Supporting Information Liedtke et al. 10.1073/pnas.1109199108 SI Results In regards to gene set enrichment analysis, there was no signifi- cantly enriched set detected in the remainder of the brain (i.e., 0 of the 54 gene sets queried showed a P value < 0.1). Specifically, the most significantly enriched set was neuroactive ligand receptor interaction, comprising 75 genes. Its normalized enrichment score was −1.26, indicating some moderate enrichment, but its nominal P value was 0.192. The second gene set, MAPK pathway, consists of 65 genes, and it had a normalized enrichment score of −0.938 and a nominal P value of 0.54. All addiction-related sets were significantly lower in ranking. It has to be considered that other locations within the central nervous system, such as in the ventral tegmentum, amygdala, and medial prefrontal cortex, probably have important involvement in sodium appetite but are diluted out by the large noninvolved brain areas when using the re- mainder of the brain as a control for hypothalamus. With regard to sodium appetite and modulation of emotion and behavior, an interesting set of findings has been made by Morris et al. (1), Na et al. (2), Morris et al. (3), Na et al. (4), Grippo et al. (5), and Morris et al. (6). They showed that nongratified salt appetite is associated with anhedonia, a key component of major depression in humans (1–6). This particular impact of sodium appetite on mood is paralleled by the depressed mood in addicted humans and experimental animals on drug withdrawal. The serum- and glucocorticoid-dependent kinase SGK1, which was characterized by Wärntges et al. (7) as expressed in various brain areas and found up-regulated in its expression by de- hydration, had a tendency to up-regulation in our microarrays. By comparing normalized gene expression for SGK1 for control vs. furosemide-depleted hypothalamus samples, the P value was 0.006 for the isolated comparison, but taking into account (as for all data shown here) the context of the genome-wide analysis, the P value was 0.31. With these data, SGK1 is more likely to be regulated than not regulated, and it ranked 173 for genes ex- clusively regulated in furosemide/Na + depletion-induced sodium appetite. Aforementioned up-regulation of SGK1 in the brain (7) did not include study of the hypothalamus, and it was con- ducted in response to dehydration, not in response to conditions that evoke sodium appetite. In regards to the functional role of SGK1 for mineral appetite, Sgk1 was found necessary for mineralocorticoid-induced sodium appetite evoked by desoxy- corticosterone acetate (DOCA) in Sgk1 null mice (7, 8). This concept is not contradicted by our data. A gene that is necessary for a certain homeostatic behavior need not necessarily be reg- ulated by the cue that evokes the respective behavior. Another interesting aspect is dependence of sodium appetite on peripheral afferents. Complete peripheral denervation abro- gates sodium appetite, which was shown decades ago. Whether dopaminergic transmission is of critical relevance in the peripheral taste system for sodium taste is currently unknown, representing an interesting subject for future studies. However, we wish to stress that we did identify one critical local relay for sodium appetite in the lateral hypothalamus by our injection studies. SI Materials and Methods DNA Microarrays. Mouse oligonucleotide arrays were printed at the Duke Microarray Facility using the Operon’s Mouse Genome Oligo Set (3.0), which contains 31,769 70-mer probes repre- senting 24,878 genes and 32,829 transcripts (Operon). Arrays were printed using an Omnigrid 300 arrayer by Genomic Sol- utions on Corning UltraGap amino-saline coated glass slides (Fisher Scientific). RNA and Microarray Probe Preparation and Hybridization. Total RNA (1 μg) from each sample (mouse hypothalamus or remainder of the brain) and the reference (Universal Mouse Reference RNA; Stratagene) were used in probe preparation. The purified anti- sense RNA was fluorescently labeled with Cy3 (reference) and Cy5 (sample; GE Healthcare). Sample and reference RNAs were pooled, mixed with 1× hybridization buffer (50% formamide, 5× SSC, and 0.1% SDS), COT-1 DNA, and poly-dA to limit non- specific binding, and heated to 95 °C for 2 min. This mixture was pipetted onto a microarray slide using a Maui Mixer hybridization chamber (BioMicro Systems) and hybridized overnight at 42 °C on an automated Maui Hybridization Station (BioMicro Systems). The array was then washed at increasing stringencies and scanned on a GenePix 4000B microarray scanner (Axon Instruments). Detailed protocols are available on the Duke Microarray Facility website (http://www.genome.duke.edu/cores/microarray/services/ spotted-arrays/protocols/). Microarray Analysis. The resulting microarray data were analyzed using the Duke Microarray Database analysis platform, a locally installed version of the Stanford Microarray Database (9, 10). Background-subtracted and normalized data were averaged for each of greater than or equal to four animals within the three groups. Regulated genes were identified by comparison of nor- malized values. Within-array Loess normalization was performed using the Limma package (11) to remove any bias that the intensity has on log-fold change. Next, normalization was carried out between arrays using variance stabilization in that