Combined proteomics and pathways analysis of collecting duct reveals a protein regulatory network activated in vasopressin escape

Abstract

Low sensitivity is characteristic of many proteomics methods. Here we present an approach that combines proteomics based on “Difference Gel Electrophoresis” (DIGE) with bioinformatic pathways analysis to identify both abundant and relatively non-abundant proteins in inner medullary collecting duct (IMCD) altered in abundance during escape from vasopressin-induced antidiuresis. Rats received the vasopressin analog dDAVP by osmotic minipump plus either a daily water load (vasopressin escape) or only enough water to replace losses (control). Immunoblotting confirmed the hallmark of vasopressin escape, a decrease in aquaporin-2, and demonstrated a decrease in the abundance of the urea transporter UT-A3. DIGE identified 22 mostly high abundance proteins regulated during vasopressin escape. These proteins were analyzed using pathways analysis software to reveal protein clusters that include those identified by DIGE. A single dominant cluster emerged that included many relatively low abundance proteins (abundances too low for DIGE identification) including several transcription factors. Immunoblotting confirmed a decrease in total and phosphorylated c-myc, a decrease in c-fos, and increases in c-jun and p53. Furthermore, immunoblotting confirmed hypothesized changes in other proteins in the proposed network: increases in c-src, RACK1, calreticulin, and caspase 3, and decreases in SRC-1, Grp78/BiP, and annexin A4. This combined approach proved capable of uncovering regulatory proteins altered in response to a specific physiological perturbation without being directly detected by DIGE. The results demonstrate a dominant protein regulatory network in IMCD cells that is altered in association with vasopressin escape, providing a new framework for further studies of signaling in IMCD.

INTRODUCTION

Escape from the antidiuretic action of vasopressin (“vasopressin escape”) is an important physiological process that limits the severity of the syndrome of inappropriate antidiuresis (SIADH) and other hyponatremic disorders (1; 2). During vasopressin escape, humans and experimental animals undergo a brisk water diuresis, despite high circulating levels of vasopressin (2–4). Vasopressin escape is of considerable clinical importance with regard to SIADH and other vasopressin-dependent dilutional hyponatremic states, because without escape further water retention and hyponatremia could be fatal (4–6).

Studies in a rat model of vasopressin escape have demonstrated that the central feature of the vasopressin-independent increase in water excretion is a marked suppression of the expression of the water channel aquaporin-2 (AQP2) (2). The fall in AQP2 protein abundance is in part due to decreased levels of AQP2 mRNA in collecting duct (2). This response is associated with a decrease in the capacity of inner medullary collecting ducts (IMCDs) to produce cyclic AMP in response to vasopressin (7) in association with a fall in vasopressin-binding capacity of the V2R receptor (8). In contrast, there is upregulation of aquaporin-3 (2) and the α-subunit of the epithelial Na channel (ENaC) (9), suggesting that the process responsible for suppression of AQP2 expression is selective. With regard to possible mediators of escape, roles have been suggested for nitric oxide and prostaglandins (10), and aldosterone (11). However, little is known about what intracellular signaling processes orchestrate the escape and how the vasopressin escape phenomenon is triggered and maintained.

To address which proteins are co-regulated with AQP2 in IMCD during vasopressin escape, we employed a relatively new differential proteomics method called DIfference Gel Electrophoresis (DIGE) coupled with MALDI-TOF/TOF mass spectrometry to identify differentially expressed proteins. Recently, we demonstrated the feasibility of such an approach using DIGE-based proteomics to identify vasopressin-responsive proteins in the collecting duct of Brattleboro rats that were treated with the selective V2R agonist dDAVP (12).

To address further the signaling pathways that are activated during vasopressin escape, we analyzed the DIGE results using bioinformatic pathways analysis (13; 14). The pathways analysis software generates hypothetical protein networks, based on large databases of protein interactions culled from the biological literature, including physical binding reactions, cis-trans interactions in transcriptional regulation, and enzyme-substrate relationships. Such networks can be used to predict signaling pathways that are activated during vasopressin escape. The rationale for using the pathways analysis approach was not only to facilitate the interpretation of the relationships between the identified proteins, but also to identify relatively low abundance proteins (abundances too low for DIGE identification) that may be involved in vasopressin escape (15). The hypothetical changes in these low abundance proteins can then be tested by immunoblotting. This integrated approach identified a single dominant network of proteins including several proteins that can be proposed to play key regulatory roles in vasopressin escape.

MATERIALS AND METHODS

Animals and sample preparation

Male Sprague-Dawley rats (Taconic Farms, Germantown, NY) were implanted subcutaneously with osmotic minipumps (model 2001; Alzet), delivering 5 ng/hr dDAVP (Peninsula Laboratories) (ACUC Protocol 2-KE-3). After three days, rats were divided into two groups: “escape” and “control”. Escape rats were given excess daily water (0.4 ml/gBW) via a gelled-agar diet (71% water, 28% finely-ground rat chow, 1% agar, BACTO-AGAR; Difco Laboratories, Detroit, MI). This diet forced the rats to take the water load to consume the food ration. Control rats were given the same amount of food but with only enough water (0.075 ml/gBW) to compensate for insensible losses plus 0.015 ml/gBW/day of urine. Rats did not receive ad libitum water. This represents a small modification from our previous studies in which control rats received no water mixed with the food but were allowed to receive ad libitum water (2; 7; 9). The rats were maintained in metabolic cages in a temperature- and humidity-controlled room with a 12:12 h light dark cycle and urine was collected daily for measurement of volume and osmolality. Because the onset of escape is known to occur between 1 and 2 days after the start of water loading (2), three time points were analyzed (4 control vs. 4 escape for each time-point): after 1 and 2 days of water loading (early stages of escape) and after 4 days of water loading (late stages of escape). Thus, a total of 24 rats were separated into 12 control rats and 12 “escape” rats, and 4 vs. 4 rats were selected arbitrary for IMCD analysis at each of the three time points. The day 1 and 4 time points were selected for DIGE analysis, whereas all three time-points were analyzed by immunoblotting. IMCD suspensions were prepared using the method of Stokes et al. (16) with some modifications (17) (See Supplementary Materials).

