Dynamics of the G Protein-coupled Vasopressin V2 Receptor Signaling Network Revealed by Quantitative Phosphoproteomics

Abstract

G protein-coupled receptors (GPCRs) regulate diverse physiological processes, and many human diseases are due to defects in GPCR signaling. To identify the dynamic response of a signaling network downstream from a prototypical G_s-coupled GPCR, the vasopressin V2 receptor, we have carried out multireplicate, quantitative phosphoproteomics with iTRAQ labeling at four time points following vasopressin exposure at a physiological concentration in cells isolated from rat kidney. A total of 12,167 phosphopeptides were identified from 2,783 proteins, with 273 changing significantly in abundance with vasopressin. Two-dimensional clustering of phosphopeptide time courses and Gene Ontology terms revealed that ligand binding to the V2 receptor affects more than simply the canonical cyclic adenosine monophosphate-protein kinase A and arrestin pathways under physiological conditions. The regulated proteins included key components of actin cytoskeleton remodeling, cell-cell adhesion, mitogen-activated protein kinase signaling, Wnt/β-catenin signaling, and apoptosis pathways. These data suggest that vasopressin can regulate an array of cellular functions well beyond its classical role in regulating water and solute transport. These results greatly expand the current view of GPCR signaling in a physiological context and shed new light on potential roles for this signaling network in disorders such as polycystic kidney disease. Finally, we provide an online resource of physiologically regulated phosphorylation sites with dynamic quantitative data (http://helixweb.nih.gov/ESBL/Database/TiPD/index.html).

G protein-coupled receptors (GPCRs)¹ mediate physiological regulation in a multiplicity of organisms and in practically every mammalian tissue. The human genome contains ∼1,000 genes that code for GPCRs, reflecting the broad importance of this receptor superfamily in physiological regulation (1). These receptors contain seven transmembrane domains and signal through two classically defined pathways: heterotrimeric G protein activation and arrestin binding (2). Ligand binding to the receptor triggers downstream changes in phosphorylation of intermediate signaling proteins and regulatory targets. An important question is whether GPCR signaling only occurs through those two classically defined pathways or involves cross-talk with other pathways. One way to address this question is through large scale analysis of protein phosphorylation (i.e. phosphoproteomics). To do this in a physiological setting, we have undertaken a dynamic, quantitative phosphoproteomic analysis of type 2 vasopressin receptor (V2R; gene Avpr2) signaling in a native rat kidney collecting duct epithelium.

Vasopressin is a nine-amino acid peptide hormone that regulates water and solute transport in the mammalian kidney but also has important physiological effects in other tissues. Through its actions in the kidney, vasopressin allows an organism to maintain its serum osmolality within a very narrow range (290–294 mosmol/kg of H₂O in human) despite varying degrees of water intake. Dysfunctions in vasopressin signaling occur in a number of clinical disorders including syndrome of inappropriate anti-diuretic hormone hypersecretion (seen in many forms of cancer), congestive heart failure, hepatic cirrhosis, and nephrogenic diabetes insipidus (3). Vasopressin signaling is also recognized to be an important factor in the progression of autosomal dominant polycystic kidney disease, one of the most common life-threatening genetic diseases with prevalence estimated to be as high as 1 in 400 individuals (4). One of the main targets of vasopressin in the kidney is the renal collecting duct cell, which expresses the vasopressin receptor V2R. In these cells, vasopressin regulates the water channel aquaporin-2 (Aqp2) (reviewed in Ref. 5) and urea channel Slc14a2 (6) through the heterotrimeric G protein G_s and subsequent activation of adenylyl cyclases that mediate a rise in intracellular cAMP. Vasopressin also increases intracellular calcium through V2R and cAMP (7).

Despite a number of recent studies that have explored the steady state response of cells to vasopressin through phosphoproteomic methodologies (8–11), a comprehensive, dynamic profile of vasopressin signaling has yet to emerge. To carry out a dynamic phosphoproteomic analysis, we utilized a multiplexed labeling strategy (iTRAQ) allowing analysis of four distinct time points following the addition of vasopressin. Other studies have utilized various phosphoproteomic strategies to quantify biological responses (12–18). To our knowledge, the present study is the first temporal quantitative phosphoproteomic analysis of a physiological response using a native mammalian model. It also represents the largest collection of quantitative phosphorylation data on vasopressin signaling to date. These new data implicate a number of previously unidentified pathways that are regulated downstream of V2R, including the Wnt/β-catenin signaling and apoptosis pathways, which are relevant to the pathogenesis of polycystic kidney disease, as well as a variety of water balance disorders.

EXPERIMENTAL PROCEDURES

Brief descriptions of key experimental procedures are provided below. For complete details, see supplemental materials online.

