Transcriptomic Analysis of the Osmotic and Reproductive Remodeling of the Female Rat Supraoptic Nucleus

Abstract

The supraoptic nucleus (SON) of the hypothalamus is an important integrative brain structure that coordinates responses to perturbations in water balance and regulates maternal physiology through the release of the neuropeptide hormones vasopressin and oxytocin into the circulation. Both dehydration and lactation evoke a dramatic morphological remodeling of the SON, a process known as function-related plasticity. We hypothesize that some of the changes seen in SON remodeling are mediated by differential gene expression, and have thus used microarrays to document global changes in transcript abundance that accompany chronic dehydration in female rats, and in lactation. In situ hybridization analysis has confirmed the differential expression of three of these genes, namely TNF-induced protein 6, gonadotropin-inducible transcription factor 1, and ornithine decarboxylase antizyme inhibitor 1. Comparison of differential gene expression patterns in male and female rats subjected to dehydration and in lactating rats has enabled the identification of common elements that are significantly enriched in gene classes with particular functions. Two of these are related to the requirement for increased protein synthesis and hormone delivery in the physiologically stimulated SON (translation initiation factor activity and endoplasmic reticulum-Golgi intermediate compartment, respectively), whereas others are consistent with the concept of SON morphological plasticity (collagen fibril organization, extracellular matrix organization and biogenesis, extracellular structure organization and biogenesis, and homophilic cell adhesion). We suggest that the genes coordinately regulated in the SON as a consequence of dehydration and lactation form a network that mediates the plastic processes operational in the physiologically activated SON.

The hypothalamo-neurohypophyseal system (HNS) consists of the large peptidergic magnocellular neurons (MCN) of the hypothalamic supraoptic nucleus (SON) and paraventricular nucleus, the axons of which course through the internal zone of the median eminence and terminate on blood capillaries of the posterior lobe of the pituitary gland (1). The SON is a homogenous collection of MCN, whereas the paraventricular nucleus is divided into a lateral subdivision of MCN and a more medial sector of parvocellular neurons the axons of which project to the external zone of the median eminence (2) and to the brainstem and spinal cord (3) and which are involved in the stress response and cardiovascular control, respectively.

The HNS is the source of two major neuropeptide hormones (4), namely vasopressin (VP) and oxytocin (OT). Both hormones are synthesized as parts of separate prepropeptides encoded by highly homologous linked genes (5). These precursors are processed during anterograde axonal transport to terminals in the posterior pituitary where biologically active VP and OT are stored until mobilized for secretion into the circulation by MCN electrical activities evoked by physiological cues (6). Single-cell RT-PCR enables VP and OT transcripts to be detected in the same MCN (7), but the expression levels of each neuropeptide RNA differ by orders of magnitude. Only a few percent of MCN express high, equivalent levels of both peptides (8), although the proportion increases after dehydration (9) and lactation (10).

Physiological activation of the HNS, resulting in massive hormone release, is a characteristic of both dehydration and lactation. VP is crucially involved in the maintenance of osmotic stability (11). After dehydration, a rise in plasma osmolality is detected by intrinsic MCN osmoreceptor mechanisms (12–14) and by specialized osmoreceptive neurons in the circumventricular organs that project to the MCN (13, 15, 16) and provide direct glutamate receptor-mediated excitatory inputs (17) to shape firing activity (18, 19) for hormone secretion (20). Upon release, VP travels through the bloodstream to specific receptor targets located in the kidney where it increases the permeability of the collecting ducts to water, reducing the renal excretion of water, thus promoting water conservation. Although also released during dehydration (21), when OT is thought to have natriuretic activity at the level of the kidney (22), OT is best known for its roles in parturition and in the milk ejection reflex during lactation (23).