Semi-quantitative immunoblotting

Immunoblotting was carried out as described previously (18). (See Supplementary Materials). The IMCD pellet was solubilized in 5× Laemmli sample buffer (1 vol per 4 vol of sample) followed by heating to 60°C for 15 minutes prior to electrophoresis. Equal loading was confirmed by staining gels loaded for all three time-points (24 samples) with Coomassie blue (18). This loading gel was scanned with a linear fluorescence scanner (Odyssey, Li-Cor Biosciences) at an excitation wavelength of 700 nm (Supplementary Figure 1).

Figure 1. Urine excretion rate, urine osmolality, plasma sodium concentration and plasma urea concentration during vasopressin escape.

A. Urine excretion rate over the course of the experiment with water loading commencing on day 0. Urine excretion rate was increased significantly from day 2 on in the escape group relative to control (4 vs. 4 rats/time-point). Negative time-points represent the equilibration period. B. Urine osmolality over the course of the experiment. Osmolality was decreased significantly from day 1 to the end of the experiment in the escape group relative to control (4 vs. 4 rats/time-point). C. Plasma sodium concentrations at days 1, 2 and 4. Plasma sodium was significantly lower in the escape group relative to control on all three days (4 vs. 4 rats/time-point). D. Plasma urea concentrations at days 1, and 4. Plasma urea was significantly lower in the escape group relative to control on both days (4 vs. 4 rats/time-point). * P < 0.05 for all.

Difference gel electrophoresis (DIGE)

DIGE analysis was carried out (for day 1 and 4 time-points, each 4 vs. 4 samples) as previously described (12; 19). (See Supplementary Materials). Briefly, prior to 2-D gel electrophoresis, IMCD proteins were solubilized in 2D sample buffer (7 M urea, 2 M thiourea, 30 mM TrisCl and 4% CHAPS, pH 8.5). The samples were labeled on lysine side chains with Cy3- (control), Cy5- (escape), or Cy2- (mixture of control + escape samples, internal standard) fluorophores using N-hydroxysuccinamide chemistry (Amersham). Isoelectric focusing was performed using an IPGphor apparatus (Amersham). Isoelectric focusing strips were loaded onto Ettan DALT-six electrophoresis unit (Amersham) and further separated on a 10% SDS-PAGE gel (5 W/gel).

Fluorescent analytical gel images were obtained (Typhoon scanner; 100 μm resolution; Amersham) using the following emission filters: Cy2 (520 BP 40), Cy3 (580 BP 30), and Cy5 (670 BP 30). Spot matching, quantitation, and statistical analyses were performed using DeCyder software (Version 5.0; Amersham). The corresponding Cy3 (control) and Cy5 (escape) images were normalized to the pooled internal standard (Cy2) for that gel using a Least-Means-Squared-Gradient-Descent algorithm. One gel was chosen as the “master” and all remaining analytical gels were matched and normalized to the Cy2 master spot map. The resulting protein abundance ratios, now represented as standardized log abundance values, were compared using an unpaired Student’s t-test. The inverse log of these values is presented in Table 1 as protein abundance ratio. A protein abundance ratio > 1 corresponds to an increase in escape compared to control samples, while a ratio < 1 corresponds to a decrease in escape (significance criterion, p ≤ 0.05). For picking, gels were fixed in 30% ethanol/7.5% acetic acid for 2 h followed by SYPRO Ruby (610 BP 30) staining overnight for total protein visualization.

Table 1.

IMCD proteins regulated at early (day 1) or later (day 4) stages of vasopressin escape

Protein Abundance Ratio * (Escape/Control)	Protein Identification	MW (kDa)	pI	GenBank No.
1A Decreased early in vasopressin escape (Day 1)
0.39	ATP synthase β	50	5.0	gi|54792127
0.40	Keratin complex 2, gene 8	51	5.7	gi|40786432
0.42	Mitochondrial aconitase	86	7.9	gi|10637996
0.43	NADH dehydrogenase	81	5.5	gi|51858651
0.51	Alpha B-crystallin	20	6.3	gi|16905067
0.58	GRP 75, stress 70 protein	75	5.9	gi|55584140
0.62	β-Tubulin-1	51	4.9	gi|56754676
0.63	Annexin A2	39	9.2	gi|584760
0.66	Malate dehydrogenase 1	37	6.2	gi|15100179
0.71	F-actin capping protein β	31	5.7	gi|53734561
0.73	Alpha enolase	52	6.7	gi|56757324
0.74	Pyruvate kinase	58	7.2	gi|125601
0.79	Transketolase	68	7.2	gi|12018252
0.82	Eukaryotic translation	96	6.4	gi|8393296
	elongation factor 2
0.82	Alpha-1-antiproteinase	46	5.7	gi|1703024
0.83	Prohibitin	29	5.6	gi|34873234
0.88	Sec23B	87	6.5	gi|34858849
0.90	T-complex protein 1	58	6.6	gi|6981642
Decreased later in vasopressin escape (Day 4)
0.71	Pyruvate kinase	58	7.2	gi|125601
0.76	Heat shock protein 70	70	5.5	gi|450932
0.81	Transketolase	68	7.2	gi|12018252
1 B Increased early in vasopressin escape (Day 1)
1.28	Calreticulin	48	4.3	gi|11693172
1.93	Pyruvate dehydrogenase β	35	5.6	gi|1352624
1C Spot shifted early in vasopressin escape (Day 1)
0.19/25.15	Protein disulfide-isomerase	57	5.9	gi|1352384

^*Calculated from average protein abundance ratios for 4 pairs of samples. Only protein spots that showed statistically significant changes by DIGE analysis were picked for MALDI-TOF/TOF identification. All of the identified proteins were in the expected ranges of pI and MW in the 2-D gel in Figure 3.

