loss (supplementary materials). However, theprominent mountains at the western margin ofSP, and the strange, multikilometer-high moundfeatures to the south are both young geologicallyand presumably composed of relatively strong,water-ice–based geologicalmaterials. Their origin,and what drove their formation so late in solarsystem history, remain uncertain. What is morecertain is that all three major Kuiper belt bodies(past or present) visited by spacecraft so far—Pluto,Charon, and Triton—are more different than sim-ilar and bear witness to the potential diversityawaiting the future exploration of their realm.

REFERENCES AND NOTES

1. H. A. Weaver, W. C. Gibson, M. B. Tapley, L. A. Young,S. A. Stern, Space Sci. Rev. 140, 75–91 (2008).

2. L. A. Young et al., Space Sci. Rev. 140, 93–127 (2008).3. J. M. Moore et al., Icarus 246, 65–81 (2015).4. Materials and methods are available as supplementary

materials on Science Online.5. S. A. Stern et al., Science 350, aad1815 (2015).6. J. Eluszkiewicz, D. J. Stevenson, Geophys. Res. Lett. 17,

1753–1756 (1990).7. Y. Yamashita, M. Kato, M. Arakawa, Icarus 207, 972–977 (2010).8. W. M. Grundy et al., Science 351, aad9189 (2016).9. G. C. Collins et al., in Planetary Tectonics, R. A. Schultz,

T. R. Watters, Eds. (Cambridge Univ. Press, New York, 2010),pp. 264–350.

10. R. Gomes, H. F. Levison, K. Tsiganis, A. Morbidelli, Nature 435,466–469 (2005).

11. S. Greenstreet, B. Gladman, W. B. McKinnon, Icarus 258,267–288 (2015).

12. G. R. Gladstone et al., Science 351, aad8866 (2016).13. K. N. Singer, S. A. Stern, Astrophys. J. 808, L50 (2015).14. G. Robuchon, F. Nimmo, Icarus 216, 426–439 (2011).15. M. Brozović, M. R. Showalter, R. A. Jacobson, M. W. Buie,

Icarus 246, 317–329 (2015).16. P. M. Schenk, J. M. Moore, in Solar System Ices, B. Schmitt,

C. de Bergh, M. Festou, Eds. (Kluwer Academic, Dordrecht,Netherlands, 1998), pp. 551–578.

17. D. G. Jankowski, S. W. Squyres, Science 241, 1322–1325(1988).

18. L. M. Prockter et al., Space Sci. Rev. 153, 63–111 (2010).19. G. Schubert et al., Space Sci. Rev. 153, 447–484 (2010).20. M. Manga, C.-Y. Wang, Geophys. Res. Lett. 34, L07202

(2007).21. E. L. Schaller, M. E. Brown, Astrophys. J. 659, L61–L64 (2007).22. W. B. McKinnon, in Treatise on Geophysics, G. Schubert, Ed.

(Elsevier, Amsterdam, ed. 2, 2015), vol. 10, pp. 637–651.23. D. C. Reuter et al., Space Sci. Rev. 140, 129–154 (2008).

ACKNOWLEDGMENTS

We thank the many engineers who have contributed to thesuccess of the New Horizons mission and NASA’s DeepSpace Network for a decade of excellent support to New Horizons.We thank the reviewers for close and meticulous reading,and P. Engebretson for contribution to figure production.S.A.S. is also affiliated with Florida Space Institute, UwinguLLC, Golden Spike Co., and World View Enterprises. H.J.R. isalso affiliated with B612 Foundation and Cornell TechnicalService. Supporting imagery is available in the supplementarymaterials. As contractually agreed to with NASA, fully calibratedNew Horizons Pluto system data will be released via theNASA Planetary Data System at https://pds.nasa.gov/ in aseries of stages in 2016 and 2017 as the data set is fullydownlinked and calibrated. This work was supported byNASA’s New Horizons project.

SUPPLEMENTARY MATERIALS

www.sciencemag.org/content/351/6279/1284/suppl/DC1Materials and MethodsSupplementary TextFigs. S1 to S15Table S1References (24–83)

16 November 2015; accepted 11 February 201610.1126/science.aad7055

FEEDING BEHAVIOR

The nutrient sensor OGT in PVNneurons regulates feedingOlof Lagerlöf,1,2 Julia E. Slocomb,3 Ingie Hong,1 Yeka Aponte,1,4 Seth Blackshaw,1

Gerald W. Hart,2 Richard L. Huganir1*

Maintaining energy homeostasis is crucial for the survival and health of organisms.The brain regulates feeding by responding to dietary factors and metabolic signals fromperipheral organs. It is unclear how the brain interprets these signals. O-GlcNActransferase (OGT) catalyzes the posttranslational modification of proteins by O-GlcNAcand is regulated by nutrient access. Here, we show that acute deletion of OGT fromaCaMKII-positive neurons in adult mice caused obesity from overeating. The hyperphagiaderived from the paraventricular nucleus (PVN) of the hypothalamus, where lossof OGTwas associated with impaired satiety. These results identify O-GlcNAcylation inaCaMKII neurons of the PVN as an important molecular mechanism that regulatesfeeding behavior.

Obesity is a major contributor to diseasethroughout the world (1). Currently, thereis no successful and available treatmentfor the majority of obese patients. Oneof the genes most commonly linked to

human obesity, Gnpda2, affects flux throughthe hexosamine biosynthesis pathway (HBP)(2–4). The HBP produces uridine-diphosphate:N-acetylglucosamine (UDP-GlcNAc), which isthe donor substrate for the enzyme O-GlcNActransferase (OGT). OGT cleaves UDP-GlcNAc andcovalently attaches GlcNAc to the hydroxyl groupof serine or threonine in nuclear and cytoplasmicproteins in b-linkage (O-GlcNAc). Nutrient accessdirectly via the HBP, and metabolic hormonessuch as insulin regulate the activity of OGT (5, 6).Although OGT has been shown to be importantfor neuronal development, the role of OGT for ma-ture brain function is almost completely unknown(7–9). To study the function of OGT in normalbehavior, we created conditional OGT knockoutmice by crossing floxed OGT mice (OGTFl) withmice expressing a tamoxifen-inducible version ofCre recombinase under the aCaMKII promoter(aCaMKII-CreERT2). This enables acute brain-specific deletion of OGT in aCaMKII-expressingneurons in adult mice, which we confirmed bymeansof immunohistochemistry,Westernblotting,and polymerase chain reaction (fig. S1). Knock-out of OGT in other tissues and cells has beenshown to lead to decreases in cell number,probably because of impaired mitosis, and in fact,constitutive knockout of OGT leads to early em-bryonic lethality (8, 10, 11). In contrast, in post-

mitotic neurons deletion of OGT did not affectcell number in vitro or in vivo (fig. S2, A and B).Acute and brain-specific loss of OGT in adult

mice caused rapid weight gain (Fig. 1, A and B,and fig. S3, A and B).Within 3weeks, the amountof adipose tissue tripled, whereas the leanweighthad not changed, as quantified with whole-bodynuclear magnetic resonance (NMR) [fat mass,n = 7 wild-type (WT) mice, 2.5 ± 0.21 g; n = 6OGT knockout mice, 8.3 ± 0.86 g] (fig. S3, C andD). The incorporation of fat was general and notparticular to any specific body region (fig. S3, Eto G). Obesity can result from either excessivecaloric intake or insufficient energy expenditure.Daily food intake rapidly increased upon knock-out of OGT and plateaued at a level more thantwice as high (Fig. 1C). If access to food was re-stricted to the same amount consumed by WTmice, the OGT knockout mice retained normalbody weight. When free access to food was re-introduced, the OGT knockout mice quicklyapproached the weight of OGT knockout litter-mates who had been fed ad libitum throughoutthe experiment (Fig. 1D). To quantify food intakeand energy expenditure simultaneously in realtime, we used comprehensive laboratory animalmonitoring system (CLAMS). CLAMS confirmedthat loss of OGT leads to hyperphagia (fig. S3H).Energy expenditure was actually increased (fig.S3I). The accelerated energy expenditure resultedfrom, at least in part, hyperactivity (fig. S3J). Asexpected from a combination of hyperphagiaand hyperactivity, knocking out OGT causedhigher vO2 and vCO2, leading to a respiratory ex-change ratio (RER) above 1 (n = 17 WT mousedays (6mice), 0.94 ± 0.004; n = 20OGT knockoutmouse days (6mice), 1.04 ± 0.007) (fig. S3, K toM).Daily food intake is a factor of both the size of

eachmeal andmeal frequency. A normal diurnalrhythm was preserved in OGT knockout mice(Fig. 1E and fig. S4, A and B). Although therewas no difference in meal frequency, loss of OGTincreased meal size as well as meal duration(meal size, n = 268 WT mouse meals (6 mice),

SCIENCE sciencemag.org 18 MARCH 2016 • VOL 351 ISSUE 6279 1293

1Solomon H. Snyder Department of Neuroscience, KavliNeuroscience Discovery Institute, Johns Hopkins UniversitySchool of Medicine, Baltimore, MD 21205, USA. 2Departmentof Biological Chemistry, Johns Hopkins University School ofMedicine, Baltimore, MD 21205, USA. 3National Institute onDrug Abuse + National Institutes of Health/Johns HopkinsUniversity Graduate Partnership Program, Baltimore, MD21224, USA. 4Intramural Research Program, NeuronalCircuits and Behavior Unit, National Institute on Drug Abuse,Baltimore, MD 21224, USA.*Corresponding author. E-mail: rhuganir@jhmi.edu

RESEARCH | RESEARCH ARTICLES

on

Nov

embe

r 5,

201

6ht

tp://

scie

nce.