The HNS offers a unique example in the adult mammalian central nervous system of a functional and structural plasticity related to a physiological state (24). Both dehydration stress and lactation evoke a remodeling of the HNS (25, 26). A plethora of activity-dependent changes in the morphology, electrical properties, and biosynthetic and secretory activity of the HNS have all been described (24), which may contribute to the facilitation of hormone production and delivery, and hence the survival of the organism. For example, alterations in the relationship between MCN and glia, the extent of terminal contact with the basal lamina in the neurohypophysis, the type and weight of synaptic inputs, and the extent of electrotonic coupling between MCN, have all been documented (27–31). This plasticity appears to be governed by a complex and dynamic interplay between the intrinsic properties of the MCN, interactions between MCN, interactions with glia, and the influences of extrinsic synaptic inputs. The response of the HNS to dehydration and lactation represents a unique and tractable model for understanding the processes whereby changes in gene expression mediate neuronal plasticity (24), but the molecular mechanics of these processes remain to be elucidated.

We have suggested that changes in the steady-state levels of specific RNA might be an effective strategy for regulating and affecting HNS remodeling (32, 33). To test this hypothesis, we used microarrays to describe the global gene expression profile of the male SON, and how this changes after chronic (3 d) dehydration (32, 33). We have now extended these studies to encompass the SON of the virgin randomly cycling female rat, under both euhydrated and dehydrated conditions, and of the lactating rat. Thus, for the first time, we have revealed changes in global gene expression patterns in the SON that accompany dehydration and lactation in the female rat. These catalogs are an important resource for researchers working on all aspects of SON physiology, particularly function-related plasticity.

Materials and Methods

Animals

Sprague Dawley (Hsd:Sprague Dawley SD) rats were obtained from Harlan (Bicester, UK). For the lactating animals, matings occurred at approximately 8–10 wk of age. Thus, by the time the dams were killed 11 d post partum, they were approximately of a similar age to the nonlactating rats (10–12 wk). Animals were maintained in standardized temperature (22 ± 1 C), humidity (50 ± 5%), and diurnal conditions (10 h light, 14 h dark; lights on at 0700 h) and had free access to food (standard laboratory rat chow). Dehydration of male rats and of randomly cycling virgin female rats involved complete fluid deprivation for exactly 72 h, while control animals had free access to drinking water (tap). Lactating rats had litter sizes ranging from eight to 14 pups per litter (mean of 11.38 ± 0.36). All experimental procedures were approved by the University of Bristol Ethical Review Committee and were carried out under UK Home Office license in accord with the Animals (Scientific Procedures) Act, 1986. Animals were always killed between 1100 h and 1300 h.

Tissue collection

In all experiments, rats were stunned according to Schedule 1 of the Animals (Scientific Procedures) Act 1986, and decapitated with a small animal guillotine (Harvard Apparatus, Holliston, MA). The brain was rapidly removed from the cranium and placed in an ice-cold brain matrix (ASI Instruments, Warren, MI). Two or three sections of approximately 1 mm thickness were taken, and the SON was carefully dissected on a bed of ice under a dissecting microscope (Leica, Nussloch, Germany). After isolation, the samples were rinsed in ribonuclease-free PBS then immediately placed in Eppendorf tubes containing RNAlater (Ambion, Huntingdon, UK) and stored for no longer than 1 month at −20 C before further processing. The above procedures were carried out in ribonuclease-free manner. A single operative carried out all dissections.

Microarray analysis

Protocols for RNA extraction, amplification, and hybridization have been described (32). Affymetrix contract research services were provided by SourceBioscience (http://lifesciences.sourcebioscience.com/). Separate microarrays (n = 5, except euhydrated female, where n = 4) were probed using independently generated targets. For each completely independent replicate, tissue from five rats was pooled for RNA extraction. The raw data (.CEL) from each of the 24 Rat GeneChip 230 2.0 microarrays were then loaded into GeneSpring GX11 (Agilent Technologies, Stockport, UK) where it was summarized with Mas5 (which incorporates a scaling normalization) and transformed to the median of all samples. GeneSpring GX11 was also used for high-level analysis, namely principal component analysis (PCA; 34), statistical analysis, and GO exploration. All raw data have been submitted to the NBCI Gene Expression Omnibus (GEO; www.ncbi.nlm.nih.gov/geo). Accession number: GSE30733. Gene annotations were checked using online tools and databases (Entrez Gene, Nucleotide, and Unigene, www.ncbi.nlm.nih.gov/; BLAST, www.ncbi.nlm.nih.gov/BLAST; NetAffyx, www.affymetrix.com). Also see: www.vasopressin.org/#/data-bank/375542.