A robotic workstation (Ettan; Amersham) was used to excise protein spots, perform in-gel tryptic digestion, extract peptides from the gel, and transfer the extracts onto a MALDI substrate. Spectra were acquired with an ABI 4700 MALDI-TOF/TOF mass spectrometer and proteins identified by database matching using Mascot.

Bioinformatic pathways analysis

Regulated proteins identified by DIGE were analyzed further by bioinformatic pathways analysis (Ingenuity Pathway Analysis [IPA]; Ingenuity Systems, Mountain View, CA, www.ingenuity.com). IPA constructs hypothetical protein interaction clusters based on a regularly updated “Ingenuity Pathways Knowledge Base”. The Ingenuity Pathways Knowledge Base is a very large curated database consisting of millions of individual relationships between proteins, culled from the biological literature. These relationships involve direct protein interactions including physical binding interactions, enzyme substrate relationships, and cis-trans relationships in translational control. The networks are displayed graphically as nodes (individual proteins) and edges (the biological relationships between the nodes).

In practice, a dataset containing the GenBank identifiers of differentially expressed proteins identified in the DIGE experiment is uploaded into IPA. IPA then builds hypothetical networks from these proteins and other non-DIGE-identified proteins from the database that are needed fill out a protein cluster. Network generation is optimized for inclusion of as many proteins from the inputted expression profile as possible, and aims for highly connected networks.

IPA computes a score for each possible network according to the fit of that network to the inputted proteins. The score is calculated as the negative base-10 logarithm of the p-value that indicates the likelihood of the inputted proteins in a given network being found together due to random chance. Therefore, scores of 2 or higher have at least a 99% confidence of not being generated by random chance alone. For previous studies using IPA, see (13; 14).

RESULTS

Verifying vasopressin escape

In this model of vasopressin escape, both control and experimental rats received a continuous dDAVP infusion starting on day -3, but only the experimental rats received a daily water load, mixed with the food, starting on day 0. As previously noted (2; 7; 9), rats began to “escape” from dDAVP-induced antidiuresis on the second day, i.e. 24–48 hours after initiation of water loading, as evidenced by a marked increase in urine excretion rate (Figure 1A). Urine osmolality (Figure 1B) was reduced significantly in escape on the second day of water loading. Plasma sodium levels (Figure 1C), showed an acute decrease between days 1 and 2 (from 137 ± 2 to 105 ± 3 mmol/l), and subsequently, a partial recovery on day 4 (115 ± 3 mmol/l) in response to vasopressin escape. Plasma urea levels were significantly lower in escape animals both at early (day 1: 7.0 ± 0.7 vs. 5.0 ± 0.4 mmol/l) and late (day 4: 7.0 ± 0.5 vs. 5.0 ± 0.4 mmol/l) stages of vasopressin escape (Figure 1D).

Changes in abundances of transport proteins in IMCD

Figure 2 shows immunoblots for AQP2, α-ENaC and collecting duct urea transporters in IMCD on day 4 of vasopressin escape and densitometry values for all three time-points. Downregulation of AQP2 and upregulation of α-ENaC is consistent with previous studies (7; 9). A novel finding was that UT-A3 showed a 50% decrease in band density on day 2 of vasopressin escape, and was further decreased on day 4. Conversely, UT-A1 did not show a significant decrease.

Figure 2. Immunoblots showing changes in the abundances of AQP2, α-ENaC, and collecting duct urea transporters in IMCDs from rats undergoing vasopressin escape vs. control rats.

Immunoblots are of IMCD cells purified from rat renal medullas at late stage of vasopressin escape (day 4 time-point). Each lane is loaded with a sample from a different rat (n = 4 rats per treatment). Thirty μg of total protein were loaded in each lane, and the resulting immunoblots were probed with anti-AQP2 antibody (L127), anti- α-ENaC (Q3560-2), anti-UT-A3 (Q2695-2), or anti-UT-A1 (L403). To the right of each immunoblot is a bar graph showing densitometry values for all three time-points studied. * P < 0.05.

Changes in the rat IMCD proteome in vasopressin escape

Figure 3 shows an example of a DIGE gel with superimposition of Cy3 (control, red) and Cy5 (escape, green) images of the gel. Spots corresponding to proteins expressed at nearly equal levels in the two samples appear yellow, those upregulated in response to vasopressin escape appear green and those downregulated in response to vasopressin escape appear red. Flanking the 2-D gel image are three-dimensional pixel density plots for 9 selected proteins that were identified by MALDI-TOF/TOF mass spectrometry, including heat-shock protein 70 (HSP70), ATP synthase, calreticulin, prohibitin, mitochondrial aconitase, Sec23B, annexin A2, malate dehydrogenase, and protein disulfide isomerase (PDI). The latter protein had an apparent shift in iso-electric point, suggestive of a post-translational modification. Only those protein spots with statistically significant abundance ratios (p ≤ 0.05 for 4 pairs of samples) were selected for MALDI-TOF/TOF identification. Moreover, only those identifications with expectation values (i.e. an indicator of the degree of certainty of an identification) larger than 95% were accepted. A total of 22 protein spots was identified by MALDI-TOF/TOF mass spectrometry (Table 1). More proteins with altered abundance levels were identified at day 1 than at day 4. Those identified included proteins that were downregulated (Table 1A), upregulated (Table 1B) and one protein that shifted position in the gel (Table 1C, Figure 3) in response to vasopressin escape. Also listed in Table 1 are the theoretical molecular weights (MW), theoretical isoelectric points (pI values). The molecular weights and pI values for all of these proteins matched those derived from the spot position on the gel, providing additional verification of the identifications.