scie

ncem

ag.o

rg/

Dow

nloa

ded

from

0.25 ± 0.02 g; n = 330 OGT knockout mousemeals (7 mice), 0.42 ± 0.02 g) (Fig. 1E and fig. S4,C to E). The point at which a meal is terminateddepends on the total caloric content of the foodbut also noncaloric determinants of the food,mainly its volume and composition. When fedeither regular carbohydrate-rich pellets or fat-based food paste with higher caloric density, theOGT knockout mice overate the same amountof calories (fig. S4F).Immunohistochemistry for OGT showed that

within the core feeding circuitry, the major lossof OGT in the knockout mice occurred in medialnuclei of the hypothalamus, most notably in theparaventricular nucleus (PVN) (Fig. 2A and fig.S5A). There was only minor loss of OGT fromnuclei of themidbrain and the brainstem (Fig. 2Aand fig. S5A).UsingaCaMKII-CreER+×TdTFlmiceand immunohistochemistry, we noticed that someaCaMKIIPVNcells expressed thyrotropin-releasinghormone (TRH) and some oxytocin (fig. S6A). Be-fore anymajor weight gain, OGTdeletion reducedthe expression of TRH, as determined with insitu hybridization, in a subpopulation of cellsin the PVN without affecting their viability,whereas several other known neuropeptidesregulating feeding behavior were largely unal-tered: proopiomelanocortin (POMC), cocaine-and amphetamine-regulated transcript (CART),orexin, agouti-related peptide (AgRP), neuropep-tide Y (NPY), vasopressin (AVP), and oxytocin(TRH, n= 4WTmice, 40 ± 4.1 cells/150 × 103 mm2;n = 4 OGT knockout mice, 23 ± 2.9 cells/150 ×103 mm2) (Fig. 2B and figs. S6, B to E, and S7A)(12, 13). Next, we infected hypothalamic organo-typic cultures with a virus that expresses greenfluorescent protein (GFP) under the aCaMKII pro-moter and treated the explants with varying con-centrations of glucose. ThePVNretained its normalmorphology and expression of neuropeptide dur-ing the extent of the experiment (~2.5 weeks)(fig. S8A). Immunohistochemistry for O-GlcNAcdemonstrated that 1 hour of treatmentwith 5mMglucose, simulating physiological andmeal-derivedfluctuations in glucose, increased O-GlcNAc inaCaMKII-positive cells [1 mM glucose, n = 360cells, 134 ± 10 intensity/cell (arbitrary units, a.u.);5mMglucose n = 330 cells, 211 ± 12 intensity/cell(a.u.)] (Fig. 2C) (14–16). In contrast, there wasno change in O-GlcNAc levels in neighboringaCaMKII-negative cells (fig. S8B). Incubation for16 hourswith 25mMglucose raised theO-GlcNAclevels in aCaMKII cells to a larger extent [5 mMglucose, n = 390 cells, 156 ± 9 intensity/cell (a.u.);25 mM glucose, n = 375 cells, 298 ± 14 intensity/cell (a.u.)] (fig. S8, C and D). Unlike cortex and thehippocampus, where O-GlcNAc levels mirror foodintake, it has recently been reported that fastingincreased O-GlcNAc in AgRP neurons, which hasbeen proposed to be due to stimulation of OGTexpression by ghrelin in these cells (6, 7, 15). Inthe PVN, fasting decreasedO-GlcNAc, but only inaCaMKII-positive cells [aCaMKII-positive cells,fedmice n = 455 cells, 158 ± 5 intensity/cell (a.u.);starved mice n = 259 cells, 81 ± 6 intensity/cell(a.u.)] (fig. S8E). In addition, we compared theexpression of the immediate early gene c-Fos

in aCaMKII PVN cells in vivo in mice fed adlibitum or refed upon starvation. This enabled usto characterize cellular responses to food intake asmice start eating after a period of starvation. Foodintake appeared to activate aCaMKII cells in thePVN, and loss of OGT blocked this activationcompletely (fig. S8, F and G).O-GlcNAc is highly expressed in neuronal

synapses in the brain (17). We crossed aCaMKII-CreER+ ×OGTFl/WT ×TdTFlmice to assesswhetherOGT regulates excitatory synaptic input ontoaCaMKII PVN cells. If deleting OGT attenuates

excitatory input, it would explain, at least in part,howOGT regulates the activity of these cells. Themean capacitance of labeled cells averaged ~15 pFand did not differ betweenWT and knockoutmice(fig. S9A). The mean miniature excitatory post-synaptic current (mEPSC) frequency decreasedby 72% in OGT knockout cells (n = 6 WT cells,1.75 ± 0.11 Hz; n = 6 OGT knockout cells, 0.50 ±0.10 Hz) (Fig. 3, A and B). The sharp decrease inmEPSC frequency suggests that OGT is essentialfor maintaining the number of functional excit-atory synapses onto aCaMKII PVN neurons.

1294 18 MARCH 2016 • VOL 351 ISSUE 6279 sciencemag.org SCIENCE

Fig. 1. Acute deletion of OGT in aCaMKII-positive neurons causes hyperphagia-dependent obe-sity. (A) Photo of mice 4 weeks after OGTdeletion. (B) Body weight time course [n = 8 WTmice, n = 8OGT knockout mice; repeated-measures two-way analysis of variance (ANOVA) with post hoc Bonferronitest, P < 0.05]. (C) Daily food intake time course (n = 6 WTmice, n = 6 OGT knockout mice; repeated-measures two-way ANOVA with post hoc Bonferroni test, P < 0.05). (D) Body weight time course uponpair-feeding (n = 4 free mice, n = 4 restricted mice, n = 3 WTmice). (E) Food intake every 15 min fromsample mice. Quantifications represent mean ± SEM.

Fig. 2. OGT-mediated hyperphagia is associatedwith feeding circuitry function in the PVN.(A) aCaMKII expression within feeding circuitry(schematized from fig. S5A). (B) TRH expression.(Left) Image of probe staining. (Right) Quantifi-cation (n = 4WTmice, n = 4 OGT knockout mice;two-tailed t test: P < 0.05). (C) (Left) GFP andO-GlcNAc in the PVN. (Right) Quantification ofO-GlcNAc intensity in GFP-positive cells (1 mM, n =360 cells, 5mM, n=330 cells; two-tailed t test:P<0.05). Quantifications represent mean ± SEM.

RESEARCH | RESEARCH ARTICLES

on

Nov

embe

r 5,

201

6ht

tp://

scie

nce.

scie

ncem

ag.o

rg/

Dow

nloa

ded

from

The mean mEPSC amplitude was also de-creased, but to a much smaller degree (n = 6WTcells 19.7 ± 1.2 pA; n = 6 knockout cells, 15.6 ±0.9 pA) (Fig. 3, A and C). Neither the rise timenor the decay time of the EPSCs changed upon

deletion of OGT (fig. S9, B and C). Pharmaco-logical inhibition of glutamatergic signaling inthe PVN has previously been shown to elicit in-tense feeding (18).Selectively deleting OGT in the aCaMKII cells

of the PVN by means of stereotactic virus injec-tion caused concurrent obesity and hyperphagia(Fig. 4, A to C, and fig. S9D). Because deletion ofOGT in aCaMKII cells in the PVN prevents theirfeeding-induced activation and leads to hyper-phagia, we predicted that stimulating those cellswould decrease food intake. Activating aCaMKIIPVN cells optogenetically decreased cumulativefood consumption over 24 hours (baseline, n = 11experiments (5 mice), 3.6 ± 0.14 g; 50 Hz, n = 11experiments (5 mice), 2.6 ± 0.33 g) (Fig. 4, D andE) (19). The amount of food ingested permeal wassmaller, whereas there was no significant effecton meal frequency (meal size, baseline, n = 125meals, 0.25 ± 0.014 g; 50Hz, n = 110meals, 0.20±0.014 g) (Fig. 4F and fig. S9E). In a second ex-periment, we started the stimulation at the onsetof darkness after fasting the mice during the lightphase. Fasting increased food intake, and 20 Hzoptogenetic stimulation for 4 hours produced asignificant decrease in food intake, whereas 50Hzstimulation for 1 or 4 hours produced large, sig-nificant drops in feeding (fig. S9F). The decreasein meal size with 50 Hz stimulation was evidentin the first meal, without any change in latencyto initiation of feeding (fig. S9, G and H).

Together, our data favor a model in which thefunction of OGT is to couple energy intake withenergy need, at least in part, by regulating theexcitatory synaptic input in aCaMKII PVN cells(Fig. 4G). The select effect on meal terminationsuggests that OGT regulates satiation. aCaMKIIneurons are often excitatory. Although vGlut2-positive excitatory neurons in the PVN decreasefood intake, there appears to be only partial over-lap between aCaMKII and vGlut2 expression inthe PVN (20, 21). Because glucose remains ele-vated in the cerebrospinal fluidmore than 1 hourupon eating, the feeding-related changes inO-GlcNAc integrates nutrient availability on a timescale longer than a single meal (14, 22, 23). Thebrain promotes satiation by coordinating adipo-kines, reflecting body energy depots, and acutefood-derived signals into circuits that turn offfeeding. Rather than constituting a link in meal-to-meal satiety feedback loops, our observationssuggest that OGT controls the threshold of suchloops through its regulation of excitatory synap-tic transmission onto aCaMKII PVN neurons.The observation thatOGTknockoutmice quicklyreached a plateau in daily food intake supportsthis idea. Thresholding satiation between mealsconfers the behavioral advantage of stabilizingcaloric intake over time so that the previousmealinforms on the caloric need of the next. Thesedata do not exclude additional, faster regulationof OGT activity and levels viametabolic hormones.These findings identify the regulation of excita-tory synapses onto aCaMKII PVNneurons byOGTas an important mechanism underlying satiation,representing a potential medicinal target for hu-man obesity.

REFERENCES AND NOTES

1. C. E. Lewis et al., Circulation 119, 3263–3271 (2009).2. E. K. Speliotes et al., Nat. Genet. 42, 937–948

(2010).3. R. Gutierrez-Aguilar, D. H. Kim, S. C. Woods, R. J. Seeley,

Obesity (Silver Spring) 20, 306–312 (2012).4. H. Wolosker et al., FASEB J. 12, 91–99 (1998).5. G. W. Hart, C. Slawson, G. Ramirez-Correa, O. Lagerlöf, Annu.

Rev. Biochem. 80, 825–858 (2011).6. X. Li, F. Lu, J. Z. Wang, C. X. Gong, Eur. J. Neurosci. 23,

2078–2086 (2006).7. H. B. Ruan et al., Cell 159, 306–317 (2014).8. N. O’Donnell, N. E. Zachara, G. W. Hart, J. D. Marth, Mol. Cell.

Biol. 24, 1680–1690 (2004).9. J. E. Rexach et al., Nat. Chem. Biol. 8, 253–261 (2012).10. C. Slawson et al., J. Biol. Chem. 280, 32944–32956

(2005).11. R. Shafi et al., Proc. Natl. Acad. Sci. U.S.A. 97, 5735–5739

(2000).12. E. Vijayan, S. M. McCann, Endocrinology 100, 1727–1730

(1977).13. J. W. Sohn, J. K. Elmquist, K. W. Williams, Trends Neurosci. 36,

504–512 (2013).14. A. B. Steffens, A. J. Scheurink, D. Porte Jr., S. C. Woods, Am. J.

Physiol. 255, R200–R204 (1988).15. G. Pekkurnaz, J. C. Trinidad, X. Wang, D. Kong, T. L. Schwarz,

Cell 158, 54–68 (2014).16. M. Karnani, D. Burdakov, Am. J. Physiol. Regul. Integr. Comp.