In situ hybridization analysis

Low- and high-resolution in situ hybridization experimental and analytical protocols have been described in detail elsewhere (33). Oligonucleotide hybridization probes were obtained from GeneDetect (Bradenton, FL). All the probes were cartridge purified to ensure more than 95% full length. Antisense probes were either single sequences, or, in the case of Giot1, were a cocktail of three sequences that recognize the same transcript (Supplemental Table 1 published on The Endocrine Society's Journals Online web site at http://endo.endojournals.org). The expression of Giot1 is not as high the other genes studied; therefore, a cocktail of multiple oligonucleotide probes was used to give better in situ signals. To confirm the specificity of the antisense probes, control experiments with all corresponding sense RNA probes were performed. No signals were seen with any of the sense probes. Processed sections were examined on a Leica DM IRB microscope (Leica Microsystems, Milton Keynes, UK). Photomicrographs were taken with a Leica DC-300F digital camera (Leica Microsystems, Milton Keynes, UK) using IM50 software (version 1.2; Leica Microsystems), saved as TIFF images at 300 dpi resolution. Semiquantitative image analysis was conducted double-blind. Cell counting was performed at ×200 magnification on a Leica DM IRB microscope with C-plan optics. Grain counting was conducted under bright-field conditions at a magnification of ×400 using the above microscope and a Leica 300DM digital camera. From each treatment group, two sections were selected from the middle region of each nucleus. These selected sections were carefully matched between groups. From each section, 10–15 labeled cells were selected at random from throughout the whole nucleus to determine the number of silver grains expressed in each cell. Statistical analyses were performed using one-way ANOVA followed by Tukey's multiple comparison test.

Results

The female SON transcriptome

We have interrogated Affymetrix rat 230 2.0 GeneChip probes, which consist of 31,009 oligonucleotide probe sets representing about 30,000 transcripts encoded by approximately 28,000 genes, with targets derived from the SON of euhydrated, randomly cycling virgin female rats, 3-d dehydrated, randomly cycling virgin female rats, and 11-d lactating rats. Data from SON of euhydrated and dehydrated male rats have previously been published (32, 33) and are here reanalyzed using GeneSpring GX11 to enable comparisons with the new female data. We first compiled list of genes called Present (P) in all five, independent euhydrated female (13,393 genes), dehydrated female (14,126 genes), euhydrated male (16,595 genes), dehydrated male (15851 genes), or lactation (14,896 genes) experiments. All marginal, absent, or unknown calls were excluded. Data, listed according to raw mean expression values, are available as files for download (Supplemental Appendix, pages 1–5, respectively). These gene lists are transcriptome catalogs that, with a high degree of confidence, represent comprehensive descriptions of the RNA populations expressed in the SON under these different conditions. Calculation of mean correlation coefficient for each experimental group demonstrated a strong correlation between microarrays of the same group (Fig. 1A); euhydrated male (r = 0.992), dehydrated male (r = 0.994) lactation (r = 0.993), euhydrated female (r = 0.978), and dehydrated female (r = 0.969). PCA (34), based on experimental condition using all available data, revealed distinct clusters corresponding to the different experimental groups (Fig. 1B).

Fig. 1.

A, Mean correlation coefficient for each experimental group in a pair-wise fashion reveals that arrays belonging to the same group have a mean correlation coefficient closer to 1 when compared with each other than when compared with any other group. However, although the euhydrated male (EM), dehydrated male (DM), and lactation (L) microarrays correlate closely, the euhydrated female (EF) and dehydrated female (DF) microarrays show more variation. B, PCA identifies directions along which the variation is maximal. The plot of PCA resolves different microarray samples in terms of how similar or different they are within in a three-dimensional space that represents three different component scores, one for each axis. PCA based on experimental condition using all available data reveals a greater degree of variation in the female samples compared with the male of lactation experimental data. C, Venn comparison (vs.) of transcripts differentially expressed in the SON as a consequence of lactation (EF vs. L), dehydration in male rats (EM vs. DM) and dehydration in female rats (EF vs. DF).