Sample Preparation

IMCD suspensions were prepared from rat kidney inner medullas (150–250 μg of protein/inner medulla) using the method of Stokes et al. (19) with modifications (20). After isolation, IMCD suspensions were incubated for 0.5, 2, 5, and 15 min at 37 °C in the presence or absence of 1 nm [deamino-Cys¹,d-Arg⁸]vasopressin (dDAVP), a V2 receptor-specific analog of vasopressin, followed by centrifugation at >10,000 × g for 10 s. Pelleted IMCD tubules were lysed in 150 μl of lysis buffer containing 8 m urea, 50 mm Tris-HCl, 75 mm NaCl with 1× HALT^TM protease and phosphatase inhibitor (Pierce). (The four time points are labeled based on the time of incubation with dDAVP and do not include the 30 s of additional preparation time for centrifugation and lysis.) Protein samples were sonicated for 1 min with 0.5-s pulses on ice. The samples were spun at 10,000 × g for 10 min to pellet the debris, and the supernatant was saved for further analysis. Five microliters of each sample was saved for analysis of protein concentration by the BCA method (Pierce). Another 15 μl was removed for protein immunoblotting. The remainder of each sample (500 μg of protein) was reduced with 10 mm DTT for 1 h at 37 °C, alkylated with 40 mm iodoacetamide for 1 h at room temperature in the dark, and then quenched with 40 mm DTT for 15 min. The samples were diluted to <1 m urea with 50 mm ammonium bicarbonate buffer and then digested with trypsin overnight at 37 °C using an enzyme-to-protein ratio of 1:25 (w:w). After acidification with 0.5% formic acid, the samples were desalted on a 1-cc Oasis HLB cartridge (Waters, Milford, MA) prior to iTRAQ labeling. This entire sample preparation process was repeated two additional times on separate days to produce a total of three biological replicates.

iTRAQ Labeling

iTRAQ labeling was performed according to the manufacturer's protocol (Applied Biosystems, Foster City, CA). Briefly, the peptide samples were resuspended in 150 μl of iTRAQ dissolution buffer (0.5 m triethylammonium bicarbonate, pH 8.5). Each peptide sample was labeled with 5 units of iTRAQ 8-plex reagent for 2 h at room temperature according to the experimental design shown in Fig. 1. The reaction was quenched by adding 0.5% formic acid. All eight iTRAQ-labeled samples were combined into a single sample, then desalted (HLB cartridges; Waters), and dried in vacuo.

Fig. 1.

Overview of phosphoproteomic profiling. A, experimental workflow for iTRAQ-based quantitative phosphoproteomics of native rat IMCD. SCX, strong cation exchange chromatography; IMAC, immobilized metal affinity chromatography. B, the distribution of phosphopeptides based on the number of phosphorylation sites identified for each peptide. C, the histogram shows the background variability for all phosphopeptides across the three biological replicates (mean log₂(0.5 min control/5 min control) = −0.02 ± 0.22 (S.E.)). D, an overview of signaling pathways regulated by vasopressin. PANTHER Pathway terms were extracted for each phosphopeptide that changed significantly (p < 0.05) with vasopressin during at least one time point and compared with pathway terms for all identified peptides (i.e. background). The gene names for the corresponding phosphoproteins that changed with vasopressin are listed for each pathway. Only the pathways that contained more than one identified protein are included.

Fractionation and Phosphopeptide Enrichment

iTRAQ-labeled peptide samples were fractionated by strong cation exchange chromatography, and phosphopeptides were enriched by Ga⁺³ IMAC.

LC-MS/MS Analysis

The samples were analyzed on an LTQ Orbitrap Velos mass spectrometer (Thermo Scientific, San Jose, CA). The average precursor isolation window was 3 m/z. MS data files are available via the Proteomic Commons Tranche Repository (https://proteomecommons.org/tranche/). (See the supplemental data online.)

Database Searching

MS2 spectra were searched with Proteome Discoverer software (version 1.1.0.263; Thermo Scientific) running the Sequest algorithm on a concatenated database containing both the forward and reversed complement of the Rat Refseq Database (National Center for Biotechnology Information, March 3, 2010, 30,734 entries), which included a list of common contaminating proteins from other species. Precursor ion tolerance was 25 ppm, whereas fragment ion tolerance was 0.05 Da. Three missed trypsin cleavage sites were allowed. Static modifications included carbamidomethylation of cysteine (+57.021 Da) and iTRAQ 8-plex modification of lysine and peptide N termini (+304.205 Da). Variable modifications included oxidation of methionine (+15.995 Da); phosphorylation of serine, threonine, and tyrosine (+79.966 Da); and iTRAQ 8-plex modification of tyrosine (+304.205 Da). Known contaminant ions were excluded. The data sets were filtered to include <1% false positive hits (estimated based on target decoy analysis (21)) based on the following Xcorr threshold values for each charge state for the each replicate: +2 (2.14, 1.965, 2.355); +3 (2.38, 2.315, 2.81); +4 (2.66, 2.4, 3.06); and +5 (2.7, 2.405, 3.065). Phosphorylation site assignment was performed using a dynamic programming algorithm (false localization rate < 1%) that is currently under review. PubMed (http://www.ncbi.nlm.nih.gov/pubmed/), PhosphoSitePlus (http://www.phosphosite.org), and HPRD (http://www.hprd.org) (22) were used to search for known phosphorylation sites. Phosphopeptides matching to multiple protein isoforms were identified using ProMatch software (23).

iTRAQ Quantification

Reporter ion intensities for redundant peptides were summed for each iTRAQ channel. (We defined peptides as redundant if the charge states, site(s) of modification, and amino acid sequences were identical.) Abundance ratios (dDAVP/control) for all four time points were calculated, and each was normalized using a correction factor that was based on the ratio of the summed reporter ion intensities for the corresponding dDAVP and vehicle control channels. The log₂ of the normalized ratio was used as the basis for calculation of the mean and standard deviation for each peptide across all three biological replicates. Unpaired t tests determined whether changes in phosphopeptide abundance were significant. Background variability across the three replicate time courses was assessed for all phosphopeptides (mean log₂ (0.5 min control/5 min control) = −0.02 ± 0.22 (S.E.)) (Fig. 1C). In addition, average nonphosphorylated peptide abundance ratios for each protein were obtained by analyzing the IMAC flow-through fractions (included in supplemental Table S1). The majority of proteins did not change abundance with vasopressin (average log₂(dDAVP/control) = −0.025 ± 0.14 (S.E.)), indicating that changes in phosphopeptide abundance are not likely due to changes in total protein abundance. Average precursor ion isolation purity (i.e. the percentage of target ion intensity compared with the complete ion intensity in the precursor isolation window) for the three replicate data sets was 83%. Because precursor isolation purity is not always a reliable indicator of accurate iTRAQ quantitation (24), data sets were not filtered for a particular isolation purity threshold. The isolation purities for individual phosphopeptides are included in supplemental Table S1.