both the red and green channels on the arrays are aligned to each other based on the assumption that most of the genes are constant across experiments (12). Spots not reaching the quality criterion were removed from additional analysis. Quality was determined in three ways. (i ) There was insufficient red or green intensity to produce a reliable ratio. (ii ) The consistency of the ratio throughout the spot was low. (iii ) Other machine-annotated flags indicated that the spot was not reliably read. After filtering, if a particular spot was found only to be present on a small number of arrays (<15), then that spot was filtered for all arrays. To be sure that the experiment consistently measured differences between groups, we performed a principle component analysis. This procedure allowed us to separate the experiments by both brain region (hypothalamus vs. whole brain without hypothalamus] and treatment [adrenocorticotropic hor- mone (ACTH)-induced sodium appetite, sodium deprivation plus furosemide, and control). Spots passing these criteria were ranked and analyzed for statistically significant differences using a linear model (13). Gene Set Enrichment Analysis. Gene set enrichment analysis (GSEA) (14) was used to test the hypothesis that addiction-related genes were regulated by sodium appetite. GSEA takes as input a set of microarray data, a set of comparisons to make between groups in that experiment, and a series of gene sets. Each of those gene sets, often pathways, represents a hypothesis of which genes are regulated in the data. The function of GSEA is to test each of those hypotheses against the data in a statistically rig- orous manner. The processed and normalized microarray data were altered by hand to conform to the .gct format. After being imported into the GSEA, a comparison was made, using a t test, between the expression of genes under the control conditions and the expression of genes in mice with sodium appetite. We tested these gene expression changes against lists of genes that Liedtke et al. www.pnas.org/cgi/content/short/1109199108 1 of 6
Transcript
Supporting InformationLiedtke et al. 10.1073/pnas.1109199108SI ResultsIn regards to gene set enrichment analysis, there was no signifi-cantly enriched set detected in the remainder of the brain (i.e., 0 ofthe 54 gene sets queried showed a P value < 0.1). Specifically, themost significantly enriched set was neuroactive ligand receptorinteraction, comprising 75 genes. Its normalized enrichment scorewas −1.26, indicating some moderate enrichment, but its nominalP value was 0.192. The second gene set, MAPK pathway, consistsof 65 genes, and it had a normalized enrichment score of −0.938and a nominal P value of 0.54. All addiction-related sets weresignificantly lower in ranking. It has to be considered that otherlocations within the central nervous system, such as in the ventraltegmentum, amygdala, and medial prefrontal cortex, probablyhave important involvement in sodium appetite but are dilutedout by the large noninvolved brain areas when using the re-mainder of the brain as a control for hypothalamus.With regard to sodium appetite andmodulation of emotion and
behavior, an interesting set of findings has been made by Morriset al. (1), Na et al. (2), Morris et al. (3), Na et al. (4), Grippoet al. (5), and Morris et al. (6). They showed that nongratifiedsalt appetite is associated with anhedonia, a key component ofmajor depression in humans (1–6). This particular impact ofsodium appetite on mood is paralleled by the depressed mood inaddicted humans and experimental animals on drug withdrawal.The serum- and glucocorticoid-dependent kinase SGK1, which
was characterized by Wärntges et al. (7) as expressed in variousbrain areas and found up-regulated in its expression by de-hydration, had a tendency to up-regulation in our microarrays.By comparing normalized gene expression for SGK1 for controlvs. furosemide-depleted hypothalamus samples, the P value was0.006 for the isolated comparison, but taking into account (as forall data shown here) the context of the genome-wide analysis, theP value was 0.31. With these data, SGK1 is more likely to beregulated than not regulated, and it ranked 173 for genes ex-clusively regulated in furosemide/Na+ depletion-induced sodiumappetite. Aforementioned up-regulation of SGK1 in the brain(7) did not include study of the hypothalamus, and it was con-ducted in response to dehydration, not in response to conditionsthat evoke sodium appetite. In regards to the functional role ofSGK1 for mineral appetite, Sgk1 was found necessary formineralocorticoid-induced sodium appetite evoked by desoxy-corticosterone acetate (DOCA) in Sgk1 null mice (7, 8). Thisconcept is not contradicted by our data. A gene that is necessaryfor a certain homeostatic behavior need not necessarily be reg-ulated by the cue that evokes the respective behavior.Another interesting aspect is dependence of sodium appetite
on peripheral afferents. Complete peripheral denervation abro-gates sodium appetite, which was shown decades ago. Whetherdopaminergic transmission is of critical relevance in the peripheraltaste system for sodium taste is currently unknown, representingan interesting subject for future studies.However, wewish to stressthat we did identify one critical local relay for sodium appetite inthe lateral hypothalamus by our injection studies.