Identification of common gene expression elements associated with SON plasticity

The P lists from each condition were combined to give an experimental list of 17,879 genes, which was subjected to statistical testing together with post hoc analysis and multiple test correction [Welch ANOVA, P < 0.05; Tukey HSD (Honestly Significant Difference) test, Benjimini-Hochberg multiple test correction]. Note that the false discovery rate of this protocol is approximately 5% of identified genes. Post hoc analysis allowed us to look at those genes significantly regulated between specific conditions. Thus, 2353 genes are significantly regulated by dehydration in female rats (Supplemental Appendix, page 6), 2141 genes are significantly regulated by dehydration in male rats (Supplemental Appendix, page 7), and 4262 genes are significantly regulated by lactation (Supplemental Appendix, page 8). Comparison of these data by Venn analysis (Fig. 1C) revealed a core of 567 genes that are commonly regulated in the SON by dehydration in the male, and by both dehydration and lactation in the female. A further 341 genes are regulated by dehydration in the male and lactation, but not by dehydration in female, 645 genes are regulated by both lactation and dehydration in female, but not by dehydration in male, and 517 genes and regulated by dehydration in both male and female, but not by lactation. Combining these lists reveals 2070 genes that are regulated by at least two conditions (Supplemental Appendix, page 9). Of these, 257 genes are differentially expressed by at least 2-fold in at least one condition (Supplemental Appendix, page 10).

In situ hybridization validation of dehydration- and lactation-regulated genes

The microarray data suggested that that the genes encoding gonadotropin-inducible ovarian transcription factor 1 (Giot1), TNF-α-induced protein 6 (Tnfaip6, also known as TSG-6), and antizyme inhibitor 1 (Azin1) are up-regulated in the female SON by dehydration. In contrast, the array suggested that lactation resulted in no significant increase in either Giot1 or Tnfaip6 expression and only a small increase in Azin1 transcript abundance. We have previously shown that dehydration results in robust and significant increases in the expression of all three genes in the male SON (33, 35); this analysis has now been extended to the female brain (Fig. 2). Taking advantage of the anatomical precision of in situ hybridization, low-low resolution analysis revealed that expression of Giot1 and Tnfaip6 mRNA was barely detectable in control female rat SON, whereas a weak Azin1 signal was observed. In accordance with the microarray data, significant up-regulation of all three genes was observed with dehydration (Table 1). With the stimulus of lactation, Giot1, Tnfaip6, and Azin1 expression were significantly increased, but by a smaller degree than in dehydration (Table 1).

Fig. 2.

Representative low-resolution in situ hybridization images of Tnfaip6, Giot1, and Azin1 mRNA expression in the euhydrated female (EF), dehydrated female (DF), and lactating (L) rat hypothalamus. Sections were exposed to film for 2 wk. Scale bar, 1 mm.

Table 1.

Quantification of changes in Giot1, Tnfaip6, and Azin1 mRNA expression in the SON dehydrated male (DM) compared (vs.) with euhydrated male (EM), dehydrated female (DF) compared with euhydrated female (EF), and lactating (L) female compared with euhydrated female (EF) as assessed by low-resolution in situ hybridization (ISH; Fig. 2; n = 5–11; one-way ANOVA followed by Tukey's Multiple Comparison Test)

Gene	DM vs. EM ISH	DM vs. EM Affy	DF vs. EF ISH	DF vs. EF Affy	L vs. EF ISH	L vs. EF Affy
Azin1	5.927 ± 3.367; P = 1E-04	2.31	2.222 ± 0.742; P = 0.0112	2.19	1.680 ± 0.342; P = 0.0269	1.70
Giot1	12.668 ± 5.889; P = 6E-06	12.86	5.174 ± 0.790; P = 6E-06	7.87	4.199 ± 1.033; P = 0.0004	ns
Tnfaip6	21.444 ± 12.224; P = 1.8E-05	11.98	5.950 ± 0.772; P = 5E-06	12.86	2.067 ± 0.481; P = 0.036	ns

Data are compared to differential gene expression identified by Affymetrix microarray analysis (Affy). Data from the male SON [dehydrated male (DM) vs. euhydrated male (EM) (32,35)] is shown for comparison. ns, Not significant; E, exponential.