Cluster Analysis

Cluster analysis of phosphopeptides with similar temporal profiles was performed using a temporal pattern mining (TPM) algorithm (http://helixweb.nih.gov/TPM/) (25). To be included for cluster analysis, a phosphopeptide had to be present in at least two of the three time courses, and all eight iTRAQ reporter ions needed to be present in each spectrum to obtain quantifiable iTRAQ ratios at each of the four time points. This analysis produced 30 distinct temporal clusters. Phosphorylation motif analysis was then performed on each of these clusters by using the motif-x algorithm (26) (supplemental Table S5). Some of the clusters contained too few phosphopeptides to produce a motif. In house software (ABE; see “Bioinformatics”) was then used to extract Gene Ontology (GO) terms for each phosphopeptide in these clusters, as well as for all identified phosphopeptides (i.e. background). The proportion of each GO biological process term in each cluster was then calculated, and this value divided by the proportion of that particular GO Biological Process term in the background data (supplemental Table S6). The values for the 100 most abundant GO biological process terms for all clusters were then hierarchically clustered using Gene Cluster software (27) to detect patterns in GO term enrichment between the various temporal phosphopeptide clusters.

Bioinformatics

Automated Bioinformatics Extractor, ABE (http://helixweb.nih.gov/ESBL/ABE/), was used to extract GO terms and conserved protein domains through NCBI Entrez Programming Utilities (http://eutils.ncbi.nlm.nih.gov). The DAVID bioinformatic tool (Database for Annotation, Visualization and Integrated Discovery, NIAID, http://david.abcc.ncifcrf.gov/) (28) was used to extract the list of PANTHER pathway terms (http://www.pantherdb.org) (29) associated with each phosphopeptide.

RESULTS

Summary of Phosphoproteomic Profiling

Three replicate 8-plex iTRAQ time course experiments were performed as indicated in Fig. 1A to quantify the effects of vasopressin at four separate time points (0.5, 2, 5, and 15 min), each with its own time point control. Immunoblotting for Aqp2 phosphorylated at Ser-256 confirmed the response of the IMCD cells to the vasopressin analog dDAVP (supplemental Fig. S1). Aqp2 phosphorylated at Ser-256 was significantly increased during all four time points, whereas the total protein abundance of Aqp2 did not change with vasopressin.

A combined total of 12,167 phosphopeptides corresponding to 2,783 proteins were identified, although not all of the phosphopeptides were present in all three replicate time courses (Fig. 1B and supplemental Table S1; nonphosphorylated peptide data in supplemental Table S2). A database of all phosphopeptide identifications including dynamic, quantitative data for both phosphopeptide and corresponding protein level abundances is available online (http://helixweb.nih.gov/ESBL/Database/TiPD/index.html). Background variability across the three replicate time courses was assessed for all phosphopeptides (mean log₂ (0.5 min control/5 min control) = −0.02 ± 0.22 (S.E.)) (Fig. 1C). The majority of phosphopeptides (90% of the total) were singly phosphorylated, and serine residues were the most commonly modified amino acid (83% of the total). 4,202 nonredundant phosphorylation sites (50% of the total) were previously unidentified based on information from various online phosphorylation site databases (see “Experimental Procedures”). Approximately 28% of phosphorylation sites were present in known “regions of interest” (based on the current NCBI Reference Sequence record), which include binding sites, enzyme active sites, regions of local secondary structure, and conserved protein domains. (Full results are provided in supplemental Table S3.) Of the phosphopeptides that were present in all three replicate time courses, 273 changed significantly during at least one time point, and 40% of these changes occurred within 2 min of exposure to vasopressin. Analysis of PANTHER Pathway terms for these 273 phosphopeptides revealed that a number of signaling pathways are regulated by vasopressin (Fig. 1D). These included pathways previously implicated in vasopressin signaling (the phosphatidylinositol 3-kinase pathway, the protein kinase B/Akt pathway, various MAP kinase pathways (30), and signaling through Rho kinases (31, 32)), as well as pathways not known to be regulated downstream from the V2 receptor (angiogenesis, Wnt signaling pathway, and heterotrimeric G protein Gα_q- and Gα_o-mediated pathway). Many of the phosphoproteins that changed with vasopressin were shared among multiple signaling pathways, suggesting that there may be considerable interpathway connectivity. The full results of the PANTHER Pathway analysis are available in supplemental Table S4.

Forty-six phosphopeptides changed significantly (p < 0.05) in abundance during at least one time point by at least 40% (−0.5 ≤ log₂(dDAVP/control) ≥ 0.5) (value is based on 2× S.E. of control:control quantitation; see “Experimental Procedures”). Phosphopeptides that increased or decreased in the presence of dDAVP based on these criteria are presented in Table I. This table includes numerous membrane channels (Aqp2 and Slc14a2), trafficking proteins (Lrba, Sec22b, Agfg1, and Sept9), and protein kinases (Camkk2, Prkar1a, Ptk2, Map3k7, Map4k6, and Pak2). There were also a number of phosphoproteins involved in actin binding and cytoskeletal reorganization (Eps8l1, Lcp1, Ctnna1, and Kif13b), a result consistent with prior studies showing that vasopressin regulates cytoskeletal dynamics (33) and that depolymerization of the cortical actin network promotes Aqp2 trafficking (31, 34). Many of the proteins listed in Table I have not been implicated previously in vasopressin signaling.