Physiol. 300, R47–R55 (2011).17. J. C. Trinidad et al., Mol. Cell. Proteomics 11, 215–229

(2012).18. S. R. Hettes et al., Brain Res. 992, 167–178 (2003).19. A. K. Sutton et al., J. Neurosci. 34, 15306–15318

(2014).20. Y. Xu et al., Cell Metab. 18, 860–870 (2013).21. A. Wallén-Mackenzie et al., J. Neurosci. 29, 2238–2251

(2009).

SCIENCE sciencemag.org 18 MARCH 2016 • VOL 351 ISSUE 6279 1295

Fig. 3. OGT regulates excitatory synaptic func-tion inaCaMKII PVNneurons. (A) Sample tracesfrom WTand OGT knockout cells. (B) mEPSC fre-quency and (C) amplitude in aCaMKII-positive PVNneurons (n = 6 WTcells, n = 6 OGT knockout cells;two-tailed t test, P < 0.05). Quantifications rep-resent mean ± SEM.

Fig. 4. OGT regulates feeding behavior in aCaMKII neurons of the PVN. (A) (Top) Schematic ofstereotactic PVN injection. (Bottom) GFP expression in the PVN after stereotactic injection. (B) Bodyweight and (C) daily food intake time course (n = 5 WT mice, n = 7 OGT knockout mice; repeated-measures two-way ANOVA with post hoc Bonferroni test, P < 0.05). (D) Schematic of optogenetic setupwith aCaMKII-driven expression of channelrhodopsin-2 in the PVN with the abutting laser. (E and F) Foodintake after optogenetic stimulation. (E) 24 hours cumulative food intake. Bars represent average intakeover all mice, and lines represent average intake per individual mouse (baseline, n = 11 experiments (5 mice);stimulation, n = 11 experiments (5 mice); two-tailed t test, P < 0.05). (F) Average meal size (baseline,n = 125 meals (5 mice); stimulation, n = 110 meals (5 mice); two-tailed t test, P < 0.05). (G) Model ofOGT-dependent hyperphagia. Quantifications represent mean ± SEM.

RESEARCH | RESEARCH ARTICLES

on

Nov

embe

r 5,

201

6ht

tp://

scie

nce.

scie

ncem

ag.o

rg/

Dow

nloa

ded

from

22. American Diabetes Association, Diabetes Care 24, 775–778(2001).

23. S. Marshall, O. Nadeau, K. Yamasaki, J. Biol. Chem. 279,35313–35319 (2004).

ACKNOWLEDGMENTS

O.L., G.W.H., and R.L.H. designed all experiments; the optogeneticsexperiments were also designed by Y.A. and J.E.S., and theelecotrophysiological experiments were also designed by I.H.O.L. performed all experiments and analyzed all data but theelectrophysiology (I.H.) and the optogenetics (J.E.S.). Theoptogenetics data were analyzed by J.E.S., Y.A., and O.L. O.L.,

G.W.H., and R.L.H. wrote the manuscript. We thank G. Schütz forproviding the aCaMKII-CreERT2 mice, J. L. Bedont for help with insitu hybridization experiments, and R. H. White for assistance withmating and genotyping. All data necessary to understand andassess the conclusions of the manuscript are in the body of thepaper and in the supplementary materials. All primary data arearchived on a secure server located in the Department ofNeuroscience at Johns Hopkins University (JHU). All data will bemade available upon request. G.W.H. receives a share of royaltyreceived by the university on sales of the CTD 110.6 antibody,which are managed by JHU. The research was supported byNIH (grant R01NS036715 to R.L.H. and grants R01DK6167,

N01-HV-00240, and P01HL107153 to G.W.H.) and the NationalInstitute on Drug Abuse Intramural Research Program (Y.A.).

SUPPLEMENTARY MATERIALS

www.sciencemag.org/content/351/6279/1293/suppl/DC1Materials and MethodsFigs. S1 to S9References (24–29)

29 September 2015; accepted 5 February 201610.1126/science.aad5494

MICROBIOTA

The maternal microbiota drives earlypostnatal innate immune developmentMercedes Gomez de Agüero,1* Stephanie C. Ganal-Vonarburg,1* Tobias Fuhrer,2

Sandra Rupp,1 Yasuhiro Uchimura,1 Hai Li,1 Anna Steinert,1 Mathias Heikenwalder,3

Siegfried Hapfelmeier,4 Uwe Sauer,2 Kathy D. McCoy,1* Andrew J. Macpherson1*†

Postnatal colonization of the body with microbes is assumed to be the main stimulus topostnatal immune development. By transiently colonizing pregnant female mice, we showthat the maternal microbiota shapes the immune system of the offspring. Gestationalcolonization increases intestinal group 3 innate lymphoid cells and F4/80+CD11c+

mononuclear cells in the pups. Maternal colonization reprograms intestinal transcriptionalprofiles of the offspring, including increased expression of genes encoding epithelialantibacterial peptides and metabolism of microbial molecules. Some of these effects aredependent on maternal antibodies that potentially retain microbial molecules and transmitthem to the offspring during pregnancy and in milk. Pups born to mothers transientlycolonized in pregnancy are better able to avoid inflammatory responses to microbialmolecules and penetration of intestinal microbes.

During pregnancy, the eutherian fetus in-habits a largely sterile environment inutero, protected from infections by mater-nal immunity. Rejection of the allogeneicfetus is avoided through maternal and fe-

tal vascular separation, the immune privilegedstatus of the placental trophoblast, and gesta-tionalmaternal tolerancemechanisms (1). At birth,the situation changes dramatically as body sur-faces becomeprogressively colonizedwithmicrobes,directly exposing the immature neonatal immunesystem to potential pathogens (2, 3). Despitecontinued protection from the immunoglobulinsand antibacterial peptides in milk, the conse-quence of this transition for human health is thatmost of the worldwide mortality in children upto 5 years old is due to infectious disease (4–6).Immune system development is both prepro-

grammed in neonatal tissues and driven later byexposure to pathogenic and nonpathogenic mi-crobes (3). Germ-free mice have low immuno-

globulin concentrations; lymphopenia of lymphoidstructures; reduced bonemarrow leukocyte pools;and aberrant innate and adaptive immune func-tions (7, 8). It has beenwidely assumed thatmostmicrobiota-driven immune alterations are post-natal effects induced by the neonate’s ownmicro-biota (2, 9, 10). Here, we challenge this assumptionby asking how the maternal microbiota in preg-nancy alone affects the early postnatal immunesystem of the offspring.To achieve gestation-only colonization under

conditionswhere themice deliver their pups spon-taneously at term, we used a system in whichpregnant dams are transiently colonized withgenetically engineered Escherichia coli HA107(11). Because this strain does not persist in theintestine, pregnant dams become germ-free againbefore term and naturally deliver germ-free pups(fig. S1A). Although E. coli is a minor componentof the adult human microbiota, it is commonerin the neonatal intestine (12) and a frequent causeof human neonatal sepsis (13).

Gestation-only colonization shapesthe intestinal mucosal innateimmune composition

Gestation-only colonization with E. coli HA107altered the numbers of early postnatal intestinalinnate leukocytes in wild-type C57BL/6 mice. Atpostnatal day 14, there was an increase in small

intestinal innate lymphoid cell (ILC) proportionsand total numbers compared with germ-free con-trols, particularly the NKp46+RORgt+ ILC3 sub-set (Fig. 1, A and B, and fig. S1B). Small intestinalNKp46+RORgt+ ILC3 are described in germ-freemice (14), but persistently increased followingtransient gestational colonization, reaching amax-imum in 14- to 21-day-old pups: This increasepersisted even after weaning (Fig. 1C and fig. S1C),consistent with increased small intestinal ILC3content of colonized compared with germ-freemice (15) and the microbiota-dependent mod-ulation of RORgt expression in this subset (16).Increases in the expression of the cytokineinterleukin-22 (IL-22) in this population havebeen observed following permanent colonizationor the introduction of segmented filamentousbacteria to themicrobiota (17, 18). Total numbersof IL-22–expressing cells increased in line withthe increased NKp46+RORgt+ ILC3 numbers asa result of gestational colonization, although in-dividual IL-22 expression levels did not change,likely because the pups were born and raisedgerm-free (fig. S1, D to F).There was also an increase in the small and

large intestinal F4/80+CD11c+ mononuclear cells(iMNCs) in day 14 (d14) pups born to gestation-only colonized dams (Fig. 1, D and E, and fig. S2,A to C), whereas the F4/80+CD11c–macrophages,F4/80loCD11c+ dendritic cells (DCs), and the CD103+

or CD11b+ DC subpopulations were not signifi-cantly affected (Fig. 1, D and E, and fig. S2, B to E).The gestational effects on increased F4/80+CD11c+

iMNCswere alsomaximal betweenpostnatal days14 to 21, and they persisted until at least 8 weeksof age in the colon (Fig. 1F). Gestational coloniza-tion causedno significant changes in small intestinalILC2 numbers (fig. S3, A and B) or in other earlypostnatal innate leukocyte populations in eithersystemic or intestinal tissues (table S1). These re-sults showed that temporary colonization of apregnant dam has long-term consequences forcertain populations of innate lymphoid andmono-nuclear cells in the intestines of her offspring.We next sought to verify that the effects of

gestational E. coli on early postnatal innate leu-kocytes would also be seen with animals stablycolonized by a different microbiota both in themother and after birth. We compared C57BL/6animals carrying the defined altered Schaedlerflora (ASF) of eight microbes with germ-free con-trols. Both small intestinal NKp46+ ILC3 and in-testinal F4/80+CD11c+ MNC populations wereincreased in pups born to stably colonized ASF

1296 18 MARCH 2016 • VOL 351 ISSUE 6279 sciencemag.org SCIENCE

1Maurice Müller Laboratories (DKF), Universitätsklinik fürViszerale Chirurgie und Medizin Inselspital, Murtenstrasse35, University of Bern, 3010 Bern, Switzerland. 2Institute ofMolecular Systems Biology, Swiss Federal Institute ofTechnology (ETH) Zürich, 8093 Zürich, Switzerland. 3Divisionof Chronic Inflammation and Cancer, German Cancer ResearchCenter (DKFZ), Heidelberg, Germany. 4Institute for InfectiousDiseases, University of Bern, 3010 Bern, Switzerland.*These authors contributed equally to this work. †Correspondingauthor: E-mail: andrew.macpherson@insel.ch

RESEARCH | RESEARCH ARTICLES

on

Nov

embe

r 5,

201

6ht

tp://

scie

nce.

scie

ncem

ag.o

rg/

Dow

nloa

ded

from

(6279), 1293-1296. [doi: 10.1126/science.aad5494]351Science 2016) Blackshaw, Gerald W. Hart and Richard L. Huganir (March 17, Olof Lagerlöf, Julia E. Slocomb, Ingie Hong, Yeka Aponte, SethThe nutrient sensor OGT in PVN neurons regulates feeding

Editor's Summary

, this issue p. 1293; see also p. 1268Sciencehelps to couple caloric intake to caloric need.animals ate more at meal times, rather than eating more often. Thus, OGT seems to regulate satiety and by Schwartz). After the loss of OGT, the animals began to overeat and rapidly gained weight. TheO-GlcNAc transferase (OGT) from a subset of neurons in the mouse hypothalamus (see the Perspective

deleted an enzyme calledet al.they balance their caloric intake with their caloric needs. Lagerlöf −−Overeating and obesity are rapidly becoming worldwide problems. Normally, mice do not overeat

When enough isn't enough

This copy is for your personal, non-commercial use only.