For the first time, high-resolution radioactive in situ hybridization followed by emulsion dipping was used to further quantify mRNA expression of Giot1, Tnfaip6, and Azin1 in the SON at the cellular level in both male and female rats (Fig. 3). In the SON, grains appeared to predominantly localize in MCN-like cells. The number of cells expressing Giot1 and Tnfaip6 and the level of expression per cell were significantly increased after either dehydration or lactation (Table 2). A significant increase in the number of cells expressing Azin1 and in expression levels per cell was seen in the SON of male rats after dehydration, but not in the SON of dehydrated or lactating female rats.

Fig. 3.

Representative high-resolution in situ hybridization images of Tnfaip6, Giot1, and Azin1 mRNA expression in the euhydrated male (EM), dehydrated male (DM), euhydrated female (EF), dehydrated female (DF), and lactating (L) rat hypothalamus (low resolution) and SON (high resolution). Sections were counterstained with toluidine blue. Sections were exposed to emulsion for 6 wk. Scale bar, 20 μm.

Table 2.

Quantification of changes in Giot1, Tnfaip6, and Azin1 mRNA expression in the SON dehydrated male (DM) compared (vs.) with euhydrated male (EM), dehydrated female (DF) compared with euhydrated female (EF), and lactating (L) female compared with euhydrated female (EF) as assessed by high-resolution in situ hybridization (Fig. 3)

Gene	Analysis method	DM vs. EM fold change	DF vs. EF fold change	L vs. EF fold change
Azin1	Cells	3.500 ± 0.764; P = 0.0112	1.225 ± 0.750; P = 0.640	1.400 ± 0.424; P = 0.280
	Grains/cell	1.830 ± 0.185; P = 0.0033	1.204 ± 0.121; P = 0.171	1.140 ± 0.080; P = 0.372
Giot1	Cells	9.667 ± 0.764; P = 8E-05	15.25 ± 7.587; P = 0.031	12.500 ± 6.016; P = 0.030
	Grains/cell	2.150 ± 0.278; P = 0.003	2.512 ± 0.313; P = 0.001	2.044 ± 0.115; P = 0.00015
Tnfaip6	Cells	3.850 ± 1.146; P = 0.0001	10.198 ± 2.615; P = 0.004	2.800 ± 0.917; P = 0.037
	Grains/cell	2.087 ± 0.333; P = 0.005	2.122 ± 0.481; P = 0.017	0.953 ± 0.013; P = 0.495

The number of expressing cells, and the expression (grains) per cell were assessed. Data were obtained from high-resolution in situ hybridization images of the SON (Fig. 3). Cell counting and grain counting were performed at ×200 and ×400 magnification, respectively, on a Leica DM microscope and digital camera. Only cells that have 5 times the background labeling (5–15 grains per cellular area) were included in the analysis. For each section, 10–15 labeled cells were selected at random for grain counting. Data were analyzed for significance using one-way ANOVA followed by Tukey's Multiple Comparison Test, n = 3 or 4. Bold text highlights significant changes. E, Exponential.

Discussion

For the first time, the transcriptome of the SON of randomly cycling virgin female euhydrated and dehydrated rats as well as female rats after 10–11 d of lactation has been catalogued. To better exploit these novel data, we have also re-mined our previously published microarray data from the SON of male rats under euhydrated and dehydrated conditions (32, 33) using the gene expression software GeneSpring GX11, which allows direct comparison of these extensive data sets. Thus, we have produced gene lists that, with a high degree of confidence, represent the total expressed gene complement of the SON in male and female rats under both euhydrated and dehydrated conditions and, in the case of the female SON, during lactation. Robust statistical comparisons of these gene lists have enabled the identification of genes that are increased or decreased in expression level as a consequence of physiological stimulation.