Table I. Phosphopeptides that significantly changed in abundance in response to dDAVP.

In the peptide sequences, # indicates a phosphorylation site, and @ indicates methionine oxidation.

Protein name	RefSeq	Gene symbol	Peptide sequence	Phosphosite	Average log2(dDAVP/control)
					0.5 min	2 min	5 min	15 min
Average log2(dDAVP/control) ≥ 0.5
1-Phosphatidylinositol 4,5-bisphosphate phosphodiesterase β-3	NP_203501	Plcb3	NNS#ISEAK	Ser-1107	0.41	0.68b	0.65b	0.67
2-Oxoisovalerate dehydrogenase subunit α, mitochondrial	NP_036914	Bckdha	IGHHS#TSDDSSAYR	Ser-338a	−0.16	−0.26	−0.07	1.00b
6-Phosphofructo-2-kinase/fructose-2,6-biphosphatase 3	NP_476476	Pfkfb3	NSVTPLAS#PEPTK	Ser-496	−0.26	0.76b	−0.08	0.40
α1-Syntrophin	NP_001094371	Snta1	NSAGGTSVGWDS#PPASPLQR	Ser-183a	−0.09	0.56	−0.33	1.12b
Ankyrin repeat and SOCS box protein 4	NP_001019489	Asb4	S#LPLSLK	Ser-408	1.40b	1.96b	1.20	1.24b
Aquaporin-2	NP_037041	Aqp2	RQS#VELHS#PQSLPR	Ser-256, Ser-261	0.72	0.66b	0.83	−0.15
Aquaporin-2	NP_037041	Aqp2	RQS#VELHSPQSLPR	Ser-256	0.89	0.83b	1.09	1.13
Bcl-2-related ovarian killer protein	NP_059008	Bok	RSS#VFAAEIMDAFDR	Ser-8	1.99	1.21	2.92b	2.28b
Calcium/calmodulin-dependent protein kinase kinase 2	NP_112628	Camkk2	S#FGNPFEGSR	Ser-494	0.26	0.22b	0.46b	0.76b
cAMP-dependent protein kinase type I-α regulatory subunit	NP_037313	Prkar1a	TDS#REDEISPPPPNPVVK	Ser-77a	0.65b	−0.16	0.16	0.32
Cytokine receptor-like factor 2 precursor	NP_604460	Crlf2	GS#FPGLFEK	Ser-278	−0.1	0.41	0.26	0.59b
Focal adhesion kinase 1	NP_037213	Ptk2	LQPQEIS#PPPTANLDR	Ser-913	0.17	−0.01	0.51b	0.20
Hypothetical protein	XP_001079852	Fam83e	ETPTTTGPALS#DILR	Ser-363	1.19b	0.16	1.70	0.47
Keratin, type I cytoskeletal 19	NP_955792	Krt19	GGS#FSGALTVTDGLLGGNEK	Ser-67	0.08	−0.11	0.19b	0.60b
Kinesin-like protein KIF13B	NP_998791	Kif13b	SS#GLQPQGAPEAR	Ser-1719a	1.09b	0.99b	0.83b	1.07b
Leucine-rich repeat flightless-interacting protein 1	NP_001014291	Lrrfip1	RGS#GDTSISM@DTEASIR	Ser-88	0.64	1.22b	0.67	0.80b
Leucine-rich repeat flightless-interacting protein 2	NP_001019932	Lrrfip2	NSASATTPLSGNSS#R	Ser-129a	0.27b	0.70b	0.53b	0.67b
Lipopolysaccharide-responsive and beige-like anchor protein	NP_001102025	Lrba	QETAPDT#GDQQR	Thr-1231	0.42b	0.33	0.52b	0.51b
Mitogen-activated protein kinase kinase kinase 7	NP_001101390	Map3k7	S#IQDLTVTGTEPGQVSSR	Ser-439	0.62b	0.55b	0.64	0.60b
Nuclear receptor coactivator 3	XP_001072953	Ncoa3	ALS#LDSPVSVGSVPPVK	Ser-848	0.55b	—c	—	—
OCIA domain-containing protein 1	NP_001013896	Ociad1	S#VTYEELR	Ser-198	0.54	0.58	0.70	0.67b
Phosphatidylinositol glycan anchor biosynthesis, class B	NP_001101636	Pigb	S#TQYVIAQEK	Ser-29	1.00	1.23b	0.84	0.83
Rap1 GTPase-activating protein	XP_001070178	Rap1gap	SQS#M@DAM@GLSNK	Ser-472	1.19	0.78b	0.65b	1.08
Signal transducer and activator of transcription 5A	NP_058760	Stat5a	LS#PPAGLFTSAR	Ser-779	−0.56	—	0.38	0.56b
Similar to Map4k6-pending protein	XP_002724558	LOC303259	SNS#AWQIYLQR	Ser-714	0.82b	0.75b	0.83b	0.70b
Splicing factor, proline- and glutamine-rich	NP_001020442	Sfpq	GM@GPGT#PAGYGR	Thr-679	−0.37	0.34	0.07	0.87b
TBC1 domain family, member 1	XP_001071842	Tbc1d1	S#LTESLESILSR	Ser-497	0.56b	0.91b	0.85b	1.08b
Tyrosine-protein phosphatase nonreceptor type 12	NP_476456	Ptpn12	DADVSEES#PPPLPER	Ser-667	−0.20	1.35	−0.13	0.72b
Urea transporter 2 isoform 1	NP_062220	Slc14a2	S#VFHIEWSSIR	Ser-486	1.65b	2.00b	1.95b	1.98b
Urea transporter 2 isoform 2	NP_808877	Slc14a2	RRES#ELPR	Ser-84	1.61	1.81b	1.74b	1.85b
Vesicle-trafficking protein SEC22b	NP_001020857	Sec22b	NLGS#INTELQDVQR	Ser-137	3.02	0.68	2.90	1.75b
Average log2(dDAVP/control) ≤ −0.5
AHNAK nucleoprotein isoform 1	XP_001078032	Ahnak	GDVAASS#PSMK	Ser-4506	−0.23	−0.55	-0.58b	-0.78b
α-Endosulfine isoform 1	NP_001029146	Ensa	YFDS#GDYNMAK	Ser-67	−0.18	−0.51b	−0.32	−0.35
Alsin	NP_001013431	Als2	LSLPGLLSQVS#PR	Ser-486	−0.29	−0.55b	—	—
Aquaporin-2	NP_037041	Aqp2	QSVELHS#PQSLPR	Ser-261	−2.79b	−3.32b	−3.03b	−2.99b
Arf-GAP domain and FG repeats-containing protein 1	NP_001129068	Agfg1	GTPTQS#PVVGR	Ser-181	−0.24b	−0.53	−0.60	−0.56b
Catenin α-1	NP_001007146	Ctnna1	TPEELDDS#DFETEDFDVR	Ser-643	0.11	−0.19	−0.57b	−0.21
CLIP-associating protein 2	NP_446174	Clasp2	MVSQSQPGS#R	Ser-465	−0.79b	0.39	0.12	−0.14
Epidermal growth factor receptor kinase substrate 8-like protein 1	NP_001101937	Eps8l1	GASPAAET#PPLQR	Thr-187	−0.22	−0.17	−0.60	−0.72b
Erythrocyte protein band 4.1-like 4a	NP_001100867	Epb4.1l4a	VAQTQPAGSNS#INR	Ser-182	−0.08	−0.50b	0.07	−0.20
Hypothetical protein LOC500552	NP_001128101	RGD1561149	CSLLSAS#PASVR	Ser-326	−0.67b	−0.24	0.48	−0.46
Plastin-2	NP_001012044	Lcp1	GS#VSDEEMM@ELR	Ser-5	−0.64b	0.24	−0.32	1.20
Septin-9 isoform 2	NP_789826	Sept9	QVESTAST#PGPSR	Thr-107	−0.74	−0.90	−0.39b	−0.88b
Serine/threonine-protein kinase PAK 2	NP_445758	Pak2	IIS#IFSSTEK	Ser-55a	−0.29	−0.58b	−0.52	−0.10
Tumor protein p53 binding protein, 2 isoform 1	XP_001063503	Tp53bp2	ENLPVS#PDGNPPQQAVSAPSR	Ser-325	−1.06b	−0.47	−0.11	−0.35