Article Tools

http://science.sciencemag.org/content/351/6279/1293article tools: Visit the online version of this article to access the personalization and

Permissionshttp://www.sciencemag.org/about/permissions.dtlObtain information about reproducing this article:

is a registered trademark of AAAS. ScienceAdvancement of Science; all rights reserved. The title Avenue NW, Washington, DC 20005. Copyright 2016 by the American Association for thein December, by the American Association for the Advancement of Science, 1200 New York

(print ISSN 0036-8075; online ISSN 1095-9203) is published weekly, except the last weekScience

on

Nov

embe

r 5,

201

6ht

tp://

scie

nce.

scie

ncem

ag.o

rg/

Dow

nloa

ded

from

www.sciencemag.org/content/351/6279/1293/suppl/DC1

Supplementary Material for

The nutrient sensor OGT in PVN neurons regulates feeding

Olof Lagerlöf, Julia E. Slocomb, Ingie Hong, Yeka Aponte, Seth Blackshaw, Gerald W. Hart, Richard L. Huganir*

*Corresponding author. E-mail: rhuganir@jhmi.edu

Published 18 March 2016, Science 351, 1293 (2016)

DOI: 10.1126/science.aad5494

This PDF file includes:

Materials and Methods Figs. S1 to S9 Full Reference List

Materials and Methods Animals OGTFl and αCaMKII-CreERT2 mice have been derived previously (8, 24). They were fully backcrossed (>N11) to the C57BL6 background. Only males were used and they were about 6 weeks old at the start of experiments. Tamoxifen was injected accordingly: 2mg twice per day for 5 days (20mg total dose). Control mice were either vehicle (sunflower seed oil injected into OGTFl x αCaMKII-CreER+ mice) or genetic (tamoxifen injected into OGTFl x αCaMKII-CreER-) controls. No difference between controls was seen. Therefore, all control animals were combined. Mice were fed standard chow (24% protein, 18% fat, 58% carbohydrate) and had free access to food unless otherwise specified. Only littermates were compared. Apart from noted exceptions, the animals were kept on a 14 h/10 h light/dark cycle. For all animal studies, animals where randomly allocated to experimental groups. The experimenter was aware of the animal genotype when conducting experiments. The TdTFl mice express the fluorescent reporter TdTomato upon exposure to Cre recombinase and were originally developed by the Allen Institute (Ai9). All animals were housed according with the Johns Hopkins University Animal Care and Use Committee guidelines. All details pertaining to optogenetics experiments, including animal work, are described below under Optogenetics. Biochemistry Primary neuronal cell culture. Cortical neurons were prepared from timed E18 embryos, as previously described (25). Lentivirus was added at DIV (days in vitro) 2. Cells were harvested for analysis at DIV 11-13. Western blot. Cultured cells were harvested in RIPA buffer (50mM tris, 150mM sodium chloride, 1% NP-40, 1% deoxycholate, 0.1% SDS, 1mM EDTA, including inhibitors for proteases, phosphatases and O-GlcNAcases), solubilized for 20 min at 4°C and then spun at 13000 rpm. The supernatant was used for further analysis. When tissue was analyzed, unless otherwise noted, after removal from the animal, it was homogenized in homogenization buffer (HB; 0.32M sucrose, 10mM HEPES, including inhibitors for proteases, phosphatases and O-GlcNAcases ). 2X RIPA was then added to a final concentration of 1X RIPA for solubilization at 4°C for 20 min, before spun at 13000 rpm. Clear lysate, with top lipid layer discarded, was used for further analysis. Note that the muscle indicated in Fig. S1A is musculus femoris. Western blotting was performed according to standard procedures. Briefly, after separation by SDS-PAGE and transfer to polyvinylidene fluoride (PVDF) membranes, membranes were blocked in 3% bovine serum albumin (BSA), or 5% bovine milk when blotting for O-GlcNAc, and then probed with the following antibodies: OGT (AL25, produced in-house, 1:5000), O-GlcNAc (CTD110.6, produced in-house, 1:10000), HSC70 (Santa Cruz Biotechnology, sc7298, 1:5000). Virus For cell culture experiments, lentivirus was used. VSVG pseudotyped lentiviral particles were produced by standard procedures. In short, a FUGW vector expressing enhanced (eGFP) alone (Wt) or eGFP and Cre recombinase (KO) together with cDNA plasmids VSVG and delta 8.9 were transfected into HEK 293T cells using Lipofectamine

2

(25). The culture supernatant was collected 24 h and 48 h after transfection, then subjected to ultracentrifugation where the resulting pellet was resuspended in Neurobasal medium. For viral infections in vivo, adeno-associated virus (AAV) was used. All AAV were of serotype 2.1 and purchased from University of Pennsylvania Vector Core: For Wt; #AV-1-PV1917 (AAV1.CaMKII0.4.eGFP.WPRE.rBG) and for KO; #AV-1-PV2521 (AAV1.CaMKII.HI.eGFP-Cre.WPRE.SV40). All details pertaining to optogenetics experiments are described below under Optogenetics. Histology Immunohistochemistry. Animals were anesthetized with avertin (0.03ml/g) and then perfused with 4% paraformaldehyde (PFA) in phosphate-buffered saline (PBS). The brain was removed and post-fixed in 4% PFA/PBS overnight. After the brain had been equilibrated in 30% sucrose/PBS, it was frozen and serial sections (25 µm) were cut on a microtome. Free-floating sections were blocked in IHC buffer (5-10% normal goat serum diluted in 0.25% triton X-100/PBS) overnight at 4°C, incubated with the primary antibody overnight at 4°C and then with the secondary antibody either for 2 hours in room temp or overnight at 4°C. DAPI was included with the secondary antibody. Washes after the primary and secondary antibody were done in 0.25% triton X-100/PBS and antibodies were diluted in IHC buffer. Primary antibodies used were OGT (AL25 (produced in-house), 1:5000), TRH (Santa Cruz Biotechnology, 1:100), c-Fos (ABE457, Millipore, 1:500), Oxytocin (MAB5296, Millipore, 1:1000), GFP (Abcam, 1:1000), in Figs. 4 and 9 the signal from virally expressed GFP alone was imaged without using an antibody against GFP. For some experiments when staining for OGT alone, the animal was not perfused but fixed in 4% PFA/PBS overnight. In situ hybridization. The brain was removed and fresh frozen in OCT and stored at -80°C. Serial sections were cut (25 µm) on a cryostat and stored at -80°C. Probes and hybridization protocol have been described at length previously (26-27). Briefly, slides were fixed in 4% PFA, treated with acetic anhydride and then blocked in pre-hybridization solution (hybridization buffer without probes added: 50% formamide, 5X saline sodium citrate (SSC), 5X Denhardt’s solution, 250ug/ml yeast tRNA, 500ug/ml sperm DNA) before application of respective probe diluted in hybridization buffer. Hybridization was done at 70°C overnight. The next day the slides were washed, re-blocked (0.1M Tris, pH 7.5, 0.15M sodium chloride, 5% sheep serum) and then incubated overnight at 4°C with an alkaline-phosphatase linked anti-digoxigenin antibody (Roche, 1:5000) diluted in blocking buffer. On the third day probes were developed after washing. Probe development was done by applying a mix of NBT (nitro-blue tetrazolium chloride), BCIP (5-bromo-4-chloro-3'-indolyphosphate p-toluidine salt) and levamisole until sufficient stain had precipitated (within the same probe, all slides were treated for the same length of time). Afterwards, slides were DAPI stained. Brain region coordinates. Different brain regions were identified by taking advantage of the Paxinos and Franklin atlas (The Mouse Brain, in stereotaxic coordinates, second edition). Counted as distance from Bregma, the following coordinates were used: SCN (-0.34 to -0.82), PVN (-0.60 to -1.20), Arc (-1.46 to -2.20). VMH (-1.30 to -2.06), LH (-1.34 to -1.75), DMH (-1.46 to -2.06). The PBN and the NTS were identified on sagittal slices, PBN (Bregma: -5.00 to -5.45, lateral: 0.85 to 1.35), NTS (Bregma: -6.30 to -7.80, lateral: 0.36 to 0.96). Abbreviations used: Paraventricular nucleus (PVN),