SI Materials and MethodsDNA Microarrays. Mouse oligonucleotide arrays were printed atthe Duke Microarray Facility using the Operon’s Mouse GenomeOligo Set (3.0), which contains 31,769 70-mer probes repre-senting 24,878 genes and 32,829 transcripts (Operon). Arrayswere printed using an Omnigrid 300 arrayer by Genomic Sol-utions on Corning UltraGap amino-saline coated glass slides(Fisher Scientific).
RNA and Microarray Probe Preparation and Hybridization. Total RNA(1 μg) from each sample (mouse hypothalamus or remainder ofthe brain) and the reference (Universal Mouse Reference RNA;Stratagene) were used in probe preparation. The purified anti-sense RNA was fluorescently labeled with Cy3 (reference) andCy5 (sample; GE Healthcare). Sample and reference RNAs werepooled, mixed with 1× hybridization buffer (50% formamide, 5×SSC, and 0.1% SDS), COT-1 DNA, and poly-dA to limit non-specific binding, and heated to 95 °C for 2 min. This mixture waspipetted onto a microarray slide using a Maui Mixer hybridizationchamber (BioMicro Systems) and hybridized overnight at 42 °C onan automated Maui Hybridization Station (BioMicro Systems).The array was then washed at increasing stringencies and scannedon a GenePix 4000B microarray scanner (Axon Instruments).Detailed protocols are available on the Duke Microarray Facilitywebsite (http://www.genome.duke.edu/cores/microarray/services/spotted-arrays/protocols/).
Microarray Analysis. The resulting microarray data were analyzedusing the Duke Microarray Database analysis platform, a locallyinstalled version of the Stanford Microarray Database (9, 10).Background-subtracted and normalized data were averaged foreach of greater than or equal to four animals within the threegroups. Regulated genes were identified by comparison of nor-malized values.Within-array Loess normalization was performed using the
Limma package (11) to remove any bias that the intensity has onlog-fold change. Next, normalization was carried out betweenarrays using variance stabilization in that both the red and greenchannels on the arrays are aligned to each other based on theassumption that most of the genes are constant across experiments(12). Spots not reaching the quality criterion were removed fromadditional analysis. Quality was determined in three ways. (i)There was insufficient red or green intensity to produce a reliableratio. (ii) The consistency of the ratio throughout the spot was low.(iii) Other machine-annotated flags indicated that the spot wasnot reliably read. After filtering, if a particular spot was found onlyto be present on a small number of arrays (<15), then that spot wasfiltered for all arrays. To be sure that the experiment consistentlymeasured differences between groups, we performed a principlecomponent analysis. This procedure allowed us to separate theexperiments by both brain region (hypothalamus vs. whole brainwithout hypothalamus] and treatment [adrenocorticotropic hor-mone (ACTH)-induced sodium appetite, sodium deprivation plusfurosemide, and control). Spots passing these criteria were rankedand analyzed for statistically significant differences using a linearmodel (13).
