproductive neuroendocrine system.To assess whether photorefractoriness is accompanied by a re-

versal of the acute SD changes in dio3 methylation, we examineddnmt3b mRNA. Hypothalamic dnmt3b expression changed sig-nificantly over time in SD (F = 6.03, P < 0.005; Fig. 3C). Initially,SD inhibited dnmt3b expression (P < 0.01), but after 42 wk in SD,dnmt3b expression was disinhibited (10 wk vs. 42 wk: P < 0.005)comparable to that of LD hamsters (P > 0.90).Lastly, we quantified DNA methylation in the dio3 proximal

promoter region of photorefractory hamsters via MSRE assay.Hypothalamic dio3 promoter methylation varied markedly overtime in SD (F = 5.41, P < 0.01; Fig. 3D), with a clear pattern ofreduced methylation on week 10, followed by a complete reversaland remethylation by week 42 in SD. Sodium bisulfite sequencingof DNA from a subset of these hamsters [LD, n = 4; SD, n = 4;SD-refractory (SD-R), n = 6] confirmed the MSRE data (Fig. 3E).Of the three CpG sites that exhibited clear decreases in methylationfollowing acute exposure to SD (compare with Fig. 1F), one site(site 4) exhibited a complete reversal in DNAmethylation followingthe development of refractoriness. Thus, refractoriness-induceddisinhibition of dnmt3b and inhibition of dio3 mRNA expressionare accompanied by a reversal in the methylation of specific DNAsequences in the dio3 proximal promoter that may provide tran-scriptional control of dio3 expression.

DiscussionHere, we report a mechanism of photoperiod- and MEL-inducedchanges in methylation of the promoter region of a gene (dio3)whose expression is critical to the transduction of photoperiodinformation into the brain and reproductive neuroendocrinesystem. Exposure to SD or MEL broadly inhibited the expressionof DNMTs, which regulated the methylation status of the dio3promoter. Furthermore, a seasonally reversible cycle of changesin DNA methylation was identified in the dio3 promoter, be-cause abundant remethylation occurred during the developmentof neuroendocrine refractoriness to SD. This work adds to anemerging literature on reversible methylation as a mechanismunderlying physiological events in the CNS and regulating phe-notypic and behavioral changes (24–26, 32, 33). Discrete regionsof the dio3 promoter exhibit extremely slow changes in methyl-ation status, occurring over the course of weeks to months; thus,the present work specifies a temporal dimension over which re-versible epigenetic mechanisms operate to regulate systems-levelbiological processes. Methylation of the dio3 promoter likelyacts as a key step for the maintenance of reproductive compe-tence during the breeding season and permits changes in envi-ronmental photoperiod to communicate with the reproductiveneuroendocrine system.Photoperiodic plasticity in deiodinase mRNA expression is

a critical step in the neuroendocrine control of reproduction inbirds (13, 14, 34) and mammals (15, 16, 35); the present workidentifies DNA methylation as a key mechanism by which daylength and MEL exert seasonal control over dio3 expression.Long summer photoperiods drive high levels of dnmt3b expres-sion, which maintain dio3 promoter methylation and ensure lowlevels of dio3 expression; collectively, this facilitates hypothalamiccatabolism of the prohormone T4 into reproductively stimulatoryT3 (15, 36) (Fig. S4A). Seasonal decreases in day length causeexpansion of the duration of nocturnal MEL secretion, and thecumulative effect of successive long MEL signals is an acutedown-regulation of dnmt3b expression; reduced methylation ofthe dio3 promoter increases accessibility of the DNA template totranscriptional control, and dio3 expression is markedly in-creased, quenching hypothalamic T3 signaling (Fig. S4B). Theregulation of dio3 expression is likely complex, and methylationstatus presumably has an impact on the manner in which addi-tional signals, such as “tuberalins,” thyroid-stimulating hor-mone β, and gonadal steroids, drive or inhibit dio3 expression