^a Ambiguous phosphorylation site.

^b p < 0.05.

^c —, no quantifiable iTRAQ ratio.

Phosphopeptide Time Course Clustering

Phosphopeptides were further analyzed by cluster analysis using a TPM algorithm (http://helixweb.nih.gov/TPM/) (25) to detect groups of peptides that changed in abundance with similar temporal profiles. A total of 30 distinct phosphopeptide time course clusters were identified from 3,198 phosphopeptides (Fig. 2), indicating the wide variety of phosphorylation dynamics triggered by exposure to vasopressin. An interactive version of this figure that lists the phosphopeptide sequences and associated data contained within each cluster is available online at http://helixweb.nih.gov/ESBL/Database/TiPD/cluster.html. Clusters showing increases in phosphopeptide abundance with dDAVP at all time points consistently contained more phosphopeptides than clusters showing decreases, indicating that vasopressin may trigger an increase in net phosphorylation at the cellular level. It also suggests that kinase activation and/or phosphatase inactivation dominates the physiological response to vasopressin.

Fig. 2.

TPM-based clustering of phosphopeptides. Phosphopeptides that were quantified in at least two of the three time courses were clustered based on temporal patterns across four different time points (0.5, 2, 5, and 15 min). The color of each node (circles) in the tree indicates whether the cluster increased (red, positive slope) or decreased (green, negative slope) at that particular time point. The zero time point is represented by a theoretical node (yellow) that is used as a baseline reference. iTRAQ ratios are represented as the average log₂(dDAVP/control) for each time point. The numbers in parentheses indicate the four-digit binary label for each cluster based on the slope of the line between each time point (0 = negative slope; 1 = positive slope). n indicates the number of phosphopeptides in each cluster.

A cluster that increased at all time points is shown in Fig. 3A. Based on linear motif analysis of the residues surrounding the site of phosphorylation (motif-x (26)), this cluster was enriched for “basophilic” motifs consistent with activation of AGC family kinases (e.g. PKA) and/or Ca⁺²-calmodulin-dependent kinases (35) (Fig. 3A, inset). A phosphopeptide cluster that decreased at all time points was enriched in “proline-directed” phosphorylation motifs, suggesting a general inactivation of MAP and/or cyclin-dependent kinases (Fig. 3B). Similar up- and down-regulated kinase motifs have been demonstrated both in a mouse cortical collecting duct cell line (8) and in rat kidney thick ascending limb (11) in the presence of dDAVP. These general motifs are apparent in many of the peptide sequences presented in Table I. It is important to realize that these motifs are predictive only of general kinase classes and that other contextual information is often required to identify kinase-substrate relationships (36). The full results of TPM clustering and kinase motif analysis are provided in supplemental Table S5.