Gene Set Enrichment Analysis. Gene set enrichment analysis(GSEA) (14) was used to test the hypothesis that addiction-relatedgenes were regulated by sodium appetite. GSEA takes as inputa set of microarray data, a set of comparisons to make betweengroups in that experiment, and a series of gene sets. Each ofthose gene sets, often pathways, represents a hypothesis of whichgenes are regulated in the data. The function of GSEA is to testeach of those hypotheses against the data in a statistically rig-orous manner. The processed and normalized microarray datawere altered by hand to conform to the .gct format. After beingimported into the GSEA, a comparison was made, using a t test,between the expression of genes under the control conditionsand the expression of genes in mice with sodium appetite. Wetested these gene expression changes against lists of genes that
Liedtke et al. www.pnas.org/cgi/content/short/1109199108 1 of 6
have been confirmed by multiple sources to play a role in dif-ferent addiction processes (15, 16). These genes included anopioids gene set, a cocaine gene set, an alcohol gene set, and anicotine gene set, a total of four gene sets. In addition, wedownloaded all known pathway gene sets of a similar size (50–350genes) from the GSEA website (http://www.broadinstitute.org/gsea), a total of 47 gene sets. These gene sets were used to gaugetheir level of enrichment vs. the addiction gene sets. Note, how-ever, that components of these pathways might be shared with anaddiction-related set. To check an overrepresentation of neuralactivity-related genes, we used three such gene sets based onexperiments across species (17, 18). Thus, 54 gene sets were fi-nally interrogated. As output, GSEA provides a nominal P valuefor each gene set, which represents how significantly up- or down-regulated the genes within that set are in the microarray datacomparison. There may be multiple gene sets tested, and there-fore, a false discovery rate q value and a family-wise error rate Pvalue are included to correct for multiple hypotheses. For a geneset to be significantly affected by sodium appetite, we require P <0.05, false discovery rate < 0.2, and family-wise error rate < 0.2.
Immunohistochemistry. Fixation of rat brains (sodium-depletedvs. controls) was either by transcardial perfusion after euthanasiawith 2% paraformaldehyde for subsequent vibratome sectioningor with short-term fixation on slide for 2 min with fresh-frozen
section. Dopamine- and cAMP-regulated neuronal phospho-protein 32 kDa (DARPP-32) was immunolabeled in 50-μm vi-bratome and 25-μm fresh-frozen sections using a monoclonalantibody supplied by Paul Greengard (The Rockefeller Univer-sity, New York, NY) (19) and detected with Alexa-488 anti-mouse secondary. Sections were double-labeled with mouseanti–DARPP-32 plus rabbit antiorexin antibody (orexin A anti-body; Santa Cruz Biotechnology) and detected with anti-mouseAlexa-488, anti-rabbit biotin, and subsequent streptavidin Alexa-595 (Invitrogen-Molecular Probes). Sections were triple-labeledwith mouse anti-DARPP-32, rabbit anti-orexin, and anti-ARCantibody raised in goats (Santa Cruz Biotechnology), and sec-ondary detection was with anti-goat Alexa-360, anti-rabbit biotinplus Streptavidin-595, and anti-mouse Alexa-488. Fluorescentmicroscopy or confocal imaging were conducted using a Olym-pus BX61 microscope and a Zeiss LSM710 confocal workstation.Additional information is available on the following topics:
general reference on sodium appetite (20), hypothalamus andinstincts (21, 22), DARPP-32 (23–25), lateral hypothalamus (ex-pression of orexin and reward-related circuits) (26–35), involve-ment of α-2C adrenoreceptor in addiction and monoaminergicreward pathways (36–38), lack of effect of antagonists used in thisinvestigation on alertness, drive, and locomotion (39–45), andaddiction as an artificial instinct (46–48).
1. Morris MJ, Na ES, Johnson AK (2010) Mineralocorticoid receptor antagonism preventshedonic deficits induced by a chronic sodium appetite. Behav Neurosci 124:211–224.
2. Na ES, Morris MJ, Johnson AK (2009) Behavioral cross-sensitization between morphine-induced locomotion and sodium depletion-induced salt appetite. Pharmacol BiochemBehav 93:368–374.
3. Morris MJ, Na ES, Johnson AK (2008) Salt craving: The psychobiology of pathogenicsodium intake. Physiol Behav 94:709–721.
4. Na ES, Morris MJ, Johnson RF, Beltz TG, Johnson AK (2007) The neural substrates ofenhanced salt appetite after repeated sodium depletions. Brain Res 1171:104–110.