Male Female

)gm(

ssam

setseT

0

200

400

600

800

1000

)gm(

ssa

msu

r et

U

0

50

100

150

200

250

300A

*** ***

dio2 dio3

AN

Rm

ev italeR

0

1

2

3

4

5

*

B

SD10 wks

SD-R42 wks

LD

evitaleR

b 3tmnd

AN

Rm

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

1.6

1.8

***

C

D

noitalyhtem

%

0

5

10

15

20

25

30

*

LDdio3 promoter CpG site

E

SD10 wks

SD-R42 wks

LD SD-10 wks SD-42 wks

LD SD-10 wks SD-42 wks

0255075

100

% C

pG M

ethy

latio

n

SDRSDLD

11 10 9 8 7 6 5 4 3 2 1

012-094-

Fig. 3. Neuroendocrine refractorinessto SD reverses patterns of DNA meth-ylation induced by acute SD exposure.(A) Acute (10 wk, SD) exposure to SDinduced gonadal regression, whereasprolonged exposure (42 wk, SD-R) trig-gered neuroendocrine refractorinessand gonadal recrudescence. Refrac-toriness in SD-R hamsters was char-acterized by a complete reversal ofhypothalamic dio3 and dnmt3bmRNAexpression (B and C) and by remethy-lation of DNA in the dio3 proximalpromoter (D). (E) Proportion of LD, SD,and SD-R hamsters in which methyl-ation was present in each of 11 CpGsites in the dio3 proximal promoter,as measured by sodium bisulfite DNAsequencing. The abscissa (not to scale)depicts the 11 CpG sites in the−490 to−210-bp region of the dio3 proximalpromoter. All data are mean ± SEM.*P < 0.05; ***P < 0.005 vs. LD value.

16654 | www.pnas.org/cgi/doi/10.1073/pnas.1310643110 Stevenson and Prendergast

Dow

nloa

ded

by g

uest

on

Janu

ary

6, 2

020

(37). Prolonged exposure to SD eventually culminates in photo-refractoriness, characterized by a complete reversal of dnmt3bexpression, remethylation of the dio3 proximal promoter (spe-cifically, site 4), and marked decreases in dio3 expression. Ab-sent hypothalamic dio3 expression, enhanced T3 signaling triggersvernal gonadal recrudescence in refractory hamsters (Fig. S4C).How short photoperiods revert from inhibiting to stimulating, ormerely disinhibiting, dnmt3b expression remains unknown, butthis is likely a key step in the induction of photorefractoriness.Reduced methylation of the dio3 promoter by shorter (10 wk)intervals of SD disinhibits dio3 expression but also grants accessto the promoter by transcription factors that cannot participate inthe regulation of dio3 at other times of year.Recent evidence suggests that hormone secretion can have

marked effects on the methylation status of specific promotersequences (38). In the adult brain, testosterone secretion main-tains methylation of CpG sites in the promoter regions of ste-roid-responsive genes (22) (e.g., vasopressin, estrogen receptorα), and during early development, gonadal steroids exert en-during effects on CNS sexual differentiation, in part, via epige-netic modifications (23, 39–41); however, whether such changesare plastic or reversible in adulthood remains unknown. Here,we demonstrate that DNMT expression and downstream dio3promoter methylation respond to the duration of elevated cir-culating MEL in a manner that mirrors reproductive responsesto MEL. Importantly, dnmt3b responses to long-duration MELlead gonadal responses by at least 1 wk, and thus are not merelyconsequences of changes in gonadal hormone secretion. Therelatively broad hypothalamic expression of DNMT3b-ir in LDsuggests that these enzymes may also be expressed in cells andnuclei that are components of CNS pathways other than thoseparticipating in the seasonal control of reproductive physiology.Epigenetic modifications are no longer viewed as being estab-

lished solely during early development and maintained across thelife span (42). Rather, several lines of evidence suggest that histoneacetylation of specific genes is a dynamic and reversible process.For example, the circadian clock gene PERIOD1 (per1) is undertranscriptional control by histone deacetylation, and repression ofper1 transcription is a rhythmic and reversible daily event (24). Inaddition, rhythmic and reversible deacetylation of histone H4 bythe histone deacetylase 3 (HDAC3)/nuclear corepressor 1 (Ncor1)complex is required for proper circadian clock function: mutationsof the Ncor1 deacetylation activation domain that block its asso-ciation with HDAC3 markedly shorten circadian period length anddysregulate the circadian expression clock genes bmal1 and rev-erbα(43). In honeybees, the reversible transition from a “nurser” toa “forager” behavioral phenotype is associated with genome-widechanges in patterns of methylation (25). The data reported hereprovide evidence for reversible DNA methylation of a specificgene, dio3, in the control of a reversible reproductive phenotype.By driving photoperiod- and refractoriness-induced changes in re-productive condition, the seasonal cycle in methylation describedhere has the potential to mediate widespread changes in physiologyand behavior. This mechanism provides flexibility to regulate bothautumnal inhibition and vernal recrudescence of reproductivephysiology. Hypothalamic deiodinase responses to photoperiodoccur in a diverse array of avian and mammalian species, sug-gesting that seasonal regulation of DNA methylation may be anevolutionarily ancient timing mechanism.The present work identifies molecular mechanisms that me-