Fig. 3.

Kinase motif analysis of temporal clusters. A, an example of a phosphopeptide cluster that decreased in abundance with dDAVP at all time points and the most predominant phosphorylation motif for this phosphopeptide cluster (inset). B, an example of a phosphopeptide cluster that increased in abundance with dDAVP at all time points and the most predominant phosphorylation motif for this phosphopeptide cluster (inset). Phosphopeptides from the following proteins showed a significant (p value < 0.05) change in abundance: Ahnak (point a), Tpcn1 (point b), Slc14a2 (points c and d), Bok (point e), and Tbc1d1 (point f).

Clustering of Gene Ontology Terms

To determine which particular classes of phosphoproteins were regulated by vasopressin, we performed hierarchical clustering of GO biological process terms in each of the 30 temporal phosphopeptide clusters relative to all identified phosphopeptides as background. The heat map in Fig. 4 shows the relative enrichment of the 100 most abundant GO biological process terms, as well as the relationship between the various phosphopeptide clusters based on these GO terms. This type of analysis allows identification of enriched GO term clusters within each phosphopeptide cluster (vertical events) as well as enrichment of individual GO terms across related phosphopeptide clusters (horizontal events). This analysis uncovered distinct functional classes of proteins that are coordinately regulated by vasopressin.

Fig. 4.

Hierarchical clustering of GO biological process terms and phosphopeptide clusters. Shown is a heat map visualizing the proportion of the 100 most abundant GO biological process terms (displayed vertically) enriched in all 30 temporal phosphopeptide clusters (displayed horizontally) relative to the background proportion of each GO term for all identified phosphopeptides. GO terms that displayed similar enrichment patterns have been colored accordingly. Squares above the heat map indicate whether a phosphopeptide increased (red squares) or decreased (green squares) in abundance with dDAVP relative to the previously measured time point based on TPM clustering. The symbols directly above these squares indicate whether the temporal profile was located entirely above the x axis (+), was located entirely below the x axis (−), or crossed the x axis (0). Numerical designations for each temporal phosphopeptide cluster are located above these values.

GO cluster D in Fig. 4 included junction plakoglobin (Jup), catenin α-1 (Ctnna1), β-catenin (Ctnnb1), and vinculin (Vcl), all components of adherens junctions (37). These proteins shared similar phosphorylation profiles across multiple phosphopeptide clusters. Indeed, three of these proteins (Jup, Ctnna1, and Ctnnb1) have been shown to directly interact and are important regulators of cell-cell adhesion (38).

GO cluster F included glucosamine-fructose-6-phosphate aminotransferase [isomerizing] 1 (Gfpt1, also known as GFAT); phosphofructokinase, liver type (Pfkl); metallothionein 2A (Mt2a); and zinc transporter 1 (Slc30a1). Phosphorylation of all four of these proteins increased with dDAVP. Pfkl and Gfpt1 are metabolic enzymes that act as major regulatory points for the glycolytic and hexosamine pathways, respectively (39). Regulated phosphorylation of the Zn⁺² carrier Mt2a and Zn⁺² transporter Slc30a1 within the same cluster suggests that dDAVP may also affect the cellular distribution of Zn⁺², a cation that has been shown to inactivate Pfk1 (40). Increased phosphorylation of Gfpt1 at Ser-243 identified in our study has been shown to activate Gfpt1 (41), which leads to an elevation in intracellular levels of O-GlcNAc.

GO cluster E included Aqp2 and Slc14a2 (UT-A1), the major channels responsible for collecting duct water and urea transport, respectively, as well as a number of other proteins that may regulate channel trafficking. The heavy chain of nonmuscle myosin IIa (Myh9) was present in this cluster and has been implicated in Aqp2 trafficking via vasopressin-mediated activation of the Ca⁺²-calmodulin pathway (42). Also included in this cluster were three regulators of Rab GTPase activity: Als2, Rabgap1, and AS160 (Tbc1d4). The Rab family of proteins has been implicated in the regulation of Aqp2 localization and trafficking (43, 44).

We found that apoptosis was one of the processes most broadly regulated by vasopressin, with the GO biological process terms “apoptosis” (GO cluster B) or “anti-apoptosis” (GO cluster A) included in 23 of 30 phosphopeptide time course clusters. Phosphoproteins associated with apoptosis included Bok, Bad, Bcl2l14 (Bcl-G), Ptk2, Tp53bp2, Ctnna1, Map3k7 (Tak1), Stat5a, Krt8, and Krt18. Three of these proteins (Bok, Bad, and Bcl2l14) contain Bcl-2 homology 3 domains and are pro-apoptotic members of the Bcl-2 family. Bad phosphorylation increased with vasopressin at Ser-137 and Ser-156 in our study, as well as in mpkCCD cells (8). Phosphorylation of Bad at these sites by PKA has been shown to block dimerization with the anti-apoptotic protein Bcl-xL, thus promoting cell survival (45, 46). Similar to Bad, which is a pro-apoptotic ligand, Bok is a pro-apoptotic channel that contains additional Bcl-2 homology 1 and 2 domains that are important for channel formation and cytochrome c release from mitochondria (47). Bok phosphorylation significantly increased at Ser-8 (Table I) in response to dDAVP nearly 8-fold (log₂(dDAVP/control) = 2.92), the largest significant increase in phosphorylation of any protein identified in this study. Ser-8 is contained within a potential PKA consensus motif (…VLRRSS*VF…). To address whether this sequence can be phosphorylated by PKA, we performed an in vitro kinase assay on a synthetic Bok peptide containing Ser-8 and quantified the signal for both the unphosphorylated and phosphorylated peptides by LC-MS/MS. The addition of PKA resulted in a prominent peak for the phosphorylated peptide that was not present without kinase (Fig. 5A), as well as a 74% reduction in the MS1 peak intensity for the unphosphorylated peptide (Fig. 5B). This result demonstrates that Bok can be phosphorylated by PKA in vitro. Although very little is known regarding the regulation of Bok, this result raises the possibility that the pro-apoptotic function of Bok may be inhibited through phosphorylation by PKA in a fashion similar to Bad. We also found increased phosphorylation of Stat5a, an anti-apoptotic transcription factor responsible for transcription of a variety of genes including Bcl-xl and Bcl-2 (48). Phosphorylation of Stat5a at this site (Ser-779) promotes its transcriptional activity (49). Vasopressin regulated the phosphorylation of the anti-apoptotic kinase Fak (Ptk2), as well as the pro-apoptotic kinase Pak2 (Table I). We also found reduced phosphorylation of cytokeratins Krt8 and Krt18 in the presence of vasopressin. Hyperphosphorylation of these particular isoforms of cytokeratin has been associated with increased apoptosis (50). Taken together, the phosphorylation data from this large array of proteins involved in apoptosis suggest that vasopressin may inhibit this process and promote IMCD cell survival.