5. Grippo AJ, Moffitt JA, Beltz TG, Johnson AK (2006) Reduced hedonic behavior andaltered cardiovascular function induced by mild sodium depletion in rats. BehavNeurosci 120:1133–1143.
6. Morris MJ, Na ES, Grippo AJ, Johnson AK (2006) The effects of deoxycorticosterone-induced sodium appetite on hedonic behaviors in the rat. Behav Neurosci 120:571–579.
7. Wärntges S, et al. (2002) Cerebral localizationand regulationof the cell volume-sensitiveserum- and glucocorticoid-dependent kinase SGK1. Pflugers Arch 443:617–624.
8. VallonV, et al. (2005) SGK1as adeterminantofkidney functionand salt intake in responseto mineralocorticoid excess. Am J Physiol Regul Integr Comp Physiol 289:R395–R401.
9. Ball CA, et al. (2005) The Stanford Microarray Database accommodates additionalmicroarray platforms and data formats. Nucleic Acids Res 33:D580–D582.
10. Gollub J, et al. (2003) The Stanford Microarray Database: Data access and qualityassessment tools. Nucleic Acids Res 31:94–96.
11. Smyth GK, Speed T (2003) Normalization of cDNA microarray data. Methods 31:265–273.
12. Huber W, von Heydebreck A, Sültmann H, Poustka A, Vingron M (2002) Variancestabilization applied to microarray data calibration and to the quantification ofdifferential expression. Bioinformatics 18(Suppl 1):S96–S104.
13. Smyth GK (2004) Linear models and empirical bayes methods for assessing differentialexpression in microarray experiments. Stat Appl Genet Mol Biol 3:3.
14. Subramanian A, et al. (2005) Gene set enrichment analysis: A knowledge-basedapproach for interpreting genome-wide expression profiles. Proc Natl Acad Sci USA102:15545–15550.
15. Li CY, Mao X, Wei L (2008) Genes and (common) pathways underlying drug addiction.PLoS Comput Biol 4:e2.
16. Suh YJ, Yang MH, Yoon SJ, Park JH (2006) GEDA: New knowledge base of geneexpression in drug addiction. J Biochem Mol Biol 39:441–447.
17. Warren WC, et al. (2010) The genome of a songbird. Nature 464:757–762.18. Xiang G, et al. (2007) Identification of activity-dependent gene expression profiles
reveals specific subsets of genes induced by different routes of Ca(2+) entry incultured rat cortical neurons. J Cell Physiol 212:126–136.
19. Ouimet CC, Miller PE, Hemmings HC Jr., Walaas SI, Greengard P (1984) DARPP-32,a dopamine- and adenosine 3′:5′-monophosphate-regulated phosphoprotein enriched indopamine-innervated brain regions. III. Immunocytochemical localization. J Neurosci 4:111–124.
20. Dethier VG (1977) The taste of salt. Am Sci 65:744–751.21. Fitzsimons JT (1998) Angiotensin, thirst, and sodium appetite. Physiol Rev 78:583–686.22. Schulkin J, Fluharty SJ (1985) Further studies on salt appetite following lateral
hypothalamic lesions: Effects of preoperative alimentary experiences. Behav Neurosci99:929–935.
23. Bateup HS, et al. (2008) Cell type-specific regulation of DARPP-32 phosphorylation bypsychostimulant and antipsychotic drugs. Nat Neurosci 11:932–939.
24. Nairn AC, et al. (2004) The role of DARPP-32 in the actions of drugs of abuse.Neuropharmacology 47(Suppl 1):14–23.
25. Santini E, et al. (2007) Critical involvement of cAMP/DARPP-32 and extracellularsignal-regulated protein kinase signaling in L-DOPA-induced dyskinesia. J Neurosci27:6995–7005.
26. Aston-Jones G, Smith RJ, Moorman DE, Richardson KA (2009) Role of lateralhypothalamic orexin neurons in reward processing and addiction.Neuropharmacology56(Suppl 1):112–121.
27. Aston-Jones G, et al. (2010) Lateral hypothalamic orexin/hypocretin neurons: A rolein reward-seeking and addiction. Brain Res 1314:74–90.