diate effects of day length on reproduction. SD MEL signalsinitially inhibit dnmt3b and reduce dio3 promoter methylation,which eventually permits dio3 expression (Fig. S4); increases indio3 quench perihypothalamic T3 signaling. Reduced methyla-tion in SD may also permit access by other transcription factorsto the dio3 promoter. Refractoriness to SD is characterized byincreases in dnmt3b expression and remethylation of the dio3proximal promoter, suggesting that the seasonal loss of re-sponsiveness to SD MEL signals may occur either at the level ofor upstream of dnmt3b expression. Because T3 regulates therelease of gonadotropin-releasing hormone (GnRH) (13, 14),

epigenetic control of dio3 mRNA expression by dnmt3b likelyfunctions as a key step in the molecular cascade of events re-sponsible for gonadotrophin signaling from the brain to the pi-tuitary (15). The extent to which changes in T3 signaling have animpact on hypothalamic neuropeptide systems that converge inGnRH neurons has not been fully elaborated, but a recent reportindicates that in SD-housed hamsters, exogenous T3 induces anLD phenotype in kisspeptin and gonadotrophin-inhibitory hor-mone peptide levels in multiple hypothalamic nuclei (44), sug-gesting that hypothalamic T3 levels may modulate GnRH activityvia multiple parallel pathways.Taken together, the present data demonstrate that a cycle of

reduced methylation and remethylation of the dio3 promoter regionis driven by photoperiod- and MEL-dependent changes in dnmtexpression. Methylation of the dio3 promoter affords transcrip-tional control of dio3 mRNA, and spontaneous remethylation ofthe dio3 promoter in late winter drives gonadal recrudescence,acting as a component of an endogenous calendar for photoperi-odic time measurement.

Materials and MethodsAnimals.Male and female Siberian hamsters (P. sungorus) were selected froma colony maintained at the University of Chicago. Hamsters were housed inpolypropylene cages illuminated for 15 h per day [LD; lights off at 1700 hCentral Standard Time (CST)]. Food (Teklad; Harlan Laboratories) and fil-tered tap water were provided ad libitum. All procedures were approved bythe Animal Care and Use Committee at the University of Chicago.

Study 1: Effects of Photoperiod on dnmt and dio2/3 Expression, and dio3Promoter Methylation. To examine the effects of acute exposure to in-hibitory SD lengths (9 h light/day, lights off at 17:00 h CST), male hamsterswere housed in the colony LD photoperiod (n = 18) or in an SD photoperiod(n = 20) for 8 wk. Testis volumes (TVs) were determined before photoperiodmanipulations (baseline), and again 4 and 8 wk later, under light isofluraneanesthesia [4% (vol/vol) induction, 2% (vol/vol) maintenance] via measure-ment of the length (L) and width (W) of the left testis. TV measurementswere collected without knowledge of the animal’s treatment condition. TVswere calculated using the equation for a prolate spheroid (TV = 4/3 × [L/2] ×[W/2]2) (45). On week 8, hamsters were euthanized via rapid decapitationat the midpoint of the light phase, and testis weights were determined(±0.1 mg). At autopsy, the whole hypothalamus was rapidly dissected aspreviously described (46), frozen on dry ice, and stored at −80 °C. FollowingRNA and DNA extraction, mRNA expression and promoter methylation weredetermined as described below. DNA from a subset of male hamsters (LD,n = 7; SD, n = 7) was used for sodium bisulfite treatment and sequencing toidentify specific nucleotide residues upon which methylation occurs.

Study 2: Effects of MEL dnmt and dio2/3 Expression, and dio3 Promoter Methylation.Male hamsterswere housed in an LD photoperiod and received daily injections ofMEL (50 μg administered s.c., N-acetyl-5-methoxytryptamine; Sigma) or 0.1 mL ofethanolic saline vehicle (SAL) 4 h before lights off. This timing and concentrationof MEL reliably induce gonadal regression in this species (47). Hamsters receivedMEL injections for 2 d (n = 12), 1 wk (n = 16), or 4 wk (n = 15); SAL injectionswere administered for 4 wk. All injection regimens terminated on the samecalendar day, such that injection treatments for hamsters receiving MEL for 2 dand 1 wk began 26 and 21 d, respectively, after hamsters receivingMEL for 4 wk.TVs were measured on weeks 0 and 2 and 4 wk later. After the proscribednumber of days of MEL treatment, hypothalamic tissue was extracted on thefinal day of the experiment as described above.