3

Dorsomedial hypothalamic nucleus (DMH), Arcuate nucleus (Arc), Ventromedial hypothalamus (VMH), Lateral hypothalamus (LH), Nucleus of the solitary tract (NTS), Parabrachial nucleus (PBN). Organotypic cultures Acute coronal slices were cut from male mice around postnatal day 10 on a vibratome. The slices were cut in oxygen-bubbled dissection medium (1 mM CaCl2, 5 mM MgCl2, 10 mM glucose, 4 mM KCl, 26 mM NaHCO3, 218 mM sucrose, 1.2 mM NaH2PO4(H2O), 30 mM HEPES). After cutting, the dorsal half of each slice was removed and then mounted on membrane inserts in a 6 well plate (one slice per insert). Each well was filled with 1 ml slice culture media (SCM: 4.16 g/500ml MEM, 20% horse serum, 1 mM L-glutamine, 0.5 mg/500ml insulin, 2 mM MgCl2, 1 mM CaCl2, 3 mM glucose, 1.3 mM NaHCO3, 3.58 g/500ml HEPES) and changed every 2-3 days. Explants were incubated at 35 °C. On DIV 10, virus was added by direct application on the top of each slice to label αCaMKII-positive cells (UPenn, #AV-1-PV1917 (AAV1.CaMKII0.4.eGFP.WPRE.rBG)). After about 1 week, cultures were submitted to 1 mM glucose SCM for 24 h (media changed once during those 24 h) and then changed into either fresh 1 mM or 5 mM glucose SCM for 1 h. In a second round of experiments, slices were put in 5 mM glucose SCM for 24 h and then transferred to either fresh 5 mM or 25 mM glucose SCM for 16 h. Slices were then fixed in 4% PFA in PBS for 2-7 h. The area around the fixed slices were cut out using a scalpel and then subjected to immunohistochemistry (O-GlcNAc and TRH) as described for free-floating slices. Stereotactic virus injection Anesthesia was induced using avertin (0.03ml/g) and thereafter kept by 1-1.5% isoflurane gas inhalation. The coordinates for the PVN were: distance from Bregma, -0.80 mm; distance from midline, 0.20 mm; depth from surface, -4.75mm. The stereotaxic instrument used was the Leica Angle Two stereotaxic instrument. The scalp was exposed by a small incision through the skin. Bregma and Lambda were identified and two small holes drilled over the injection site. Stored virus was thawed, diluted with sterile saline to a final concentration of 1x109 particles/ul and then spun for 5 min at 3000 rpm in room temperature. The virus was then loaded into a pulled glass pipette (20-40 µm) and connected to a syringe pump. In total, 2ul virus was injected at 0.300 ul/min. After the surgery, the scalp was sutured and the animal rehydrated with saline (0.02 ml/g, intraperitoneal injection) and given buprenorphine to relieve pain (0.01 ml/g, subcutaneous injection). Upon animal harvest, the injection site and spread of the virus was controlled by immunohistochemistry. In total, 5 Wt animals and 8 KO animals were injected. In all but one case, the PVN was successfully targeted. Although there was minor spread outside the PVN, the PVN was the common denominator of areas hit across animals. In the one case when the PVN was hit only partially, the daily food intake and body weight increased but to a lesser degree, as expected (data not shown). This animal was removed from analysis. Excluding animals where the PVN had not been successfully targeted was a pre-established criterion. All details pertaining to optogenetics experiments are described below under Optogenetics. Behavior

4

General feeding behavior. Animals were kept in individual cages after tamoxifen treatment or virus injection. Food intake was measured by weighing leftover food pellets. The bottom of the cage was searched to include all leftover food. For some Wt and KO tamoxifen treated animals, the animals were not separated to individual cages until two weeks after the treatment had finished. Data from these animals were compiled with the rest as the effect on feeding behavior and body weight for all animals followed the same time course. Whereas tamoxifen treated animals were kept on a 14 h/10 h light/dark cycle, the virus injected animals were kept on a 12 h/12 h light/dark cycle. Restricted feeding. 10 days after tamoxifen treatment, the KO animals were split into two groups. One group had continuous free access to food whereas the other group was fed 1 gram at the start of light hours and 2.5 grams at the start of dark hours, totaling 3.5 grams/day. At 21 days, free access to food was reintroduced to the latter group. Animals were kept on a 12 h/12 h light/dark cycle. Comprehensive laboratory animal monitoring system (CLAMS). CLAMS is a closed chamber system that measures ventilation gases and food intake automatically. Our system was also equipped with laser beams that register physical movement. We submitted 7 Wt and 7 KO mice to CLAMS analysis two weeks after tamoxifen treatment. After several days of acclimatization, data were collected for 3 days. Data were binned in 15 min increments. Values that were the obvious result of aberrant mouse behavior or technical failure, such as spillage off the scale, were removed from analysis. There were a few small negative values recorded. These were included in the analysis, except for Fig. 1E for which negative values were removed. The total number of beam breaks is shown as ‘physical activity’ in all figures. A ‘meal’ was defined as any multiplicative of 15 min where the amount of intake the first 15 min was 50 mg or larger and the previous 15 min had no intake registered. A meal was considered finished when no intake had been registered for another 15 min. The formulas used to calculate different endpoints can be found at www.colinst.com. Animals were kept on a 12 h/12 h light/dark cycle. Refeeding. For refeeding experiments in Fig. S8, the food was removed for 24 h and then reintroduced for 2.5 h before animal harvest. Throughout the experiment, the animals had free access to drinking water. For the refeeding paradigm coupled with optogenetics, see Optogenetics. Fat-based diet. Mice were fed either regular chow (3.1 kcal/g, 24% protein, 18% fat, 58% carbohydrate) or energy-dense fat-based food (6.7 kcal/g, 9.1% protein, 90.5% fat, 0.4% carbohydrate). The mice were raised on regular chow and then switched to fat-based diet either about 1 week before injected with tamoxifen or about 14 days after injections had finished. The same behavior was noticed in both regiments and the data were combined. As the data for regular chow mimicked the data in general feeding behavior experiments, all those data were combined also. The data shown are the averages during the period when daily food intake had stabilized (days 20-26 after tamoxifen injection). Optogenetics Animals. All experimental protocols were conducted according to U.S. National Institutes of Health guidelines for animal research and were approved by the Institutional Animal Care and Use Committee at NIDA. Male C57BL/6J mice (2–5 months old, Jackson Laboratories) were used for optogenetic experiments.

5

Stereotactic viral delivery and fiber guide system implantation. Mice were anesthetized with isoflurane and were placed into a stereotaxic apparatus, as previously described (28). A pulled glass pipette with 30-40µm tip diameter was inserted into the brain, and one unilateral injection (30nL) of AAV2/1-CamKIIα-hChR2(H134R)-EYFP virus (titer: 3.3x1012 viral genomes/mL, Optogenetics and Transgenic Technology Core, NIDA) was made at coordinates in the paraventricular nucleus of the hypothalamus (PVN, bregma: -0.70mm, midline: +0.15mm, dorsal surface: -4.60mm). A micromanipulator was used to control the injection speed (30nL/min), and the pipette was withdrawn 15 min after the injection. This was followed by the insertion of a ferrule-capped optical fiber (200 μm diameter core; NA 0.48) through the craniotomy. Metabond cement was used to anchor the ferrule to the skull. The fiber tip was implanted 0.7mm above the PVN. Mice were returned to their home cages for 10 – 14 days to recover and for expression of ChR2:EYFP. Components for food consumption and photostimulation. In home cages, mice had ad libitum access to rodent chow. For behavioral testing, mice were transferred into feeding cages with automatic pellet dispensers and supplied with food pellets (20 mg each) of identical composition to the food in the home cage. Pellet removal was sensed by the offset of a beam break and an additional pellet was administered after a delay (10 s). Food consumption was monitored continuously using Graphic State software (Coulbourn Instruments). Water was also available ad libitum during behavioral experiments, and in some cases, was monitored by optical detection (Coulbourn Instruments) of licks emitted towards the drinking spout. In addition, water bottles were manually weighed to estimate water intake every 24 hours. Mice were allowed to acclimate for 3 days before initiating the photostimulation protocols. Light was delivered to the brain through an optical fiber. The relationship of light scattering and absorption in the brain as a function of distance has been described previously (29). Using this relationship, the light power exiting the fiber tip (10 mW) was estimated to correspond to 2.0 mW/mm2 at the PVN, which was sufficient to drive a behavioral response. For optical delivery of light pulses with millisecond precision to multiple mice, the output beam from a 300 mW diode laser (473 nm) was split into four beams using a combination of 50/50 beam splitters and turning mirrors. Each beam was controlled using an acousto-optic modulator (AOM) to generate light pulses that were launched into separate fiber ports and their corresponding optical fibers. Using these components, four mice could be simultaneously and independently photostimulated. LabVIEW software was used control the AOMs. Behavioral experiments. General feeding behavior. Food intake was recorded during a pre-stimulus (baseline) for 3 days followed by a 24 h photostimulation period. In two experiments, stimulation started during light hours and in one experiment stimulation started at the onset of darkness. In all experiments, the same behavioral response was recorded and the data were averaged over all experiments. The primary photostimulation protocol consisted of 50 light pulses (10 ms) for 1 s at 50 Hz, with the sequence repeated every 4 s for 24 h. Meals were quantified as described for CLAMS. Hunger-induced feeding behavior. Mice were fasted for 9 hours during the light phase (with ad libitum access to water) before testing for evoked feeding behavior at the onset of feeding at the start of the dark phase. The primary photostimulation protocol consisted of 50 or 20 light pulses (10 ms each) which were delivered for 1 s (50 Hz or 20

6

Hz, respectively), and the sequence repeated every 4 seconds for 12 hours. For the no starvation control, we used the average intake during the equivalent hours over 2 days. For meals analysis, food intake was binned over 5 min. A ‘meal’ was defined as any multiplicative of 5 min where the amount of intake the first 5 min was 10 mg or larger and no intake had been registered prior to that period. A meal was considered finished when no intake had been registered for another 10 min. Whole body composition analysis Quantitative NMR to measure total fat and lean mass was done using EchoMRI. Animals were analyzed about three weeks after tamoxifen treatment. Electrophysiology Whole-cell patch-clamp recordings were performed to assess the excitatory synaptic function in OGT KO PVN neurons. For this, we took advantage of αCaMKII-CreER+ x OGTFl/Wt x TdTFl mice. Acute PVN slices were prepared 17-19 days after tamoxifen injection at a similar time of day. Mice were anesthetized with isoflurane and decapitated. Brains were removed rapidly and placed in ice-cold cutting solution containing 210.3 mM sucrose, 2.5 mM KCl, 1 mM NaH2PO4, 26.2 mM NaHCO3, 4 mM MgCl2, 0.5 mM CaCl2 and 11 mM D-(+)-glucose and oxygenated with 95% O2/5% CO2. Coronal slices (300 µm thick) were cut with a vibratome and were kept in ACSF (125 mM NaCl, 2.5 mM KCl, 2 mM MgCl2, 2 mM CaCl2, 1.0 mM NaH2PO4, 26.2 mM NaHCO3 and 11 mM glucose, oxygenated with 95% O2/5% CO2) at 23–25 °C until recordings. Slices were placed in a submerged chamber and perfused with ACSF supplemented with 50 µM picrotoxin and 1 µM TTX. Targeted whole cell recordings of PVN tdTomato-positive neurons were made using pipettes of 2-5 MΩ resistance. The intracellular solution contained 115 mM CsMeSO4, 0.4 mM EGTA, 5.0 mM TEA-Cl, 1 mM QX314, 2.8 mM NaCl, 20 mM HEPES, 3.0 mM MgATP, 0.5 mM GTP and 10 mM sodium phosphocreatine (pH = 7.2 and osmolality of 285–290 mOsm). Cells were held at -70mV holding potential and recording was performed at room temperature. The junction potential was left uncorrected. Signals were measured with a amplifier and digitized using an analog-to-digital board. Data were acquired with pClamp 10 software and digitized at 20 kHz. Cell capacitance was measured. Miniature EPSCs (mEPSCs) were analyzed with MiniAnalysis using a detection threshold of 7 pA (>2 times root mean square noise). Decay times were obtained by single-exponential fitting of the average mEPSC traces. Image analysis Neuronal cell number. In vitro: Cell number was quantified by counting the number of DAPI stained primary cultured neurons. On each of three Wt and KO coverslips, 10 random images were taken on a confocal microscope. In ImageJ, after setting a pre-defined threshold, the number of cells was counted automatically by applying the ‘Analyze particles’ command. In vivo: Cell number was quantified by counting the number of DAPI stained cells within the CA1 region of the hippocampus and the PVN in several slices from 3 Wt and 3 KO animals (confocal microscope). All slices were matched in distance from Bregma between Wt and KO. The number of cells was counted