It is pertinent at this point to highlight the limitations of this analysis. First, although we are confining our analysis to finely dissected specific brain regions, we acknowledge that our samples will be contaminated with neighboring tissues. Similarly, the samples are composed of numerous different cell types, including neurons and glia. Further detailed analysis of a target gene requires confirmation of the anatomical location of its expression using, for example, in situ hybridization or immunocytochemistry. A second caveat relates to the possibility that our use of randomly cycling female rats might introduce variability. Although it would have been preferable to examine each stage of the estrus cycle, issues of cost unfortunately precluded this. The cost of an already expensive experiment (in terms of animal use and microarrays) would have tripled (from three new experimental groups to nine). We therefore used randomly cycling female animals, as has been reported previously for both molecular (e.g. Ref. 36) and electrophysiological studies (e.g. Ref. 37). That said, our array data were obtained from pools of five animals, and replicated four or five times, thus averaging out variation within groups. Indeed, all of the female groups exhibited high correlation coefficients, enabling us to reliably focus on changes in gene expression as a consequence of the physiological cues of dehydration and lactation. Indeed, although mean correlation coefficient for each experimental group in a pair-wise fashion (Fig. 1A) and PCA analysis (Fig. 1B) revealed slightly more variation between the female euhydrated samples and the female dehydrated samples, compared with the data from males and the lactating animals, variation is still low. Thus we are confident that our data enable us to reliably focus on changes in gene expression as a consequence of the physiological cues of dehydration and lactation and to identify common patterns of differential gene expression that define and may mediate SON plasticity. However, in addition to common patterns of differential gene expression, transcriptome analysis of the SON of male and female rats can also reveal sex-specific patterns of gene expression, both in the euhydrated and the dehydrated state, but our use of randomly cycling females precludes further analysis and comment. An additional caveat of our study relates to the upper limit of the sensitivity range of the microarrays. As previously noted (33), but contrary to expectation based on previous reports (4, 38), no significant change in VP gene expression was seen after dehydration in the male rat SON. Rather than being a biological phenomenon, we suggest that this is a technical limitation, with the raw expression level of VP in the control SON being close to the maximum sensitivity of the Affymetrix array. This signal saturation precluded the detection of any significant increase after dehydration. We suggest that the new data presented here also suffer from this sensitivity ceiling with respect to both VP and OT expression, and possibly other genes, after dehydration and lactation. The only way to address this ceiling effect would be to reduce the amount of labeled target used to hybridize to the array probe, but this would reduce the sensitivity at the lower end of the expression level scale. Finally, it must be recognized that, although the microarray analysis provides us with a global picture of the abundance of thousands of specific transcripts, this is but one level at which gene expression and function are regulated.

Technical and statistical issues inherent in array analysis require that the data are tested using independent methodological criteria. We have done this in two ways. First, the literature provides ample evidence to support some of our findings. Lactation-regulated transcripts identified in this study that have previously been described include prodynorphin (Pdyn; 39), Vgf (40), cocaine, and amphetamine-regulated transcript (Cart; 41) and c-Fos (42). There are few studies on chronic dehydration in females, but we note that c-Fos has previously been shown to be up-regulated (36, 43, 44). Second, we have used in situ hybridization to validate three novel targets; Giot1, Tnfaip6, and Azin1, the possible physiological functions of which are discussed in Supplemental Information File 1. Giot1, Tnfaip6, and Azin1 have all been previously shown to be robustly up-regulated in the SON of male rats subject to dehydration (33, 35). Similarly, in accordance with the microarray data, significant up-regulation of all three genes in the SON was observed in the brains of dehydrated female rats compared with corresponding euhydrated controls (Fig. 2, Table 1, Fig. 3, and Table 2). High-resolution in situ hybridization revealed that expression of all three genes was confined to MCN. In male SON, the dehydration-induced increase in the abundance of transcripts encoded by all three genes comprises both an increase in the number of expressing MCN, as well as an increase in the expression per cell (Table 2). This was also the case in female rats for Giot1 and Tnfaip6, but these parameters did not reach significance for Azin1 (Table 2). Less pronounced and consistent changes in the expression patterns of these genes were observed after lactation. The array predicted a 1.7-fold increase in the expression of Azin1 in lactation, and this was confirmed by the in situ hydridization analysis (Fig. 2 and Table 1), although the number of expressing MCN and the expression level per cell did not change significantly (Table 2). For Giot1 and Tnfaip6, the conservative and robust analysis of out array data predicted no significant change in the expression of lactation. However, in situ hybridization revealed small but significant increases in expression but both much lower than the dehydration-induced changes. The number of MCN expressing Giot1, and the level of expression per cell, were significantly increased after lactation (Table 2). Interestingly, for Tnfaip6, the number of expressing cells increased during lactation, but the expression level per cell was unchanged (Table 2). Within the constraints of the caveats discussed above, these data nonetheless suggest that Tnfaip6 and Azin1 have a less important role in SON plasticity in lactation compared with dehydration. In this context, it should be noted that dehydration and lactation are distinct stimuli with different outcomes. In dehydration the firing rate of VP cells increases, changing from a slow, irregular mode to a phasic firing pattern (45). During suckling, nipple stimulation gives rise, via the spinal pathway of the milk-ejection reflex (46), to a synchronized short high-frequency burst of action potentials in most OT cells (47), which precedes bolus OT release (48). However, it should be noted that lactation also appears to be an osmotic stimulus (10). Thus it is to be anticipated that gene expression responses within the SON in response to dehydration and lactation will show similarities and differences, as noted above. This concept has support in the literature; corticotropin-releasing factor (Crf) mRNA has previously been shown to be up-regulated in response to osmotic stimulation but not 10-d lactation (49, 50), despite its localization in OT neurons (51), an observation confirmed by our array data.