Study 3: Effects of Photorefractoriness on dnmt and dio2/3 Expression, andPromoter Methylation. On week 0, male and female hamsters were housedin an LD photoperiod (n = 12 male, n = 11 female) or SD photoperiod (n = 21male, n = 24 female) for the next 42 wk; an age-matched cohort of hamsterswas housed in an LD photoperiod for 32 wk and was transferred to an SDphotoperiod for the last 10 wk of the experiment (n = 10 male, n = 11 fe-male). TVs of male hamsters and body masses and fur scores (1 = dark“summer” fur, 4 = white “winter” fur) of both sexes were determined atmultiple intervals between weeks 0 and 42 to document gonadal and so-matic responsiveness and refractoriness to SD photoperiods. Testes and uteriwere dissected and weighed at autopsy, and hypothalamic tissue was ex-tracted as described above.

Stevenson and Prendergast PNAS | October 8, 2013 | vol. 110 | no. 41 | 16655

PHYS

IOLO

GY

Dow

nloa

ded

by g

uest

on

Janu

ary

6, 2

020

RNA/DNA Isolation. DNA and RNA were extracted from whole hypothalamictissue using a QIAGEN RNeasy/DNeasy kit. Nucleic acid concentration and qualitywere determined by means of a spectrophotometer (Nanodrop; Thermo Scien-tific). cDNA was synthesized using SuperScript III (Invitrogen), and genomic DNAand cDNA were stored at −20 °C until quantitative PCR (qPCR) was performed.

DNA Sequencing and qPCR. Primers for the dio3 proximal promoter and dnmt1,dnmt3a, and dnmt3b were designed based on conserved regions of mouse, rat,and human sequences using PrimerExpress software (Life Technologies) andoptimized for use in Siberian hamsters (Table S2). All sequences were de-termined at the University of Chicago Comprehensive Cancer Center DNA se-quencing facility. A standard nucleotide BLAST was used to determine sequencespecificity. qPCR reactions were performed with 2 μL of cDNA using a Bio-RadCFX384 system.

MSRE Assay. Hypothalamic DNA from all hamsters was subjected to MSREanalyses. DNAmethylation at CpG nucleotides in the dio3 proximal promoterregion was quantified using the MSRE assay (22).

Sodium Bisulfite Conversion and DNA Sequencing. Bisulfite conversion wasconducted using EpiTect Bisulfite kits (QIAGEN). Ligation, transformation,growth, PCR, and sequencing procedures are described in SI Materialsand Methods.

Statistical Analyses. TVs were analyzed using ANOVA. mRNA expression anddio3 promoter methylation were assessed using ANOVAs and unpairedt tests where appropriate. Sex differences were not evident in any measureof gene expression or methylation (P > 0.10, all comparisons), and data fromboth sexes were combined to increase statistical power. When violationsof normality were observed, data were log-transformed. Differences wereconsidered significant if P < 0.05.

ACKNOWLEDGMENTS.We thank Irv Zucker for comments on a manuscriptdraft, and Kenneth Onishi and Betty Theriault for expert technicalassistance. This project was supported by the National Center for ResearchResources and the National Center for Advancing Translational Sciences ofthe National Institutes of Health (NIH) (Grant UL1 RR024999). This workwas funded by NIH/National Institute of Allergy and Infectious DiseasesGrant AI-67406.

1. Stevenson TJ, Ball GF (2011) Information theory and the neuropeptidergic regulationof seasonal reproduction in mammals and birds. Proc Biol Sci 278(1717):2477–2485.

2. Dawson A, King VM, Bentley GE, Ball GF (2001) Photoperiodic control of seasonality inbirds. J Biol Rhythms 16(4):365–380.

3. Goldman BD (2001) Mammalian photoperiodic system: Formal properties and neu-roendocrine mechanisms of photoperiodic time measurement. J Biol Rhythms 16(4):283–301.

4. Hoffman RA, Reiter RJ (1965) Pineal gland: Influence on gonads of male hamsters.Science 148(3677):1609–1611.

5. Reiter RJ (1980) The pineal and its hormones in the control of reproduction inmammals. Endocr Rev 1(2):109–131.

6. Carter DS, Goldman BD (1983) Antigonadal effects of timed melatonin infusion inpinealectomized male Djungarian hamsters: Duration is the critical parameter.Endocrinology 113:1261–1267.

7. Badura LL, Goldman BD (1992) Central sites mediating reproductive responses tomelatonin in juvenile male Siberian hamsters. Brain Res 598(1-2):98–106.

8. Maywood ES, Hastings MH (1995) Lesions of the iodomelatonin-binding sites of themediobasal hypothalamus spare the lactotropic, but block the gonadotropic responseof male Syrian hamsters to short photoperiod and to melatonin. Endocrinology136(1):144–153.

9. Bittman EL (1978) Hamster refractoriness: The role of insensitivity of pineal targettissues. Science 202(4368):648–650.