Figure 3. 2-D gel showing changes in the IMCD proteome in vasopressin escape.

Superimposed images from samples labeled with Cy3 (control, red pseudocolor), and Cy5 (escape, green pseudocolor) and three dimensional representation of spot intensities. Spots that appear red and green represent proteins that are respectively down- and upregulated in vasopressin escape, whereas proteins that are equally abundant in both samples appear yellow. Full range of horizontal axis is from 3 pH units (left) to 10 pH units (right). Full range of vertical axis is 15 kDa (bottom) to 120 kDa (top). pI, isoelectric point; MW, molecular weight.

Pathways analysis of vasopressin escape

Figure 4 shows the largest protein cluster that was generated by the pathways analysis of the proteins listed in Table 1. This “vasopressin escape cluster” consists of a network of 33 proteins, including 8 of the 22 proteins that were identified by DIGE-based proteomics, and 25 additional proteins, that were recognized as being related because of their reported interactions with the proteins identified by DIGE. The nodes represent individual proteins listed by gene name, while the edges represent the interactions, which include direct physical binding, substrate-enzyme interactions, and/or cis-/trans-relationship in transcriptional regulation. In Figure 4, nodes are displayed using different gray levels that represent how the protein was identified and studied. Proteins identified by DIGE only are represented as light gray nodes, while proteins identified by DIGE and further studied by immunoblotting are represented as black nodes. Proteins identified by IPA only are represented as white nodes, while proteins identified by IPA and further studied by immunoblotting are represented as dark gray nodes. Finally, the index protein AQP2 is shown. Because we were chiefly interested in identifying candidate proteins for follow-up by immunoblotting, we did not discriminate between the early and late time-points for the pathways analysis. Annotation of the interactions in Figure 4 is provided as Supplementary Materials.

Figure 4. Protein regulatory network associated with vasopressin escape.

A. Protein regulatory network generated by bioinformatic pathways analysis through the use of the Ingenuity Pathways Analysis (IPA) software. Proteins listed in Table 1 were analyzed. Individual proteins are displayed as nodes, utilizing different shades of gray to represent how the protein was identified and studied. Proteins identified by DIGE only are represented as light gray nodes, while proteins identified by DIGE and further studied by immunoblotting are represented as black nodes. Proteins identified by IPA only are represented as white nodes, while proteins identified by IPA and further studied by immunoblotting are represented as dark gray nodes. In addition, different shapes are used to represent the functional class of the gene product (see figure insert). The edges describe the nature of the relationship between the nodes: an edge with arrow-head means that protein A acts on protein B, whereas an edge without an arrow-head represents binding only between two proteins. P indicates phosphorylation as a special case of the former. An edge without an arrow-head represents binding only between two proteins. The gene names associated with the proteins are shown; see accompanying table for protein names. See Supplementary Materials for detailed descriptions of individual protein-protein interactions, including literature references. The overall score for the depicted network was 34 indicating that the probability of matching the indicated proteins by a purely random event was 10⁻³⁷. B: Glossary of proteins in network.

Confirmation by semi-quantitative immunoblotting

Four of the 22 proteins identified by DIGE were selected for semi-quantitative immunoblotting based on the availability of high quality antibodies. These were: PDI, calreticulin, β-tubulin, and HSP70 (Figure 5). Although DIGE and immunoblotting gave percent changes that differed somewhat for these four proteins, in general, immunoblotting confirmed the qualitative responses detected with the DIGE technique, as found previously (12; 19). Of the 25 proteins identified by pathways analysis as potentially involved in vasopressin escape (Figure 4), 10 were selected for semi-quantitative immunoblotting, based on the central positions in the protein cluster and on antibody availability. One of these is c-myc, a transcription factor that is regulated in part by phosphorylation. At early stages of escape, total c-myc abundance was decreased, whereas at later stages, the abundance of the phosphorylated form of c-myc was decreased (Figure 6). The abundances of four additional proteins identified by pathways analysis were decreased (Figure 7A: annexin A4, GRP78/BiP, SRC-1, and c-fos), whereas the abundances of five other proteins were increased (Figure 7B: caspase 3, p53, RACK-1, c-src, and c-jun). The abundances of the proteins shown in Figures 5, 6, and 7 were changed only at the reported time-point, and were unchanged at the other time-points (not shown). To address whether the protein abundance changes are specific for the IMCD or occur in other tissues, we immunoblotted renal cortex and brain samples from the same experiments and probed with selected antibodies (Figure 8). Among the responding proteins in IMCD, only c-src showed a similar response in renal cortex, and only HSP70 showed a similar response in brain.

Figure 5. Immunoblots confirming selected protein abundance changes identified by DIGE-based proteomics.

IMCD suspensions, prepared at the day 1 and day 4 time-points. Each lane is loaded with a sample from a different rat (n = 4 rats/treatment). Thirty μg of total protein were loaded in each lane, and the resulting immunoblots were probed with (for Day 1) anti-protein disulfide isomerase (PDI), anti-calreticulin, anti-β-tubulin, and (for Day 4) anti-heat shock protein 70 (HSP70). the apparent change in mobility of beta-tubulin may represent an unidentified post-translational modification. * P < 0.05.

Figure 6. Immunoblots of IMCD proteins using antibodies to total and phosphorylated c-myc at early and late stages of vasopressin escape.

Each lane is loaded with a sample from a different rat (n = 4 rats/treatment). Thirty μg of total protein were loaded in each lane, and the resulting immunoblots were probed with anti-c-myc and anti-phosporylated-c-myc. * P < 0.05.

Figure 7. Immunoblots for selected proteins identified by pathways analysis.