7

manually in the CA1 region. For the PVN, cells were counted automatically in ImageJ as described above. Neuropeptide transcript expression. Within the PVN, the expression of the following probes was quantified: TRH, AVP, Oxytocin. Within the arcuate nucleus, the expression of the following probes was quantified: POMC, CART, AgRP, NPY. In the lateral hypothalamus the expression of orexin was quantified. Images for quantification were taken on a Nikon Eclipse Ti, except for orexin where an Axiophot was used. Images shown in Fig. 2 and S7 were all taken on the Axiophot. Animals had been harvested in the early phase when KOs had just started to develop hyperphagia without developing any major obesity (about two weeks after tamoxifen treatment). Slices from the same 4 Wt and 4 KO animals were used for all probes, except for Wt CART where slices from 3 of the animals were used. Quantification was done in ImageJ. A pre-defined region of interest (ROI) was placed over each cell. The number of cells per image and the average intensity of each cell were collected. The right and left side of each brain region were imaged separately. Data were averaged per animal. Within the same probe, all sections were matched in distance from Bregma. The investigator was blinded to whether images came from Wt or KO animals during counting. c-Fos and OGT expression in virus-injected PVN. Prior to harvest, αCaMKII-positive cells in the PVN had been labeled with virus expressing GFP under the αCaMKII promoter by stereotaxic injection. Within the PVN, in ImageJ, all c-Fos positive cells per brain slice of the PVN were counted manually and marked as GFP-positive or GFP-negative. All sections were matched in distance from Bregma. Note that the Wt virus GFP expression was higher than the KO virus GFP expression, a result we attributed to the combined GFP and Cre recombinase expression by the KO virus while the Wt virus expressed GFP alone. Therefore, for the GFP channel in KO slices, the microscope gain was set higher to include all GFP positive cells (confocal microscope). O-GlcNAc expression in organotypic slices and starved mice. All image analysis was done on raw, unmanipulated images in ImageJ. Cells and background were marked with a pre-defined ROI. For TdT/GFP positive cells, the ROI was placed over positive cells in the TdT/GFP channel. For TdT/GFP negative cells, cells were picked randomly in the DAPI channel and then verified as TdT/GFP negative in the TdT/GFP channel. If the ROI marked a TdT/GFP positive cell, the ROI was moved to a cell nearby. The background in each image was quantified by measuring the average intensity of three ROIs placed over cell-empty areas as judged by no DAPI signal. After placing all ROIs, the mean intensity was measured in the O-GlcNAc channel. The background value was subtracted from each cell value. The experimenter was blinded to the O-GlcNAc channel when placing the ROIs. Image presentation. All image analysis was done on raw, unmanipulated images. For presentation purposes, most shown immunohistochemistry images and DAPI stained images were improved post hoc. All manipulations were applied equally within the same experiment and to Wt and KO images. The manipulations were the following: change of Levels and Brightness/Contrast and Filtering (‘Despeckle’ and ‘Gaussian blur’ (2.0pixels)). Statistical analysis

8

Statistical analyses were done using the program Prism 5. Repeated-measures two-way analysis of variance (ANOVA) analyses detected significant interaction of ‘time’ and ‘genotype’, and post hoc Bonferroni analyses detected statistical significance between individual time points. In Fig. S3A, one-way ANOVA was applied to calculate significant interaction between all groups and then the difference between all separate group combinations was tested post hoc with Tukey’s Multiple Comparison Test. For in situ hybridization, statistical analysis compared between animals mean data averaged over all imaged slices. All Student’s t-tests were two-tailed and unpaired. Star, ‘*’, signifies P<0.05. All error bars represent mean +/- standard error of the mean (s.e.m.). A few animals where either some food intake or body weight value had been missed were omitted in repeated-measures two-way ANOVA analyses and post-hoc Bonferroni analyses. However, these values did not change the overall result and were included in shown graphs. The normality distribution of data used for the F-test in Fig. S6C was verified using the D'Agostino-Pearson omnibus normality test.

9

Fig. S1. αCaMKII-CreER x OGTFl deletes OGT in αCaMKII-positive neurons of adult mice but does not affect OGT expression in peripheral organs. (A) Schematic of the Ogt locus, including loxP sites (triangles) in OGTFl mice and translational start site (empty circle). (B) PCR analysis showing the genotype of Wt (αCaMKII-CreER- x OGTFl) and KO (αCaMKII-CreER+ x OGTFl) mice. (C) Immunohistochemistry for OGT. Top: neocortex. Bottom: the hippocampus. (D) Primary cultured OGTFl neurons were infected with either a wildtype (Wt) virus or a virus expressing Cre recombinase (KO). Western blot characterizing the O-GlcNAc and OGT expression in Wt and KO neurons. (E) Western blot characterizing O-GlcNAc and OGT expression in peripheral tissues. (F) Quantification of OGT expression in peripheral tissues from (E) (Wt n=7, KO n=8; two-tailed t-test: P<0.05). (G) PCR analysis detecting Cre-dependent Ogt recombination in the brain but not in the liver. Quantifications represent mean +/- s.e.m.

10

Fig. S2. Deletion of OGT in postmitotic neurons does not affect cell number. (A) Left: DAPI stain of primary cultured neurons that were infected with either a wildtype virus (Wt) or a virus expressing Cre recombinase (KO). Right: quantification of DAPI stained cells (Wt n=3, KO n=3; two-tailed t-test: P<0.05). (B) Left: DAPI stain of the CA1 region of the hippocampus, where almost all cells express αCaMKII. Right: quantification of DAPI stained cells (Wt n=3, KO n=3; two-tailed t-test: P<0.05). Quantifications represent mean +/- s.e.m.

11

Fig. S3. Loss of OGT leads to adiposity, hyperphagia and hyperactivity. (A-B) Induction of Cre recombinase by tamoxifen treatment leads to weight gain only in the presence of floxed Ogt. (A) Quantification of body weight; tamoxifen treatment alone does not affect body weight (n=3 for all groups; one-way ANOVA with post hoc Tukey’s Multiple Comparison Test: P<0.05). (B) Quantification of body weight; αCaMKII-CreER crossed with an unrelated floxed mouse and treated with tamoxifen does not affect body weight (Wt n=5, KO n=5; two-tailed t-test: P<0.05). (C-D) Quantitative NMR of total body fat (C) and total lean body mass (D) (Wt n=7, KO n=6; two-tailed t-test: P<0.05). (E-G) Quantification of amount fat in different regions the body. (E) Intraperitoneal fat. (F) Subcutaneous fat. (G) Total liver weight. (For all regions: Wt n=3, KO n=3; two-tailed t-test: P<0.05). (H-M): Data from CLAMS analysis; (H) Daily food intake. (I) Total energy expenditure as measured by exchange of respiratory gases, see Methods. (J) Physical activity. (K) Volume O2 inhaled. (L) Volume CO2 exhaled. (M) The respiratory exchange ratio (RER). (H-M: Wt n=17, KO n=20; two-tailed t-test: P<0.05). Quantifications represent mean +/- s.e.m.

12

Fig. S4. Diurnal rhythm and meal frequency are not perturbed in OGT KO mice while meal size and duration are increased. The OGT-dependent hyperphagia correlates with total caloric intake. (A) Quantification of total activity during CLAMS as dispersed over dark and light hours (Wt n=21, KO n=21; two-tailed t-test: P<0.05). (B) Quantification of food intake during CLAMS as dispersed over dark and light hours (Wt n=17, KO n=20; two-tailed t-test: P<0.05). (C) Quantification of the number of initiated meals during CLAMS (Wt n=17, KO n=20; two-tailed t-test: P<0.05). (D-E) Quantification of meal size (D) and meal length (E) during CLAMS (Wt n=268, KO n=330; two-tailed t-test: P<0.05). (F) Quantification of food intake as it differs between carbohydrate-based pellets and fat-based paste (KD). (Chow (3.1 kcal/g): Wt n=10, KO n=10; KD (6.7 kcal/g): Wt n=7, KO n=5). Quantifications represent mean +/- s.e.m.

13

Fig. S5. αCaMKII-CreER x OGTFl deletes OGT from several core feeding nuclei. (A) For each panel: left; immunohistochemistry for OGT in αCaMKII-CreER x OGTFl mice. Arrowheads indicate cells that lack OGT. Right; schematics showing general anatomical landmarks for slices that were used for imaging. The areas within each slice containing purple dots were the areas that were imaged and used for analysis. Purple dots within the schematics indicate cells that lost OGT in KO mice. Abbreviations used: Paraventricular nucleus (PVN), Dorsomedial hypothalamic nucleus (DMH), Arcuate nucleus (Arc), Ventromedial hypothalamus (VMH), Lateral hypothalamus (LH), Nucleus of the solitary tract (NTS), Parabrachial nucleus (PBN).

14

Fig. S6 αCaMKII-CreER x OGTFl deletes OGT from a subpopulation of cells in the PVN with no effect on total cell number in the PVN. (A) αCaMKII-positive cells (purple) were labeled using αCaMKII-CreER+ x TdTFl mice. Left: DAPI (green). Middle: immunohistochemistry for TRH (green). Right: immunohistochemistry for Oxytocin (green). Arrow; cell positive for both the cell-specific marker and TdT. Arrowhead; cell positive for the cell-specific marker but not TdT. (B) Double-labeling of TRH cells and αCaMKII cells. The PVN was injected stereotactically with Wt or KO virus. TRH cells were labeled with immunohistochemistry for TRH (purple) and αCaMKII cells with virus expressing GFP under the αCaMKII promoter (green). Arrow; cell positive for both TRH and GFP. Arrowhead; cell positive for TRH but not GFP. (C) Serial sections of the PVN were cut and subjected to in situ hybridization for TRH as shown in Fig. 2B. Shown is the quantification of the number of TRH positive cells per PVN slice. The distribution of TRH cells in Wts and KOs showed statistical un-equal variance, indicating that a particular subpopulation of TRH cells is affected by OGT deletion (Wt n=45, KO n=48; F-test: P<0.05). (D) The same data as in (C) but plotted is the number of TRH positive cells per PVN slice in relation to distance from Bregma. (E) Left: DAPI stain of the PVN. Right: quantification of DAPI stained cells (Wt n=3, KO n=3; two-tailed t-test: P<0.05). Quantifications represent mean +/- s.e.m.