10. Freeman DA, Zucker I (2001) Refractoriness to melatonin occurs independently atmultiple brain sites in Siberian hamsters. Proc Natl Acad Sci USA 98(11):6447–6452.

11. O’Jile JR, Bartness TJ (1992) Effects of thyroxine on the photoperiodic control ofenergy balance and reproductive status in Siberian hamsters. Physiol Behav 52(2):267–270.

12. Hanon EA, et al. (2008) Ancestral TSH mechanism signals summer in a photoperiodicmammal. Curr Biol 18(15):1147–1152.

13. Yoshimura T, et al. (2003) Light-induced hormone conversion of T4 to T3 regulatesphotoperiodic response of gonads in birds. Nature 426(6963):178–181.

14. Nakao N, et al. (2008) Thyrotrophin in the pars tuberalis triggers photoperiodic re-sponse. Nature 452(7185):317–322.

15. Barrett P, et al. (2007) Hypothalamic thyroid hormone catabolism acts as a gatekeeperfor the seasonal control of body weight and reproduction. Endocrinology 148(8):3608–3617.

16. Ono H, et al. (2008) Involvement of thyrotropin in photoperiodic signal transductionin mice. Proc Natl Acad Sci USA 105(47):18238–18242.

17. Yasuo S, Yoshimura T (2009) Comparative analysis of the molecular basis of photo-periodic signal transduction in vertebrates. Integr Comp Biol 49(5):507–518.

18. Li E (2002) Chromatin modification and epigenetic reprogramming in mammaliandevelopment. Nat Rev Genet 3(9):662–673.

19. Suzuki MM, Bird A (2008) DNA methylation landscapes: Provocative insights fromepigenomics. Nat Rev Genet 9(6):465–476.

20. Hollis F, Duclot F, Gunjan A, Kabbaj M (2011) Individual differences in the effect ofsocial defeat on anhedonia and histone acetylation in the rat hippocampus. HormBehav 59(3):331–337.

21. Bonthuis PJ, Patteson JK, Rissman EF (2011) Acquisition of sexual receptivity: Roles ofchromatin acetylation, estrogen receptor-alpha, and ovarian hormones. Endocrinology152(8):3172–3181.

22. Auger CJ, Coss D, Auger AP, Forbes-Lorman RM (2011) Epigenetic control of vaso-pressin expression is maintained by steroid hormones in the adult male rat brain. ProcNatl Acad Sci USA 108(10):4242–4247.

23. Nugent BM, McCarthy MM (2011) Epigenetic underpinnings of developmental sexdifferences in the brain. Neuroendocrinology 93(3):150–158.

24. Duong HA, Robles MS, Knutti D, Weitz CJ (2011) A molecular mechanism for circadianclock negative feedback. Science 332(6036):1436–1439.

25. Herb BR, et al. (2012) Reversible switching between epigenetic states in honeybeebehavioral subcastes. Nat Neurosci 15(10):1371–1373.

26. Métivier R, et al. (2008) Cyclical DNA methylation of a transcriptionally active pro-moter. Nature 452(7183):45–50.

27. Gardiner-Garden M, Frommer M (1987) CpG islands in vertebrate genomes. J Mol Biol196(2):261–282.

28. Bartness TJ, Powers JB, Hastings MH, Bittman EL, Goldman BD (1993) The timed in-fusion paradigm for melatonin delivery: What has it taught us about the melatoninsignal, its reception, and the photoperiodic control of seasonal responses? J Pineal Res15(4):161–190.

29. Purnell MR, Whish WJD (1980) Novel inhibitors of poly(ADP-ribose) synthetase.Biochem J 185(3):775–777.

30. Prendergast BJ, Mosinger B, Jr., Kolattukudy PE, Nelson RJ (2002) Hypothalamic geneexpression in reproductively photoresponsive and photorefractory Siberian hamsters.Proc Natl Acad Sci USA 99(25):16291–16296.

31. Paul MJ, Zucker I, Schwartz WJ (2008) Tracking the seasons: The internal calendars ofvertebrates. Philos Trans R Soc Lond B Biol Sci 363(1490):341–361.

32. Day JJ, Sweatt JD (2011) Epigenetic mechanisms in cognition. Neuron 70(5):813–829.33. Guo JU, et al. (2011) Neuronal activity modifies the DNA methylation landscape in the

adult brain. Nat Neurosci 14(10):1345–1351.34. Perfito N, et al. (2012) Anticipating spring: Wild populations of great tits (Parus

major) differ in expression of key genes for photoperiodic time measurement. PLoSONE 7(4):e34997.