IMCD proteins identified by pathways analysis were immunoblotted (30 μg of total protein per lane). Immunoblots show the regulation of proteins at late stages of vasopressin escape (day 4 time-point); all proteins were unchanged at early stages of vasopressin escape (day 1, not shown). Each lane is loaded with a sample from a different rat (n = 4 rats/treatment). Immunoblots were probed with anti-annexin A4, anti-GRP78/BiP, anti-SRC-1, anti-c-fos (Figure 7A), and anti-caspase 3, anti-p53, anti-RACK1, anti-c-jun (Figure 7B). * P < 0.05.

Legend Figure 8.

Immunoblots for selected proteins in renal cortex and brain tissue during vasopressin escape (day 4). Each lane is loaded with a sample from a different rat (n = 4 rats/treatment), using renal cortex (A) and whole brain (B) homogenates. Thirty μg of total protein were loaded in each lane, and the resulting immunoblots were probed with anti-calreticulin, anti-annexin A4, anti-HSP70, anti-GRP78/BiP, anti-c-src, and anti-RACK1. * P < 0.05

DISCUSSION

Proteomics analysis is seeing increasing use as a means of identifying new mechanistic hypotheses in physiology (20). An important disadvantage of all 2-D gel-based proteomics approaches is that low abundance proteins, including many cell signaling proteins, are unlikely to be identified (15; 21). Hence, alternative approaches are needed to discover transcription factors and other regulatory proteins involved in the responses to physiological perturbations. In the present study, we used a combination of DIGE-based proteomics and pathways analysis to identify a protein regulatory network whose state is altered in association with the vasopressin escape process and the associated downregulation of AQP2 expression. Proteins identified by DIGE were used as input data for bioinformatic pathways analysis, which pointed to several additional proteins that could hypothetically be involved in the escape process. These additional proteins corresponded individually to ‘hypotheses’ that could be tested by immunoblotting asking whether each protein was up- or down-regulated as part of the escape process. Indeed, 10 of 10 proteins tested in this way displayed the hypothesized regulation (Figures 6 and 7). The outcome of this integrated approach was a cluster of proteins, discussed below, that could participate in the onset and maintenance of escape or alternatively in secondary responses following vasopressin escape. Most of the demonstrated protein abundance changes occurred only in IMCD, and not in renal cortex or brain, indicative of a selective response in the IMCD.

UT-A3, but not UT-A1, is downregulated during vasopressin escape

As a preliminary step in these experiments, we carried out immunoblotting to verify the downregulation of aquaporin-2 seen in previous studies (2; 9). The present study confirmed the decrease in AQP2 expression and showed that the α-subunit of ENaC is upregulated in IMCD in the same manner as in more proximal parts of the collecting duct (9). In addition, we have extended our studies to collecting duct urea transporter abundances and demonstrated that UT-A3, but not UT-A1, is markedly decreased in abundance during vasopressin escape (Figure 2). The decrease occurred in parallel with the reduction in AQP2. Since both UT-A1 (unchanged) and UT-A3 (downregulated) share the same transcription start site and upstream regulatory regions (22), it is unlikely that transcriptional regulation is the basis of the decrease in UT-A3 expression. Instead, recognizing that the 3′ end of UT-A1 and UT-A3 transcripts differ (22), it seems possible that the differential regulation of these two proteins is based on the different 3′ ends of the mRNAs. Both mRNA stability regulation and translational regulation are based on specialized processes involving the 3′ end of mRNA molecules (23), raising the hypothesis that either of these mechanisms may be involved in UT-A3 regulation.

The observed downregulation of UT-A3 could contribute to the increase in water excretion seen during the escape process by a mechanism similar to that demonstrated in knock-out mice lacking both UT-A1 and UT-A3 (24). The knockout mice developed a urinary concentrating defect because urea failed to accumulate in the inner medulla, resulting in a urea-dependent osmotic diuresis. Vasopressin escape is associated with modest extracellular volume expansion and hypertension (10). This ECF volume expansion could play a role in the decreased expression of collecting duct urea transporters, as suggested in a prior study implicating aldosterone/salt-induced extracellular volume expansion in regulation of collecting duct urea transporter expression (25). In the present study, serum urea levels were markedly decreased as seen previously (11), a presumed consequence of UT-A3 downregulation.

The abundances of several transcription factors are altered during the vasopressin escape process

Amidst the protein regulatory network identified in this study are several transcription factors, whose cellular abundances are presumably too low to be detected via the DIGE technique. Several of these were demonstrated to undergo changes in abundance by immunoblotting. The earliest transcription factor to exhibit abundance changes was c-myc, whose abundance was found to be decreased just prior to the increase in water excretion (day 1). c-Myc is a basic helix-loop-helix leucine-zipper protein that binds as a heterodimer to so-called E-box cis-elements. The presence of three such E-box elements in the 5′-flanking region of the AQP2 gene (26) raises the possibility that c-myc abundance changes could contribute to the fall in AQP2 expression during the escape response. Although c-myc is best known as a tumor promotor oncogene, it also has physiological functions in all cells that appear to be related to the state of differentiation and proliferation through general regulation of transcription and translation (27). Its role in translational regulation has been recently revealed in DNA array studies, which showed that changes in c-myc levels are associated with parallel changes in the expression of a host of ribosomal proteins (28). The demonstrated fall in c-myc expression may therefore be expected to be associated with a decrease in total cellular transcription and translation, at least at early stages of escape.

At later stages of vasopressin escape, we found a decrease in phosphorylated c-myc. The antibody recognizes c-myc phosphorylated at threonine-58 and serine-62. These modifications result in a decrease in half-life of the c-myc protein thereby increasing the abundance of total c-myc. The phosphorylation is mediated largely by two kinases: c-jun N-terminal kinase (29) and glycogen synthase kinase-3 (30). The decrease in phosphorylation may play a role in the restoration of total c-myc toward control levels on day 4 of the vasopressin escape protocol.