28. Carter ME, Borg JS, de Lecea L (2009) The brain hypocretins and their receptors:Mediators of allostatic arousal. Curr Opin Pharmacol 9:39–45.
29. Cason AM, et al. (2009) Role of orexin/hypocretin in reward-seeking and addiction:Implications for obesity. Physiol Behav 100:419–428.
30. Kelley AE (2004) Memory and addiction: Shared neural circuitry and molecularmechanisms. Neuron 44:161–179.
31. Kelley AE, Berridge KC (2002) The neuroscience of natural rewards: Relevance toaddictive drugs. J Neurosci 22:3306–3311.
32. Kenny PJ, Boutrel B, Gasparini F, Koob GF, Markou A (2005) Metabotropic glutamate5 receptor blockade may attenuate cocaine self-administration by decreasing brainreward function in rats. Psychopharmacology (Berl) 179:247–254.
33. Rao Y, et al. (2008) Regulation of synaptic efficacy in hypocretin/orexin-containingneurons by melanin concentrating hormone in the lateral hypothalamus. J Neurosci28:9101–9110.
34. Richards JK, Simms JA, Bartlett SE (2009) Conditioned cues and yohimbine inducereinstatement of beer and near-beer seeking in Long-Evans rats. Addict Biol 14:144–151.
35. Sharf R, Sarhan M, Dileone RJ (2008) Orexin mediates the expression of precipitatedmorphine withdrawal and concurrent activation of the nucleus accumbens shell. BiolPsychiatry 64:175–183.
36. Hahn B, Stolerman IP (2005) Modulation of nicotine-induced attentional enhancementin rats by adrenoceptor antagonists. Psychopharmacology (Berl) 177:438–447.
37. Ihalainen JA, Tanila H, Scheinin M, Riekkinen P Jr. (2001) alpha(2C)-Adrenoceptorsmodulate the effect of methylphenidate on response rate and discriminationaccuracy in an operant test. Brain Res Bull 54:553–557.
38. Zhang W, Klimek V, Farley JT, Zhu MY, Ordway GA (1999) alpha2C adrenoceptorsinhibit adenylyl cyclase in mouse striatum: Potential activation by dopamine.J Pharmacol Exp Ther 289:1286–1292.
39. Cowen MS, Krstew E, Lawrence AJ (2007) Assessing appetitive and consummatoryphases of ethanol self-administration in C57BL/6J mice under operant conditions:Regulation by mGlu5 receptor antagonism. Psychopharmacology (Berl) 190:21–29.
40. Moore R, Krstew EV, Kirchhoff J, Davisson RL, Lawrence AJ (2007) Centraloverexpression of angiotensin AT(1A) receptors prevents dopamine D(2) receptorregulation of alcohol consumption in mice. Alcohol Clin Exp Res 31:1128–1137.
41. McNamara FN, et al. (2003) Congenic D1A dopamine receptor mutants: Ethologicallybased resolution of behavioural topography indicates genetic background asa determinant of knockout phenotype. Neuropsychopharmacology 28:86–99.
42. Ceccarelli SM, et al. (2007) Rational design, synthesis, and structure-activityrelationship of benzoxazolones: New potent mglu5 receptor antagonists based onthe fenobam structure. Bioorg Med Chem Lett 17:1302–1306.
Liedtke et al. www.pnas.org/cgi/content/short/1109199108 2 of 6
43. Jaeschke G, et al. (2007) Synthesis and biological evaluation of fenobam analogs asmGlu5 receptor antagonists. Bioorg Med Chem Lett 17:1307–1311.
46. Anonymus (1971) Addiction—an artificially induced instinct. Can Med Assoc J 104:825.
47. Bejerot N (1971) Addiction as an artificially induced instinct. A theory.NordMed 85:20–27.48. Bejerot N (1971) Addiction: Its nature, spread and treatment. Isr Ann Psychiatr Relat
Discip 9:155–169.