35. Watanabe T, et al. (2007) Hypothalamic expression of thyroid hormone-activatingand -inactivating enzyme genes in relation to photorefractoriness in birds andmammals. Am J Physiol Regul Integr Comp Physiol 292(1):R568–R572.

36. Freeman DA, Teubner BJ, Smith CD, Prendergast BJ (2007) Exogenous T3 mimics longday lengths in Siberian hamsters. Am J Physiol Regul Integr Comp Physiol 292(6):R2368–R2372.

37. Dupré SM, et al. (2010) Identification of Eya3 and TAC1 as long-day signals in thesheep pituitary. Curr Biol 20(9):829–835.

38. Kouzmenko A, Ohtake F, Fujiki R, Kato S (2010) Hormonal gene regulation throughDNA methylation and demethylation. Epigenomics 2(6):765–774.

39. Matsuda KI, et al. (2011) Histone deacetylation during brain development is essentialfor permanent masculinization of sexual behavior. Endocrinology 152(7):2760–2767.

40. Tsai HW, Grant PA, Rissman EF (2009) Sex differences in histone modifications in theneonatal mouse brain. Epigenetics 4(1):47–53.

41. Nugent BM, Schwarz JM, McCarthy MM (2011) Hormonally mediated epigeneticchanges to steroid receptors in the developing brain: Implications for sexual differ-entiation. Horm Behav 59(3):338–344.

42. Roth TL, Sweatt JD (2011) Annual Research Review: Epigenetic mechanisms and en-vironmental shaping of the brain during sensitive periods of development. J ChildPsychol Psychiatry 52(4):398–408.

43. Alenghat T, et al. (2008) Nuclear receptor corepressor and histone deacetylase 3govern circadian metabolic physiology. Nature 456(7224):997–1000.

44. Henson JR, Carter SN, Freeman DA (2013) Exogenous T3 elicits long day-like alter-ations in testis size and the RFamides Kisspeptin and gonadotropin-inhibitory hor-mone in short-day Siberian hamsters. J Biol Rhythms 28(3):193–200.

45. Watson-Whitmyre M, Stetson MH (1985) A mathematical method for estimatingpaired testes weight from in situ testicular measurements in three species of hamster.Anat Rec 213(3):473–476.

46. Prendergast BJ, Pyter LM, Kampf-Lassin A, Patel PN, Stevenson TJ (2013) Rapid in-duction of hypothalamic iodothyronine deiodinase expression by photoperiod andmelatonin in juvenile Siberian hamsters (Phodopus sungorus). Endocrinology 154(2):831–841.

47. Prendergast BJ, Hotchkiss AK, Nelson RJ (2003) Photoperiodic regulation of circulatingleukocytes in juvenile Siberian hamsters: Mediation by melatonin and testosterone.J Biol Rhythms 18(6):473–480.

16656 | www.pnas.org/cgi/doi/10.1073/pnas.1310643110 Stevenson and Prendergast

Dow

nloa

ded

by g

uest

on

Janu

ary

6, 2

020

Corrections

PERSPECTIVECorrection for “Integrating the invisible fabric of nature intofisheries management,” by Joseph Travis, Felicia C. Coleman,Peter J. Auster, Philippe M. Cury, James A. Estes, Jose Orensanz,Charles H. Peterson, Mary E. Power, Robert S. Steneck, andJ. Timothy Wootton, which appeared in issue 2, January 14,2014, of Proc Natl Acad Sci USA (111:581–584; first publishedDecember 23, 2013; 10.1073/pnas.1305853111).The authors note that on page 582, left column, first paragraph,

line 3, “10 metric tons (MT)” should instead appear as “10 millionmetric tons (MT).”

www.pnas.org/cgi/doi/10.1073/pnas.1402460111

BIOPHYSICS AND COMPUTATIONAL BIOLOGYCorrection for “Dynamics transitions at the outer vestibule ofthe KcsA potassium channel during gating,” by H. Raghuraman,Shahidul M. Islam, Soumi Mukherjee, Benoit Roux, and EduardoPerozo, which appeared in issue 5, February 4, 2014, of ProcNatl Acad Sci USA (111:1831–1836; first published January 15,2014; 10.1073/pnas.1314875111).The authors note that, due to a PNAS error, the author con-

tributions footnote appeared incorrectly. The corrected authorcontributions footnote appears below.