In addition to c-myc, several other transcription factors were identified whose abundances were altered at later stages of vasopressin escape, viz. c-fos (decreased), c-jun (increased), and p53 (increased). In addition, the abundance of a transcriptional cofactor, the steroid receptor co-activator 1 (SRC-1), was decreased.

c-Fos and c-jun together form the transcription factor called “activated protein 1” (AP-1), for which there is an enhancer site in the 5′-flanking region of the human AQP2 gene (31–33). Previously, we demonstrated that c-fos and c-jun are upregulated in the rat renal inner medulla in response to long-term vasopressin infusion (34). Prior studies have demonstrated that AP-1 and a cAMP response element are necessary for maximal transcriptional activation of the AQP2 gene in response to increased intracellular cAMP (31). Conceivably, the demonstrated decrease in c-fos expression contributes to the fall in AQP2 expression in vasopressin escape. c-Fos is itself regulated in part via a cyclic-AMP binding element in its 5′-flanking region (33). SRC-1 is a transcriptional coregulator that interacts with AP-1 and other transcription factors to mediate transcriptional regulation (35; 36).

The abundances of several other regulatory proteins are altered during vasopressin escape

Aside from transcription factors, the protein regulatory network identified by pathways analysis included several other regulatory proteins that potentially could play a role in vasopressin escape. One of these was c-src, a non-receptor tyrosine kinase, whose abundance was increased 3-fold during late stages of vasopressin escape (Figure 7). c-Src is a critical protein in the coupling between G-protein coupled receptors and mitogen-activated protein kinase (MAPK) pathways (37). c-Src as well as some of its substrates bind to β-arrestin 2 (38), a key protein in the internalization of the V2R. Note that c-src was also increased in renal cortical samples (Figure 8), suggesting that c-src upregulation may have been due to a systemic factor. Similarly, the heat shock protein HSP70 was decreased in abundance not only in IMCD, but also in brain, perhaps related to the fall in systemic tonicity (39). Another protein whose abundance was upregulated was the Receptor for Activated C Kinase 1 (RACK1). Its function appears to be broader than its name suggests, since it constitutes a component of the ribosome, and thus may play a role in translational regulation (40). RACK1 is also anti-apoptotic, and a binding partner for c-src (41). Finally, many of the identified regulated proteins also play a role in the endoplasmic reticulum (ER) stress response, including calreticulin, GRP78/BiP, PDI, and caspase 3. ER stress results from situations in which the protein folding capacity of the ER is exceeded such as generalized acceleration of translation (42).

In conclusion, combined proteomics and pathways analysis served to identify a protein network associated with vasopressin escape, containing both high and low abundance proteins. The network included several transcription factors that may be involved in vasopressin escape as well as other relatively low abundance regulatory proteins. These findings provide a new framework for the study of AQP2 regulation in the collecting duct critical to the understanding of SIADH and other forms of hyponatremia.

SUPPLEMENTARY MATERIALS

Supplementary Methods

IMCD sample preparation

After harvesting the kidneys from the animals, inner medullas were dissected, finely minced with a razor blade and transferred to a 12 × 75-mm glass tube, containing tubule suspension solution (118 mM NaCl, 5 mM KCl, 4 mM Na₂HPO₄, 25 mM NaHCO₃, 2 mM CaCl₂, 1.2 mM MgSO₄, 5.5 mM glucose, 5 mM sodium acetate) supplemented with 2 mg/ml collagenase B (Boehringer Mannheim, Indianapolis, IN) and 540 U/ml hyaluronidase (Worthington Biochemical, Freehold, NJ). Suspensions were incubated at 37ºC with 95% air - 5% CO₂ superfusion. Every 15 minutes, suspensions were aspirated with a large-bore Pasteur pipette until all large tissue clumps were digested (75–90 minutes). Then, suspensions were centrifuged at 50 × g, after which the supernatant was discarded and the pellet was resuspended in tubule suspension solution. The procedure was repeated twice, and after the third centrifugation, the resuspended pellet was centrifuged at 1,000 × g for 5 min. Previously, we have demonstrated that this procedure gives an approximately 10-fold enrichment of the collecting duct marker AQP2 and that the resulting pellet is contaminated with only very small amounts of non-IMCD cells (17).

Difference gel electrophoresis

For sample labeling, 50 μg of protein were mixed with 400 pmol of NHS-Cy dye and incubated on ice in the dark for 30 min. 50 μg of each Cy3 and Cy5-labeled protein as well as 50 μg of pooled internal standard (Cy2) were mixed with 50–900 μg of unlabeled protein. Rehydration buffer (7 M urea, 2 M thiourea, and 4 % CHAPS, 2 % Dithiothreitol (DTT), 2 % pH 3-10 Pharmalyte^™) was added to give a total volume of 450 μl. This was loaded on a 24 cm Immobiline DryStrip (pH 3-10 linear) by active rehydration for 10 h (Ettan IPGphor, Amersham). Isoelectric focusing was accomplished by subsequently applying 500 V for 500 V-hrs, 1000 V for 1000 V-hrs and 8000 V for 62533 V-hrs. After isoelectric focusing, strips were equilibrated in SDS equilibration buffer (50 mM TrisHCl pH 8.8, 6 M Urea, 30 % glycerol, 2 % SDS, 0.025 % bromophenol blue) containing 0.5 % DTT for 10 min at room temperature followed by incubation in SDS equilibration buffer containing 2.5 % iodoacetemide for 10 min at room temperature.