15

Fig. S7. Deletion of OGT in αCaMKII-positive neurons leaves several major feeding circuits intact. (A) In situ hybridization for neuropeptide modulators in core feeding circuits two weeks after OGT KO prior to significant weight increase. For each peptide: left; image of probe staining, right; quantification of number of stained cells within the respective area (cells / 150 x 103 µm2) and quantification of intensity of staining per cell (a.u.) (Wt n=4 (except for CART, and number of oxytocin positive cells, where n=3), KO n=4; two-tailed t-test: P<0.05). Quantifications represent mean +/- s.e.m.

16

Fig. S8. αCaMKII-positive PVN neurons are uniquely sensitive to physiological variations in glucose in terms of O-GlcNAc and deletion of OGT blunts feeding-induced activation of the same cells. (A-D) Hypothalamic explant with GFP expression (green) and immunohistochemistry for TRH (A) or O-GlcNAc (C) (purple). Boxed magnification in (A) shows the PVN. (B) Quantification of O-GlcNAc intensity in αCaMKII-negative cells in the PVN from Fig. 2C (1 mM n=360, 5 mM n=330; two-tailed t-test: P<0.05). (D) Quantification of O-GlcNAc intensity in αCaMKII-positive cells in the PVN from (C) (5 mM n=390, 25 mM n=375; two-tailed t-test: P<0.05). (E) O-GlcNAc expression in the PVN in fed and starved animals (αCaMKII-CreER+ x TdTFl mice). Left; double-labeling of TdT expression (purple) and immunohistochemistry for O-GlcNAc (green). Arrow; cell positive for TdT. Arrowhead; cell negative for TdT. Right; quantification of O-GlcNAc expression in TdT-positive and TdT-negative cells (TdT-positive: fed n=455, starved n=259. TdT-negative: fed n=357, starved n=375; two-tailed t-test: P<0.05). (F) The PVN was injected stereotactically with Wt or KO virus. Double-labeling of GFP expression (green) and c-Fos (orange) in the PVN in fed and refed animals. Arrow; cell positive for both c-Fos and GFP. Arrowhead; cell positive for c-Fos but not GFP. (G) Quantification of number of cells expressing both GFP and c-Fos from (F) (Fed: Wt n=8, KO n=7. Refed: Wt n=12, KO n=10; two-tailed t-test: P<0.05). Quantifications represent mean +/- s.e.m.

17

Fig. S9. OGT KO does not affect membrane capacitance or EPSC shape in αCaMKII-positive cells in the PVN and optogenetic stimulation of the same cells decreases food intake. (A-C) Patch-clamp analysis of αCaMKII-positive cells in the PVN (Wt n=6, KO n=6; two-tailed t-test: P<0.05). (A) Membrane capacitance. (B) mEPSC peak rise time. (C) mEPSC peak decay tau. (D) GFP expression (green) and immunohistochemistry for OGT (purple) in the PVN after stereotactic injection of Wt or KO virus. Arrow; cell positive for OGT and GFP. Arrowhead; cell positive for OGT only. (E-H) Optogenetic stimulation of αCaMKII-positive cells in the PVN. (E) Quantification of meal frequency (baseline n=11, stimulation n=11; two-tailed t-test: P<0.05). (F) Quantification of food intake after a light fast. Bars represent average intake over all mice and lines represent average intake per individual mouse (All conditions n=5; two-tailed t-test: '*' refers to P<0.05). (G) Quantification of the size of the first meal after a light fast, 50 Hz stimulation (baseline n=5, stimulation n=5; two-tailed t-test: P<0.05). (H) Quantification of the latency to the initiation of the first meal after a light fast, 50 Hz stimulation (baseline n=5, stimulation n=5; two-tailed t-test: P<0.05). Quantifications represent mean +/- s.e.m.

18

References and Notes 1. C. E. Lewis, K. M. McTigue, L. E. Burke, P. Poirier, R. H. Eckel, B. V. Howard, D. B.

Allison, S. Kumanyika, F. X. Pi-Sunyer, Mortality, health outcomes, and body mass index in the overweight range: A science advisory from the American Heart Association. Circulation 119, 3263–3271 (2009). Medline doi:10.1161/CIRCULATIONAHA.109.192574

2. E. K. Speliotes, C. J. Willer, S. I. Berndt, K. L. Monda, G. Thorleifsson, A. U. Jackson, H. Lango Allen, C. M. Lindgren, J. Luan, R. Mägi, J. C. Randall, S. Vedantam, T. W. Winkler, L. Qi, T. Workalemahu, I. M. Heid, V. Steinthorsdottir, H. M. Stringham, M. N. Weedon, E. Wheeler, A. R. Wood, T. Ferreira, R. J. Weyant, A. V. Segrè, K. Estrada, L. Liang, J. Nemesh, J. H. Park, S. Gustafsson, T. O. Kilpeläinen, J. Yang, N. Bouatia-Naji, T. Esko, M. F. Feitosa, Z. Kutalik, M. Mangino, S. Raychaudhuri, A. Scherag, A. V. Smith, R. Welch, J. H. Zhao, K. K. Aben, D. M. Absher, N. Amin, A. L. Dixon, E. Fisher, N. L. Glazer, M. E. Goddard, N. L. Heard-Costa, V. Hoesel, J. J. Hottenga, A. Johansson, T. Johnson, S. Ketkar, C. Lamina, S. Li, M. F. Moffatt, R. H. Myers, N. Narisu, J. R. Perry, M. J. Peters, M. Preuss, S. Ripatti, F. Rivadeneira, C. Sandholt, L. J. Scott, N. J. Timpson, J. P. Tyrer, S. van Wingerden, R. M. Watanabe, C. C. White, F. Wiklund, C. Barlassina, D. I. Chasman, M. N. Cooper, J. O. Jansson, R. W. Lawrence, N. Pellikka, I. Prokopenko, J. Shi, E. Thiering, H. Alavere, M. T. Alibrandi, P. Almgren, A. M. Arnold, T. Aspelund, L. D. Atwood, B. Balkau, A. J. Balmforth, A. J. Bennett, Y. Ben-Shlomo, R. N. Bergman, S. Bergmann, H. Biebermann, A. I. Blakemore, T. Boes, L. L. Bonnycastle, S. R. Bornstein, M. J. Brown, T. A. Buchanan, F. Busonero, H. Campbell, F. P. Cappuccio, C. Cavalcanti-Proença, Y. D. Chen, C. M. Chen, P. S. Chines, R. Clarke, L. Coin, J. Connell, I. N. Day, M. den Heijer, J. Duan, S. Ebrahim, P. Elliott, R. Elosua, G. Eiriksdottir, M. R. Erdos, J. G. Eriksson, M. F. Facheris, S. B. Felix, P. Fischer-Posovszky, A. R. Folsom, N. Friedrich, N. B. Freimer, M. Fu, S. Gaget, P. V. Gejman, E. J. Geus, C. Gieger, A. P. Gjesing, A. Goel, P. Goyette, H. Grallert, J. Grässler, D. M. Greenawalt, C. J. Groves, V. Gudnason, C. Guiducci, A. L. Hartikainen, N. Hassanali, A. S. Hall, A. S. Havulinna, C. Hayward, A. C. Heath, C. Hengstenberg, A. A. Hicks, A. Hinney, A. Hofman, G. Homuth, J. Hui, W. Igl, C. Iribarren, B. Isomaa, K. B. Jacobs, I. Jarick, E. Jewell, U. John, T. Jørgensen, P. Jousilahti, A. Jula, M. Kaakinen, E. Kajantie, L. M. Kaplan, S. Kathiresan, J. Kettunen, L. Kinnunen, J. W. Knowles, I. Kolcic, I. R. König, S. Koskinen, P. Kovacs, J. Kuusisto, P. Kraft, K. Kvaløy, J. Laitinen, O. Lantieri, C. Lanzani, L. J. Launer, C. Lecoeur, T. Lehtimäki, G. Lettre, J. Liu, M. L. Lokki, M. Lorentzon, R. N. Luben, B. Ludwig, P. Manunta, D. Marek, M. Marre, N. G. Martin, W. L. McArdle, A. McCarthy, B. McKnight, T. Meitinger, O. Melander, D. Meyre, K. Midthjell, G. W. Montgomery, M. A. Morken, A. P. Morris, R. Mulic, J. S. Ngwa, M. Nelis, M. J. Neville, D. R. Nyholt, C. J. O’Donnell, S. O’Rahilly, K. K. Ong, B. Oostra, G. Paré, A. N. Parker, M. Perola, I. Pichler, K. H. Pietiläinen, C. G. Platou, O. Polasek, A. Pouta, S. Rafelt, O. Raitakari, N. W. Rayner, M. Ridderstråle, W. Rief, A. Ruokonen, N. R. Robertson, P. Rzehak, V. Salomaa, A. R. Sanders, M. S. Sandhu, S. Sanna, J. Saramies, M. J. Savolainen, S. Scherag, S. Schipf, S. Schreiber, H. Schunkert, K. Silander, J. Sinisalo, D. S. Siscovick, J. H. Smit, N. Soranzo, U. Sovio, J. Stephens, I. Surakka, A. J. Swift, M. L. Tammesoo, J. C. Tardif, M. Teder-Laving, T. M. Teslovich, J. R. Thompson, B. Thomson, A. Tönjes, T. Tuomi, J. B. van Meurs, G. J. van Ommen,