Author contributions: H.R. and E.P. designed research; H.R. performed research; H.R. andS.M. performed fluorescence measurements; S.M.I. and B.R. contributed the computa-tional analysis; H.R., S.M.I., S.M., B.R., and E.P. analyzed data; and H.R. and E.P. wrotethe paper.

www.pnas.org/cgi/doi/10.1073/pnas.1402205111

MEDICAL SCIENCESCorrection for “Blocking CD40-TRAF6 signaling is a therapeu-tic target in obesity-associated insulin resistance,” by AntoniosChatzigeorgiou, Tom Seijkens, Barbara Zarzycka, David Engel,Marjorie Poggi, Susan van den Berg, Sjoerd van den Berg, OliverSoehnlein, Holger Winkels, Linda Beckers, Dirk Lievens, AnnDriessen, Pascal Kusters, Erik Biessen, Ruben Garcia-Martin,Anne Klotzsche-von Ameln, Marion Gijbels, Randolph Noelle,Louis Boon, Tilman Hackeng, Klaus Schulte, Aimin Xu, GertVriend, Sander Nabuurs, Kyoung-Jin Chung, Ko Willems van Dijk,Patrick C. N. Rensen, Norbert Gerdes, Menno de Winther,Norman L. Block, Andrew V. Schally, Christian Weber, Stefan R.Bornstein, Gerry Nicolaes, Triantafyllos Chavakis, and EstherLutgens, which appeared in issue 7, February 18, 2014, of ProcNatl Acad Sci USA (111:2686–2691; first published February 3,2014; 10.1073/pnas.1400419111).The authors note that the author name Klaus Schulte should

instead appear as Klaus-Martin Schulte. The corrected authorline appears below. The online version has been corrected.

Antonios Chatzigeorgiou, Tom Seijkens, BarbaraZarzycka, David Engel, Marjorie Poggi, Susanvan den Berg, Sjoerd van den Berg, Oliver Soehnlein,Holger Winkels, Linda Beckers, Dirk Lievens, AnnDriessen, Pascal Kusters, Erik Biessen, RubenGarcia-Martin, Anne Klotzsche-von Ameln, MarionGijbels, Randolph Noelle, Louis Boon, Tilman Hackeng,Klaus-Martin Schulte, Aimin Xu, Gert Vriend, SanderNabuurs, Kyoung-Jin Chung, Ko Willems van Dijk,Patrick C. N. Rensen, Norbert Gerdes, Menno de Winther,Norman L. Block, Andrew V. Schally, Christian Weber,Stefan R. Bornstein, Gerry Nicolaes, TriantafyllosChavakis, and Esther Lutgens

www.pnas.org/cgi/doi/10.1073/pnas.1403231111

4644–4646 | PNAS | March 25, 2014 | vol. 111 | no. 12 www.pnas.org

PHYSIOLOGYCorrection for “Reversible DNA methylation regulates seasonalphotoperiodic time measurement,” by Tyler J. Stevenson andBrian J. Prendergast, which appeared in issue 41, October 8, 2013,of Proc Natl Acad Sci USA (110:16651–16656; first publishedSeptember 25, 2013; 10.1073/pnas.1310643110).The authors note the following: “Recent high throughput se-

quencing has indicated that an upstream region of the dio3promoter sequence in our paper was the result of a PCR fusionerror. The reverse primer located –140bp upstream of the startcodon was not specific to Siberian hamsters. As a result, the tran-scription factor binding site analysis and sodium bisulfite-treatedDNA sequence analyses in the original publication were incorrect.To correct this error, we have resequenced the dio3 proximalpromoter and conducted replications of the transcription factorbinding site analyses and sequencing of sodium bisulfite-treatedDNA, using primer sequences with confirmed specificity to Siberianhamster DNA. The corrected dio3 promoter sequence exhibitedgreater homology with mice and human dio3 promoter (revised

Fig. S1), a greater number of CpG sites and a higher CpG fre-quency (revised Fig. S2) than previously reported. Analysis ofsodium bisulfite-treated DNA in the acute LD-SD (Fig. 1F) andphotorefractory experiments (Fig. 3E) yielded results consistentwith the originally-published report: dio3 promoter DNA meth-ylation was reduced in SD (revised Fig. 1F) and increased in SD-R(revised Fig. 3E). Revised transcription factor binding site analyseshave also been performed (revised Table S1). See corrected TableS2 for primers used on sodium bisulfite treated DNA. In addition,the reverse primers for the sequencing and MSRE PCR reactions(Table S2) were originally listed in the incorrect (3′-5′) orientation;the correct orientation (5′-3′) now appears in the revised version ofTable S2.“We thank Drs. Hugues Dardente and David Hazlerigg for

their assistance in identifying these errors.”As a result of this error, Figs. 1 and 3 and their legends appeared

incorrectly. The figures and their corrected legends appear below.These errors do not affect the main conclusions of the article.