Fig. S1. Gene expression analysis using microarrays. A depicts the heat map, as shown in Fig. 1A, with spelled-out abbreviations for gene symbols andmaximum log-fold change. B shows the original enrichment plots of GSEA analysis. Depicted here are enrichment plots for the gene sets for opioids, cocaine,alcohol, and neural activity (1, 2), and the location of each gene in the set is represented against the ranking of all genes (red to blue) by a black vertical line(below the running score is the green line). The location of the gene set members is used to calculate the enrichment score, which is defined as the maximum/minimum distance of the running score (green line) from baseline. Note the different scales of the y axis for different sets: opioids minimum = −0.5, cocaineminimum = −0.4, alcohol minimum = −0.35, and neural activity minimum = −0.2. The even distribution in the activity set is conspicuous, indicating lack ofenrichment, as is the robustly enriched feature (right-shifted to overexpressed genes) in the addiction-related subsets. These four subsets depict the combineddata from both methods of induction of sodium appetite.
Liedtke et al. www.pnas.org/cgi/content/short/1109199108 3 of 6
1. Li CY, Mao X, Wei L (2008) Genes and (common) pathways underlying drug addiction. PLoS Comput Biol 4:e2.2. Xiang G, et al. (2007) Identification of activity-dependent gene expression profiles reveals specific subsets of genes induced by different routes of Ca(2+) entry in cultured rat cortical
neurons. J Cell Physiol 212:126–136.
Fig. S2. DARPP-32 expression in the periventricular nucleus of a sodium-depleted rat is shown. A shows the anatomic orientation, including anatomicalhallmarks of this section of the hypothalamus, shown in the micrograph. B depicts representative examples of up-regulated DARPP-32 expression in theperiventricular nucleus in a sodium-depleted animal. Note the positive label in the ependymal cells as well. This expression pattern was not detected in controlanimals; counts were 18, 14, and 23 labeled periventricular cells in three animals vs. 0, 0, and 0 (three animals) in sodium depletion vs. control. (Scale bar:30 μm.) III indicates the third ventricle for orientation. C depicts lack of regulation of DARPP-32 immunoreactivity in the suprachiasmatic (SCN) and supraoptic(SON) nucleus of two representative animals. (Scale bar: 30 μm.)
Liedtke et al. www.pnas.org/cgi/content/short/1109199108 4 of 6
Fig. S3. Pharmacological antagonism attenuates gratification behavior of sodium appetite. (A) This figure shows results of specific inhibition of metabotropicglutamate receptor 5 receptors in mouse and rat. Left shows the time course of gratification behavior in sodium-depleted mice as significantly attenuated bythe specific metabotropic glutamate receptor 5 receptor antagonist, MTEP, applied at 20 mg/kg (n = 5 mice per group). Differences to vehicle-treated mice arestatistically significant. Right shows the robust effects of fenobam, an mGlu5 inhibitor, strongly reducing gratification behavior in rats with induced sodiumappetite. Fenobam was used at 25 mg/kg fenobam (1, 2). Upper Right is the amount of 0.3 M NaCl consumed within 20 min; in Lower Right, the latency to firstdrinking is shown. Note that, in the rat model (similar to mouse models), ≥90% of 0.3 M NaCl is consumed within the first 20 min after 0.3 M NaCl is madeavailable (n = 6 rats per group; P < 0.001, t test). (B Left) This diagram shows the attenuation of sodium appetite by intra-lateral hypothalamus (LH) micro-injection [D1(5)-R–specific antagonist SCH23390 at 100 nM in 200 nL injection volume] in individual rats 1–4. Schematics (B Right) depicting the location of theinjection needle within the lateral hypothalamus are shown. The diagram with red dots for the verified injection site depicts the two rats that had completeelimination of sodium appetite with SCH23390. Note that the size of a calculated sphere with 362 μm radius correlating to an injection volume of 200 nL isshown in light red. The schematic in green depicts the injection sites for rats 1 and 4, with the same rendering of 200 nL injection volume. The micrograph onthe right is of an injection site with a fluorescently labeled dextran tracer. (Scale bar: 400 μm.)
1. Montana MC, et al. (2009) The metabotropic glutamate receptor subtype 5 antagonist fenobam is analgesic and has improved in vivo selectivity compared with the prototypicalantagonist 2-methyl-6-(phenylethynyl)-pyridine. J Pharmacol Exp Ther 330:834–843.
2. Porter RH, et al. (2005) Fenobam: A clinically validated nonbenzodiazepine anxiolytic is a potent, selective, and noncompetitive mGlu5 receptor antagonist with inverse agonist activity.J Pharmacol Exp Ther 315:711–721.