Fig. 5.

Vasopressin regulates a variety of apoptotic proteins in the collecting duct. A kinase assay was performed to determine whether residue Ser-8 of the pro-apoptotic factor Bok can be phosphorylated by PKA (A and B). A synthetic Bok peptide containing Ser-8 was incubated in the absence (no kinase) or presence (PKA) of the catalytic subunit of PKA. LC-MS/MS quantitation of the relative abundances of the phosphorylated peptide signal (A) and the unphosphorylated peptide signal (B). The asterisk indicates the site of phosphorylation. C, immunoblots of the cleaved forms of caspase-3 and caspase-7 in the presence (+) or absence (−) of dDAVP. D, band densities from replicate immunoblots (n = 5) are plotted as the mean log₂(dDAVP/control). The error bars indicate S.E. *, p < 0.05.

The anti-apoptotic effects of vasopressin mediated via the V1a receptor have been documented in neurons (51, 52) and glomerular mesangial cells (53). To further address whether vasopressin, via the V2 receptor, may have similar effects in kidney collecting duct, we incubated IMCD tubule suspensions in the presence and absence of dDAVP for 1 h and then quantified the levels of caspase-3 and caspase-7 cleavage products, which are hallmarks of apoptosis (54–57). The amounts of the cleaved forms of both caspases were significantly reduced in the presence of dDAVP (Fig. 5, C and D), consistent with the conclusion that vasopressin has a net anti-apoptotic effect in the renal collecting duct. Inhibition of apoptosis may provide a protective effect in the kidney inner medulla, which is often subjected to high osmotic stress during the urine concentration process (58). Inhibition of the apoptosis pathway has also been implicated in the pathogenesis of polycystic kidney disease (4), but a role for vasopressin in inhibition of apoptosis was not previously recognized.

Another pathway with many phosphoproteins regulated by vasopressin was the Wnt receptor signaling pathway (Figs. 1D and 4, GO cluster “C”). The Wnt signaling pathway plays a critical role in the development of the renal collecting duct system. Wnt family members are ligands for the Frizzled family of G protein-coupled receptors (59). Transcripts for both Wnt (Wnt4 and Wnt5a) and Frizzled (Fzd1 and Fzd4) gene family members are highly enriched in rat inner medullary collecting duct (http://dir.nhlbi.nih.gov/papers/lkem/imcdtr) (60). We identified a total of 15 Wnt pathway phosphoproteins based on GO term analysis, including β-catenin, an integral downstream component of this pathway that has been implicated in the pathogenesis of polycystic kidney disease (61, 62). Akt and PKA, which are both activated by vasopressin (30), phosphorylate β-catenin at Ser-552, which induces β-catenin accumulation in the nucleus and increases its ability to regulate gene transcription (63–65). Prior studies have demonstrated that phosphorylation of β-catenin at Ser-552 increases with vasopressin (9) (8, 11). We confirmed this increase in phosphorylation by LC-MS/MS and demonstrated through quantitative immunoblotting that this increase occurs within 0.5 min of vasopressin exposure (Fig. 6, A and B). A number of additional proteins in this GO cluster (Apc, Lrrfip1, Lrrfip2, Tnik, and Csnk1e) have been shown to regulate the function of β-catenin. Phosphorylation of Apc (adenomatosis polyposis coli) was decreased by vasopressin. Apc promotes the degradation of β-catenin, and its activity directly correlates with its phosphorylation state (66). Phosphorylation of Lrrfip1 at Ser-88 and Lrrfip2 at Ser-133, both part of putative PKA consensus sites, were increased by vasopressin. Lrrfip isoforms positively regulate β-catenin-dependent gene transcription (67–69). These data are consistent with the hypothesis that vasopressin may act as a potentiator of Wnt pathway signaling.

Fig. 6.

Confirmation of selected dDAVP-regulated phosphoproteins by immunoblotting. IMCD suspensions were treated with or without 1 nm dDAVP for the designated times and then processed for immunoblotting. Representative immunoblots probed with antibodies recognizing either β-catenin phosphorylated at Ser-552 (pS552-Ctnnb1) or total Ctnnb1 (A), either Rap1gap phosphorylated at Ser-556 (pS556-Rap1gap) or total Rap1gap (C), or Map3k7 phosphorylated at Ser-439 (pS439-Map3k7) or total Map3k7 (E). Band densities from three replicate immunoblots of pS552-Ctnnb1 (B), pS556-Rap1gap (D), and pS439-Map3k7 (F) were normalized to the corresponding total signal and then plotted as the mean log₂(dDAVP/control) for each time point. The error bars indicate S.E. *, p < 0.05.