Semi-quantitative immunoblotting

The following affinity-purified primary antibodies were used, AQP2, α-ENaC, UT-A1, and UT-A3, which were all characterized previously. Other primary antibodies were obtained from Santa Cruz Biotechnologies (Santa Cruz, CA): annexin A4 [sc-1930], HSP70 [sc-1060]), c-fos [sc-253], c-jun [sc-45], RACK1 [sc-17754], GRP78 [sc-1050], phosphorylated c-myc (Thr 58/Ser 62)-R [sc-8000-R], c-myc [sc-42], c-src [sc-19], β-tubulin [sc-5274]; Upstate (Lake Placid, NY): caspase 3 [06–735], SRC-1 [05–522]; Sigma (St. Louis, MO): protein disulfide isomerase [P7496]. Additional antibodies were supplied by independent investigators: p53 (Dr. N.I. Dmitrieva), calreticulin (Dr. M. Michalak). Antibodies were diluted in a diluent containing 50 mM NaPO₄, 150 mM NaCl, 0.05% Tween-20, and 0.1% BSA. The peroxidase-conjugated secondary antibodies were diluted at 1:5000 in blot wash buffer (50 mM NaPO₄, 150 mM NaCl, 0.05% Tween-20) containing 5% non-fat dried milk. Band visualization was achieved using an enhanced chemiluminescence substrate (LumiGLO for Western blotting, Kirkegaard and Perry no. VC110) before exposure to X-ray film (Kodak 165–1579). The band densities were quantitated by laser densitometry (model PDS1-P90, Molecular Dynamics). To facilitate comparisons, the densitometry values were normalized to control, defining the mean for the control group as 100%.

Legend Supplemental Materials Figure 1. Loading gel confirming equal loading of all three time-points.

Immunoblot of inner medullary collecting duct suspensions, prepared at the early (Days 1 and 2) and later (Day 4) stages of vasopressin escape. Equal loading was confirmed by staining identically loaded gels with Coomassie blue and scanning these with a linear fluorescence scanner (Odyssey, Li-Cor Biosciences) at an excitation wavelength of 700 nm. Densitometry was performed with Odyssey software (v. 1.1), yielding an overall average variation of 2.2 ± 0.3 %.

Supplemental Materials Table 1.

Detailed description of the protein-protein interactions in the “vasopressin escape pathway” represented by edges in network diagram (Figure 4)