Week0 4 8

mc(e

mulovsitseT

3 )

0.0

0.1

0.2

0.3

0.4

0.5

LD SD

*** ***

A

dio2 dio3

AN

Rm

evitaleR

0

2

4

6

8

10

12LDSD

*

B

dnmt1 dnmt3a dnmt3b

AN

Rm

evit aleR

0.0

0.5

1.0

1.5

2.0

2.5

3.0

*

C

*

LD SD

noitalyhtemlabol

G%

0

20

40

60

80

*

E

D

F

EC

III

LDSD

LDSD

% M

ethy

latio

n

dio3 promoter CpG site

0

20

40

60

80

100

SDLD

16 15 8 714 13 6 512 11 4 310 9 2 117

-12-249

Fig. 1. Short photoperiods inhibit reproduction and activate hypothalamic mRNA expression via epigenetic mechanisms. Acute transfer from LD to SD pho-toperiods caused gonadal regression (A), increased hypothalamic dio3 mRNA expression (B), and decreased hypothalamic dnmt1 and dnmt3b mRNA expression(C) after 8 wk. (D) Immunocytochemical localization of DNMT3b (DNMT3b-ir) in the hamster mediobasal hypothalamus (MBH). DNMT3b-ir was evidentthroughout the MBH and in the ependymal cell (EC) layer along the third ventricle (III). (E) Transfer from LD to SD reduced DNA methylation in the dio3 proximalpromoter, as measured using an MSRE assay. (F) Proportion of LD and SD hamsters in which no unmethylated DNA was detected at each of 17 CpG sites in thedio3 proximal promoter, as assessed by direct sequencing of sodium bisulfite-treated DNA. The abscissa (not to scale) depicts the 17 CpG sites from –249 to thestart codon of the dio3 proximal promoter. Averaging across the entire promoter region that was sequenced, evidence of unmethylation was evident on 15% ofCpG sites in LD (i.e., on 85% of CpG sites examined in LD hamsters, no detectable C-to-T bisulfite conversion occurred), whereas in SD, evidence of unmethylatedDNA was present on 42% of CpG sites (χ2 = 18.4, P < 0.002). All data in panels A–E are mean ± SEM. *P < 0.05, ***P < 0.005 vs. LD value.

PNAS | March 25, 2014 | vol. 111 | no. 12 | 4645

CORR

ECTIONS

www.pnas.org/cgi/doi/10.1073/pnas.1321493110

Male Female

)gm(

ssam

setseT

0

200

400

600

800

1000

)gm(

ssa

ms u

ret

U

0

50

100

150

200

250

300A

*** ***

dio2 dio3

AN

Rm

evita leR

0

1

2

3

4

5

*

B

SD10 wks

SD-R42 wks

LD

evitaleR

b3tm nd

AN

Rm

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

1.6

1.8

***

C

D

noitalyhtem

%

0

5

10

15

20

25

30

*

LD SD10 wks

SD-R42 wks

LD SD-10 wks SD-42 wks

LD SD-10 wks SD-42 wks

dio3 promoter CpG site

E

% M

ethy

latio

n

SD-RSDLD

0

20

40

60

80

100

16 15 8 714 13 6 512 11 4 310 9 2 117

-12-249

Fig. 3. Neuroendocrine refractoriness to SD reverses patterns of DNA methylation induced by acute SD exposure. (A) Acute (10 wk, SD) exposure to SDinduced gonadal regression, whereas prolonged exposure (42 wk, SD-R) triggered neuroendocrine refractoriness and gonadal recrudescence. Refractorinessin SD-R hamsters was characterized by a complete reversal of hypothalamic dio3 and dnmt3b mRNA expression (B and C) and by remethylation of DNA in thedio3 proximal promoter (D). (E) Proportion of LD, SD, and SD-R hamsters in which no unmethylated DNA was detected in each of 17 CpG sites in the dio3proximal promoter, as assessed by direct sequencing of sodium bisulfite DNA from the whole hypothalamus. The abscissa (not to scale) depicts the 17 CpGsites from –249 to the start codon of the dio3 proximal promoter. Averaging across the promoter region that was sequenced, evidence of unmethylation wasevident on 25% of CpG sites in LD hamsters, whereas in SD hamsters, evidence of unmethylation was present on 39% of CpG sites (χ2 = 5.08, P < 0.03). In SD-Rhamsters, methylation patterns returned to LD-like values, and evidence of unmethylation was detected on 29% of CpG sites examined (χ2 = 0.46, P > 0.40 vs.LD). Five of 17 sites (sites 16, 13, 12, 2, and 1) exhibited reversals in the pattern of methylation in SD-R hamsters. All data in panels A–D are mean ± SEM. *P <0.05; ***P < 0.005 vs. LD value.

4646 | www.pnas.org