Liedtke et al. www.pnas.org/cgi/content/short/1109199108 5 of 6
Dataset S1, Table S1. Top 600 regulated genes for both conditions of sodium appetite combined
Dataset S1, Table S1
This dataset lists the 600 most regulated genes (fold regulation log2 for each animal) by their gene symbol with data for each animal, and it also provides theP value of combined sodium appetite groups vs. controls (column B) and the P value for combined gratified vs. controls (column C). Columns AF and AG depictprimer sequences for the qRT-PCR assays (genes that were probed by qRT-PCR are highlighted in orange), and column AH is the addiction subset to whichgenes belonged (A, alcohol; C, cocaine; O, opioid). Genes belonging to the addiction gene set are shown in bold.
Dataset S1, Table S2. Top 100 regulated genes for sodium depletion
Dataset S1, Table S2
This dataset shows genes specifically regulated for Na+ depletion. Fold regulation is shown as log2. The top 100 regulated genes are shown, which areexclusively regulated in one condition, not regulated in the other, and not regulated in the combined data set. Turquoise highlighting indicates members ofthe addiction gene set.
1. Horike N, et al. (2003) Adipose-specific expression, phosphorylation of Ser794 in insulin receptor substrate-1, and activation in diabetic animals of salt-inducible kinase-2. J Biol Chem278:18440–18447.
2. Katoh Y, et al. (2004) Salt-inducible kinase (SIK) isoforms: Their involvement in steroidogenesis and adipogenesis. Mol Cell Endocrinol 217:109–112.3. Okamoto M, Takemori H, Katoh Y (2004) Salt-inducible kinase in steroidogenesis and adipogenesis. Trends Endocrinol Metab 15:21–26.4. Takemori H, Okamoto M (2008) Regulation of CREB-mediated gene expression by salt inducible kinase. J Steroid Biochem Mol Biol 108:287–291.
Table S1. Top 10 enriched gene sets
Name Size ES NES NOM P value FDR q value FWER P value
Values for pooled data based on both sodium appetite conditions are shown (increased over single sodium appetite condition); also,note the enrichment parameters for gene set neural activity, which indicate relative lack of enrichment. The neural activity gene set is alsoin Fig. 1 and SI Material and Methods. Addiction gene subsets are highlighted in turquoise. ES, enrichment score; FDR, false discovery rate;NES, normalized enrichment score; NOM, nominal; FWER, family-wise error rate.
Dataset S1, Table S3. Top 100 regulated genes for sodium appetite induced by ACTH
Dataset S1, Table S3
This dataset shows genes specifically regulated for ACTH-evoked Na+ appetite. Fold regulation is shown as log2. The top 100 regulated genes are shown,which are exclusively regulated in one condition, not regulated in the other, and not regulated in the combined data set. Turquoise indicates members of theaddiction gene set, orange indicates synaptic function-related genes, and green indicates other genes of interest. Regarding the latter, note the up-regulationof salt-inducible kinase 2 (SIK2), as well as a functional subunit of the glucocorticoid receptor, suggesting the possibility of a feed-forward dysregulationunderlying the effects of chronic ACTH administration. Specifically, we report expression and regulation for ACTH-mediated SIK2 overexpression in the centralnervous system. SIK2 refers to the concept of a cellular sodium sensor that can also regulate chromatin (1–4).
1. Horike N, et al. (2003) Adipose-specific expression, phosphorylation of Ser794 in insulin receptor substrate-1, and activation in diabetic animals of salt-inducible kinase-2. J Biol Chem278:18440–18447.
2. Katoh Y, et al. (2004) Salt-inducible kinase (SIK) isoforms: Their involvement in steroidogenesis and adipogenesis. Mol Cell Endocrinol 217:109–112.3. Okamoto M, Takemori H, Katoh Y (2004) Salt-inducible kinase in steroidogenesis and adipogenesis. Trends Endocrinol Metab 15:21–26.4. Takemori H, Okamoto M (2008) Regulation of CREB-mediated gene expression by salt inducible kinase. J Steroid Biochem Mol Biol 108:287–291.
Liedtke et al. www.pnas.org/cgi/content/short/1109199108 6 of 6