Central node	Interaction with	Nature of interaction	Reference(s)
c-Myc	p53	p53 decreases expression of c-Myc	A.o. Soengas, Science 1999
	Caspase 3	c-Myc decreases expression of of caspase 3 mRNA	-Louro, Canc Res 2002 -Grassili, JBC 2004
	Annexin A4	c-Myc decreases expression of Annexin A4 mRNA	Louro, Canc Res 2002
	Calnexin	c-Myc protein decreases expression of rat Calnexin mRNA	Louro, Canc Res 2002
	Elongation Factor 2	c-Myc protein increases expression of rat Elongation Factor 2 mRNA	Louro, Canc Res 2002
	Alpha B crystallin	c-Myc protein increases expression of rat Alpha B crystallin mRNA	Louro, Canc Res 2002
	F-actin capping protein β	In cells, human Myc protein decreases expression of rat F-actin capping protein β protein.	Shiio, EMBO J 2002
	Glucose phosphate isomerase	In Rat1a cells, c-Myc protein increases expression of rat Gpi mRNA.	Osthus, JBC 2000
	Lysosomal associated membrane protein 2	c-Myc protein decreases expression of rat Lamp2 mRNA	Watson, JBC 2002
	Transgelin 2	In cells, human c-Myc protein decreases expression of rat Transgelin2 protein.	Shiio Y et al, EMBO J 2002
	Retinoic acid receptor α	In mutant cells with a homozygous knockout of rat c-Myc, c-Myc protein decreases expression of rat retinoic acid receptorα mRNA.	Guo QM et al, Canc Res 2000
	Ubiquitin-like protein SMT3B	In mutant cells with a homozygous knockout of rat c-Myc, c-Myc protein increases expression of rat Smt3b mRNA.	Guo QM et al, Canc Res 2000
	Immunoglobulin kappa locus	c-Myc protein decreases transcription of immunoglobulin kappa locus gene	Oster SK et al, Adv Canc Res 2002
Glucocorticoid receptor (GR)	c-Jun	-GR protein decreases activation of a protein-protein complex consisting of c-Fos and of c-Jun -Heterodimerization of 39 kd c-Jun protein and GR protein occurs. -Binding of AP1 binding site and dimer consisting of c-Jun and GR protein occurs. -GR interacts with c-Jun.	-Touray M et al, Oncogene 1991 -Hsu TC et al, Biochem Biophys Res Commun 1993 -Ribeiro RC, Annu Rev Med 1995 -Chen H et al, J Biol Chem 1997 -Herdegen T et al, Brain Res Brain Res Rev 1998
	SRC-1	GR interacts with SRC-1	-Makishima M et al, Science 2002 -Kucera T et al, J Biol Chem 2002
	c-Myc	GR binds c-Myc promoter	Ma T et al, Mol Endocrinol 2000
	p53	-In cytoplasm, dexamethasone increases binding of human HDM2 protein and human GR protein and human p53 protein. -Binding of GR protein and P53 [TP53] protein occurs in HSC-2 cells. -Wild-type p53 interacts with GR.	-Yu C et al, Cancer Lett 1997 -Sengupta S et al, EMBO J 2000 -Sengupta S et al, Genes Dev 2001
	Calreticulin	-GR binds calreticulin. -GR binds N-domain of calreticulin.	-Burns K et al, Nature 1994 -Roderick HL et al, FEBS Lett 1997
	GRP78	GR binds GRP78.	Hutchison KA et al, J Steroid Biochem Mol Biol 1996
	c-Src	Binding of mouse Hsp90 protein(s) and mouse GR protein and mouse Src protein occurs in mouse thymocytes.	Marchetti MC, Blood 2003
c-Fos	c-Jun	c-Fos and c-Jun form a dimer representing the AP1 transcription factor.
	SRC-1	-Binding of c-Jun protein and SRC-1 protein occurs in a cell-free system. -Binding of c-Jun protein and human SRC-1 protein occurs in a system of purified components	Lee SK et al, Mol Endocrinol 2000 Lee SK et al, J Biol Chem 1998
	CREB	-In mouse CA1 neuron, transgenic CREB protein increases expression of mouse c-Fos protein. -CREB protein increases expression of c-Fos mRNA. -A protein-protein complex has members mouse CREB protein and mouse Crem alpha protein and mouse c-Fos protein and mouse c-Jun protein and mouse Junb protein and mouse Jund protein(s).	Barco A et al, Cell 2002
	Protein kinase A	PKA protein increases phosphorylation of c-Fos protein	Herdegen T et al, Brain Res Brain Res Rev 1998
	Annexin A4	In 208F cells, v-Fos protein increases expression of rat Annexin 4 mRNA.	Johnston IM et al, Oncogene 2000
	Keratin 8	In 208F cells, v-Fos protein increases expression of rat Cytokeratin-8 mRNA	Johnston IM et al, Oncogene 2000
	Sec23B	In 208F cells, v-Fos protein decreases expression of rat Sec23b mRNA.	Johnston IM et al, Oncogene 2000
	p53	p53 protein increases transcriptional activation of c-Fos gene.	Kohn KW et al, Mol Biol Cell 1999
	X-box binding protein 1	Binding of c-Fos protein and human XBP1 protein occurs in vitro	Ono SJ et al, Proc Natl Acad Sci U S A 1991
	Lysosomal associated membrane protein 2	In 208F cells, v-Fos protein decreases expression of rat Lamp2 mRNA.	Johnston IM et al, Oncogene 2000
	Ubiquitin-like protein SMT3B	In 208F cells, v-Fos protein increases expression of rat Smt3b mRNA.	Johnston IM et al, Oncogene 2000
c-Jun	CREB	-PD98059 in cell culture decreases binding of human CREB protein in a nuclear fraction of serum-deprived JEG3 cells treated with EGF protein and cAMP response element in a system of purified components and human c-Jun protein in a nuclear fraction of serum-deprived JEG3 cells treated with EGF protein. -EGF protein in cell culture increases binding of human CREB protein in a nuclear fraction of serum-deprived JEG3 cells and CRE in a system of purified components and human c-Jun protein in a nuclear fraction of serum-deprived JEG3 cells. -A protein-protein complex has members mouse Atf2 protein and mouse CREB protein and mouse c-Jun protein and mouse Junb protein. -A protein-protein complex has members mouse CREB protein and mouse Crem alpha protein and mouse Fos protein and mouse C-jun [Jun] protein and mouse Junb protein and mouse Jund protein(s).	Rutberg SE et al, Oncogene 1999 Roberson MS et al, Mol Cell Biol 2000 Andrecht S et al, J Biol Chem 2002
	p53	-c-Jun protein increases expression of mouse p53 protein. -Mouse p53 protein is necessary for the expression of mouse c-Jun mRNA that involves gamma radiation.	Fuchs SY et al, Genes Dev 1998
	Src	In Rat-1 cells, v-Src protein increases (S73) serine phosphorylation of rat c-Jun protein	Paasinen-Sohns A et al, Oncogene 1997
p53	HSP70	In Hct 116 cells, human p53 protein decreases expression of human HSP70 mRNA	Daoud SS et al, Cancer Res 2003
	Transgelin 2	In Hct 116 cells, human p53 protein is necessary for the Topotecan-independent expression of human transgelin 2 mRNA	Daoud SS et al, Cancer Res 2003
	Glucose phosphate isomerase	Mouse p53 protein is necessary for the expression of mouse glucose phosphate isomerase mRNA that involves gamma radiation.	Komarova EA et al, Oncogene1998
c-Src	Annexin A2	pp60src phosphorylates annexin A2	Ohnishi M et al, J Cell Sci 1994
	RACK1	Binding of human RACK1 protein and mouse c-Src protein occurs in rabbit reticulocyte lysates and NIH/3T3 cells.	Chang BY et al, Mol Cell Biol 1998
Protein kinase A (PKA)	CREB	-PKA protein increases (S133) phosphorylation of CREB protein. -BMP2 protein increases the mouse PKA protein-dependent phosphorylation of mouse Creb [Creb1] protein. -PKA activates CREB.	Rao A et al, Annu Rev Immunol 1997 Gupta IR et al, J Biol Chem 1999 Richards JP et al, J Biol Chem 1996
	AQP2	PKA phosphorylates AQP2.	Bouley R et al, J Clin Invest 2000
GRP78/BiP	Translation initiation factor 2	-GRP78/BiP protein decreases activation of PERK protein. -Thapsigargin decreases binding of human PERK protein and human GRP78/BiP protein. -Binding of human PERK protein and human GRP78/BiP protein occurs in 293t cells.	Bertolotti A et al, Nat Cell Biol 2000
	Immunoblogin	A protein-protein complex has members mouse GRP78/BiP	Chillarón J, Mol Biol Cell 2000
	kappa locus	protein and Immunoblogin kappa locus protein.
Activating TF 6	GRP78/BiP	-In a microsomal fraction of 293t cells, binding of human ATF6 protein and Grp78/BiP protein is the same as binding of mutant human ATF6 protein (T645I with its N-linked glycosylation site mutated) and Grp78/BiP protein. -Binding of human ATF6 protein and Grp78/BiP protein occurs in a microsomal fraction of 293t cells.	Yoshida H et al, Mol Cell Biol 2000 Hong M et al, J Biol Chem 2004
	X-Box binding protein 1	Soluble ATF6 protein increases expression of human XBP1 mRNA.	Yoshida H et al, Mol Cell Biol 2000
Calreticulin	Retinoic acid receptor α	Retinoic acid receptor α interacts with calreticulin	Desai D et al, J Biol Chem 1996