For the first time, we have revealed changes in global gene expression patterns in the SON that accompany dehydration and lactation in the female rat. By comparing these gene lists with previously published data on dehydration in the male rat, we have identified 2070 genes that are regulated in at least two of these conditions (Supplemental Appendix, page 9). In an attempt to place these differentially expressed genes into physiological and functional context, and to identify genes that might underpin plasticity within the SON, we have explored gene ontology (GO) terms associated with genes in this list using a GO enrichment algorithm in GeneSpring GX11. In total, six GO were considered enriched in this list (Table 3), relating to 72 probe sets from 49 genes (see Supplemental Table 2). We can assume that some of these changes are related to the necessity for increased protein synthesis and hormone delivery in the physiologically stimulated MCN (translation initiation factor activity and endoplasmic reticulum-Golgi intermediate compartment, respectively). Further, we note with great interest that four enriched GO terms represent molecular functions, biological processes, and cellular components that are consistent with the concept of morphological remodeling and extracellular organization within this tissue as a result of physiological plasticity (24–26), namely collagen fibril organization, extracellular matrix organization and biogenesis, extracellular structure organization and biogenesis, and homophilic cell adhesion. We suggest that the genes coordinately regulated in the SON as a consequence of dehydration and lactation form a network that mediates the plastic processes operational in the physiologically activated SON. Functional studies will emerge from these comparative transcriptome analyses that will address the physiological role of these genes in the processes that facilitate efficient peptide release in dehydration and lactation.

Table 3.

GO exploration of common differentially expressed genes

GO ID	GO Accession	GO term	Corrected P value	Count in selection	P value	% Count in selection	Count in total	% Count in total
1925	GO:0003743	Translation initiation factor activity	0.046967838	17	7.55E-06	22.368422	50	0.41240513
3490	GO:0005793	ER-Golgi intermediate compartment	0.019212848	12	2.57E-06	15.789474	25	0.20620257
4648	GO:0007156	Homophilic cell adhesion	0.014607531	26	1.17E-06	34.210526	90	0.74232924
11691	GO:0030198	Extracellular matrix organization and biogenesis	0.009371977	23	5.02E-07	30.263159	76	0.62685585
11692	GO:0030199	Collagen fibril organization	0.003857188	12	1.03E-07	15.789474	20	0.16496205
15653	GO:0043062	Extracellular structure organization and biogenesis	0.015612268	23	1.67E-06	30.263159	120	0.9897724

The GO Spreadsheet shows the following for each GO term displayed: P value, the probability of obtaining the specified GO accession number from a list of random entities (the smaller the P value, the more significant is the GO accession number); Corrected P value, as multiple GO accession number are tested for their significance, a multiple testing correction is performed (note that the GO spreadsheet is sorted on the basis of this parameter); Count in selection, the number of genes in the selected entity list that have that particular GO term; % Count in selection, the percentage of genes in the input entity list which have that GO term; Count in total, the number of genes in all entities which have that GO term; % Count in total, the percentage of genes in the All entities list that have that GO term. ER, Endoplasmic reticulum.