V. Vatin, J. Viikari, S. Visvikis-Siest, V. Vitart, C. I. Vogel, B. F. Voight, L. L. Waite, H. Wallaschofski, G. B. Walters, E. Widen, S. Wiegand, S. H. Wild, G. Willemsen, D. R. Witte, J. C. Witteman, J. Xu, Q. Zhang, L. Zgaga, A. Ziegler, P. Zitting, J. P. Beilby, I. S. Farooqi, J. Hebebrand, H. V. Huikuri, A. L. James, M. Kähönen, D. F. Levinson, F. Macciardi, M. S. Nieminen, C. Ohlsson, L. J. Palmer, P. M. Ridker, M. Stumvoll, J. S. Beckmann, H. Boeing, E. Boerwinkle, D. I. Boomsma, M. J. Caulfield, S. J. Chanock, F. S. Collins, L. A. Cupples, G. D. Smith, J. Erdmann, P. Froguel, H. Grönberg, U. Gyllensten, P. Hall, T. Hansen, T. B. Harris, A. T. Hattersley, R. B. Hayes, J. Heinrich, F. B. Hu, K. Hveem, T. Illig, M. R. Jarvelin, J. Kaprio, F. Karpe, K. T. Khaw, L. A. Kiemeney, H. Krude, M. Laakso, D. A. Lawlor, A. Metspalu, P. B. Munroe, W. H. Ouwehand, O. Pedersen, B. W. Penninx, A. Peters, P. P. Pramstaller, T. Quertermous, T. Reinehr, A. Rissanen, I. Rudan, N. J. Samani, P. E. Schwarz, A. R. Shuldiner, T. D. Spector, J. Tuomilehto, M. Uda, A. Uitterlinden, T. T. Valle, M. Wabitsch, G. Waeber, N. J. Wareham, H. Watkins, J. F. Wilson, A. F. Wright, M. C. Zillikens, N. Chatterjee, S. A. McCarroll, S. Purcell, E. E. Schadt, P. M. Visscher, T. L. Assimes, I. B. Borecki, P. Deloukas, C. S. Fox, L. C. Groop, T. Haritunians, D. J. Hunter, R. C. Kaplan, K. L. Mohlke, J. R. O’Connell, L. Peltonen, D. Schlessinger, D. P. Strachan, C. M. van Duijn, H. E. Wichmann, T. M. Frayling, U. Thorsteinsdottir, G. R. Abecasis, I. Barroso, M. Boehnke, K. Stefansson, K. E. North, M. I. McCarthy, J. N. Hirschhorn, E. Ingelsson, R. J. Loos, MAGIC, Procardis Consortium, Association analyses of 249,796 individuals reveal 18 new loci associated with body mass index. Nat. Genet. 42, 937–948 (2010). Medline doi:10.1038/ng.686

3. R. Gutierrez-Aguilar, D. H. Kim, S. C. Woods, R. J. Seeley, Expression of new loci associated with obesity in diet-induced obese rats: From genetics to physiology. Obesity (Silver Spring) 20, 306–312 (2012). Medline doi:10.1038/oby.2011.236

4. H. Wolosker, D. Kline, Y. Bian, S. Blackshaw, A. M. Cameron, T. J. Fralich, R. L. Schnaar, S. H. Snyder, Molecularly cloned mammalian glucosamine-6-phosphate deaminase localizes to transporting epithelium and lacks oscillin activity. FASEB J. 12, 91–99 (1998). Medline

5. G. W. Hart, C. Slawson, G. Ramirez-Correa, O. Lagerlöf, Cross talk between O-GlcNAcylation and phosphorylation: Roles in signaling, transcription, and chronic disease. Annu. Rev. Biochem. 80, 825–858 (2011). Medline doi:10.1146/annurev-biochem-060608-102511

6. X. Li, F. Lu, J. Z. Wang, C. X. Gong, Concurrent alterations of O-GlcNAcylation and phosphorylation of tau in mouse brains during fasting. Eur. J. Neurosci. 23, 2078–2086 (2006). Medline doi:10.1111/j.1460-9568.2006.04735.x

7. H. B. Ruan, M. O. Dietrich, Z. W. Liu, M. R. Zimmer, M. D. Li, J. P. Singh, K. Zhang, R. Yin, J. Wu, T. L. Horvath, X. Yang, O-GlcNAc transferase enables AgRP neurons to suppress browning of white fat. Cell 159, 306–317 (2014). Medline doi:10.1016/j.cell.2014.09.010

8. N. O’Donnell, N. E. Zachara, G. W. Hart, J. D. Marth, Ogt-dependent X-chromosome-linked protein glycosylation is a requisite modification in somatic cell function and embryo viability. Mol. Cell. Biol. 24, 1680–1690 (2004). Medline doi:10.1128/MCB.24.4.1680-1690.2004

9. J. E. Rexach, P. M. Clark, D. E. Mason, R. L. Neve, E. C. Peters, L. C. Hsieh-Wilson, Dynamic O-GlcNAc modification regulates CREB-mediated gene expression and memory formation. Nat. Chem. Biol. 8, 253–261 (2012). Medline doi:10.1038/nchembio.770

10. C. Slawson, N. E. Zachara, K. Vosseller, W. D. Cheung, M. D. Lane, G. W. Hart, Perturbations in O-linked β-N-acetylglucosamine protein modification cause severe defects in mitotic progression and cytokinesis. J. Biol. Chem. 280, 32944–32956 (2005). Medline doi:10.1074/jbc.M503396200

11. R. Shafi, S. P. Iyer, L. G. Ellies, N. O’Donnell, K. W. Marek, D. Chui, G. W. Hart, J. D. Marth, The O-GlcNAc transferase gene resides on the X chromosome and is essential for embryonic stem cell viability and mouse ontogeny. Proc. Natl. Acad. Sci. U.S.A. 97, 5735–5739 (2000). Medline doi:10.1073/pnas.100471497

12. E. Vijayan, S. M. McCann, Suppression of feeding and drinking activity in rats following intraventricular injection of thyrotropin releasing hormone (TRH). Endocrinology 100, 1727–1730 (1977). Medline doi:10.1210/endo-100-6-1727

13. J. W. Sohn, J. K. Elmquist, K. W. Williams, Neuronal circuits that regulate feeding behavior and metabolism. Trends Neurosci. 36, 504–512 (2013). Medline doi:10.1016/j.tins.2013.05.003

14. A. B. Steffens, A. J. Scheurink, D. Porte Jr., S. C. Woods, Penetration of peripheral glucose and insulin into cerebrospinal fluid in rats. Am. J. Physiol. 255, R200–R204 (1988). Medline

15. G. Pekkurnaz, J. C. Trinidad, X. Wang, D. Kong, T. L. Schwarz, Glucose regulates mitochondrial motility via Milton modification by O-GlcNAc transferase. Cell 158, 54–68 (2014). Medline doi:10.1016/j.cell.2014.06.007

16. M. Karnani, D. Burdakov, Multiple hypothalamic circuits sense and regulate glucose levels. Am. J. Physiol. Regul. Integr. Comp. Physiol. 300, R47–R55 (2011). Medline doi:10.1152/ajpregu.00527.2010

17. J. C. Trinidad, D. T. Barkan, B. F. Gulledge, A. Thalhammer, A. Sali, R. Schoepfer, A. L. Burlingame, Global identification and characterization of both O-GlcNAcylation and phosphorylation at the murine synapse. Mol. Cell. Proteomics 11, 215–229 (2012). Medline doi:10.1074/mcp.O112.018366

18. S. R. Hettes, J. Gonzaga, T. W. Heyming, S. Perez, S. Wolfsohn, B. G. Stanley, Dual roles in feeding for AMPA/kainate receptors: Receptor activation or inactivation within distinct hypothalamic regions elicits feeding behavior. Brain Res. 992, 167–178 (2003). Medline doi:10.1016/j.brainres.2003.08.032

19. A. K. Sutton, H. Pei, K. H. Burnett, M. G. Myers Jr., C. J. Rhodes, D. P. Olson, Control of food intake and energy expenditure by Nos1 neurons of the paraventricular hypothalamus. J. Neurosci. 34, 15306–15318 (2014). Medline doi:10.1523/JNEUROSCI.0226-14.2014

20. Y. Xu, Z. Wu, H. Sun, Y. Zhu, E. R. Kim, B. B. Lowell, B. R. Arenkiel, Y. Xu, Q. Tong, Glutamate mediates the function of melanocortin receptor 4 on Sim1 neurons in body

weight regulation. Cell Metab. 18, 860–870 (2013). Medline doi:10.1016/j.cmet.2013.11.003

21. A. Wallén-Mackenzie, K. Nordenankar, K. Fejgin, M. C. Lagerström, L. Emilsson, R. Fredriksson, C. Wass, D. Andersson, E. Egecioglu, M. Andersson, J. Strandberg, O. Lindhe, H. B. Schiöth, K. Chergui, E. Hanse, B. Långström, A. Fredriksson, L. Svensson, E. Roman, K. Kullander, Restricted cortical and amygdaloid removal of vesicular glutamate transporter 2 in preadolescent mice impacts dopaminergic activity and neuronal circuitry of higher brain function. J. Neurosci. 29, 2238–2251 (2009). Medline doi:10.1523/JNEUROSCI.5851-08.2009

22. American Diabetes Association, Postprandial blood glucose. Diabetes Care 24, 775–778 (2001). Medline doi:10.2337/diacare.24.4.775

23. S. Marshall, O. Nadeau, K. Yamasaki, Dynamic actions of glucose and glucosamine on hexosamine biosynthesis in isolated adipocytes: Differential effects on glucosamine 6-phosphate, UDP-N-acetylglucosamine, and ATP levels. J. Biol. Chem. 279, 35313–35319 (2004). Medline doi:10.1074/jbc.M404133200

24. G. Erdmann, G. Schütz, S. Berger, Inducible gene inactivation in neurons of the adult mouse forebrain. BMC Neurosci. 8, 63 (2007). Medline doi:10.1186/1471-2202-8-63

25. K. Takamiya, L. Mao, R. L. Huganir, D. J. Linden, The glutamate receptor-interacting protein family of GluR2-binding proteins is required for long-term synaptic depression expression in cerebellar Purkinje cells. J. Neurosci. 28, 5752–5755 (2008). Medline doi:10.1523/JNEUROSCI.0654-08.2008

26. T. Shimogori, D. A. Lee, A. Miranda-Angulo, Y. Yang, H. Wang, L. Jiang, A. C. Yoshida, A. Kataoka, H. Mashiko, M. Avetisyan, L. Qi, J. Qian, S. Blackshaw, A genomic atlas of mouse hypothalamic development. Nat. Neurosci. 13, 767–775 (2010). Medline doi:10.1038/nn.2545

27. S. Blackshaw, High-throughput RNA in situ hybridization in mouse retina. Methods Mol. Biol. 935, 215–226 (2013). Medline doi:10.1007/978-1-62703-080-9_15

28. Y. Aponte, D. Atasoy, S. M. Sternson, AGRP neurons are sufficient to orchestrate feeding behavior rapidly and without training. Nat. Neurosci. 14, 351–355 (2011). Medline doi:10.1038/nn.2739

29. A. M. Aravanis, L. P. Wang, F. Zhang, L. A. Meltzer, M. Z. Mogri, M. B. Schneider, K. Deisseroth, An optical neural interface: In vivo control of rodent motor cortex with integrated fiberoptic and optogenetic technology. J. Neural Eng. 4, S143–S156 (2007). 10.1088/1741-2560/4/3/S02 Medline doi:10.1088/1741-2560/4/3/S02