Global Analysis of Protein Domains

We next extracted conserved protein domains for all vasopressin-regulated phosphoproteins. The most prevalent domain was the catalytic domain of serine/threonine protein kinases (smart00220: S_TKc) (12 hits) (supplemental Table S7). We identified a total of 17 serine/threonine protein kinases, including kinase regulatory subunits, that changed phosphorylation status with vasopressin. Among these were nine members of MAP kinase signaling pathways. Vasopressin has been shown to decrease signaling through MAP kinases (8, 30, 70), and the results of the current study confirm and expand on this hypothesis. We were able to detect significant changes in phosphorylation of A-raf and Raf1 (upstream kinases in the ERK pathway), which were consistent with down-regulation of the ERK pathway. In addition, Raf1 signaling can be negatively regulated by interaction with the small GTPase Rap1 (71). In the current study, vasopressin increased phosphorylation of Rap1gap (an inhibitor of Rap1) at Ser-472 (Table I) and Ser-556 (Fig. 6, C and D), two sites that are known to inhibit GAP activity when phosphorylated by PKA (72). Theoretically, this event could lead to an increase in Rap1 activity and subsequent inhibition of the Raf-MEK (MAP kinase/ERK kinase)-ERK pathway. This mechanism could contribute to the vasopressin-mediated decrease in ERK MAP kinase activity found in previous studies (8, 30).

We also found reduced phosphorylation of Jnk2 (Mapk9) and p38α (Mapk14), as well as the upstream activating kinases Map2k3/6. In contrast, phosphorylation of Map3k7 (Tak1) at Ser-439, a known target of PKA (73), was significantly increased by vasopressin as detected by LC-MS/MS, as well as immunoblotting (Fig. 6, E and F). Reduced phosphorylation of Map2k5 at Ser-311 (a known activating site (74)) along with increased phosphorylation of its corresponding regulatory kinase Map3k2 at Ser-331 (a known inhibitory site (75)) provides initial evidence that vasopressin can also down-regulate members of the ERK5 pathway. Thus, the majority of phosphorylation changes in MAP kinase family members detected in the present study are consistent with vasopressin triggering an overall reduction in MAP kinase signaling under normal physiologic conditions.

DISCUSSION

We have applied a temporal quantitative phosphoproteomic strategy to probe the dynamics of signaling through a prototypical GPCR, V2R, in native kidney cells. This study has greatly expanded knowledge about vasopressin signaling, producing a network model that serves as a basis for further study of GPCR signaling (Fig. 7). The study provides a temporal ordering of phosphorylation changes following vasopressin binding. These temporal data extend our knowledge about vasopressin signaling well beyond the existing evidence obtained in experiments in which measurements were made at a single time point (8–11). Although causality cannot be established using this approach, these time course data can provide the basis for hypotheses about causality. Clustering of the time courses revealed that the phosphorylation dynamics triggered by the activated V2 receptor are complex, with many different response patterns. The initial response was very rapid, with a large number of phosphorylation changes occurring within 30 s of exposure to ligand. These phosphorylation changes coincide temporally with increases in water and solute permeability in isolated, perfused renal collecting ducts following vasopressin exposure (76).

Fig. 7.

An updated view of the vasopressin signaling network. Select phosphoproteins identified in this study that increased (red nodes) or decreased (green nodes) with dDAVP at one or more time points (small squares) have been integrated with the established vasopressin signaling network (30) (gray nodes). Orange nodes indicate proteins that contain multiple dDAVP-regulated phosphorylation sites; some sites increased, whereas others decreased with dDAVP. A small letter a indicates a known activating phosphorylation site, whereas a small letter i indicates a known inhibitory phosphorylation site. The network was formatted using Ingenuity IPA software (www.ingenuity.com).

This study has demonstrated that binding of the V2R by vasopressin affects more than simply the canonical cAMP-PKA and β-arrestin pathways in the kidney collecting duct. Previous studies have demonstrated that arrestin binding to liganded GPCRs causes activation of MAP kinases (77–79), whereas our current results, as well as prior studies of the general response to vasopressin (8, 30, 70), show an inactivation of multiple MAP kinase pathways. Thus, it appears that the effect of vasopressin on MAP kinase signaling cannot be attributed to β-arrestin binding and MAP kinase scaffolding, but rather to additional effects, possibly related to PKA-mediated phosphorylation of Rap1gap (72). Additional evidence for the varying effects of GPCR signaling to MAP kinases comes from a recent phosphoproteomic study of lysophosphatidic acid (LPA) signaling (18). In that study, Schreiber et al. report that LPA (signaling through a GPCR) and EGF (signaling through the EGF receptor) synergistically activate the ERK MAP kinase pathway, which potentiates downstream mitogenic effects in a renal carcinoma cell line. This interplay between GPCRs and receptor tyrosine kinases has not been implicated in vasopressin signaling. In fact, perfusion of isolated collecting ducts with EGF actually inhibits vasopressin-induced osmotic water permeability (supplemental Fig. S2). These apparent differences in LPA and vasopressin signaling may be due to differences in the specific G protein isoforms activated by these receptors; AVP triggers G_s-mediated signaling, whereas many LPA receptors signal through G_i, G_q, and G_12/13 (18).

From this study we have demonstrated that vasopressin affects many other pathways including Wnt/β-catenin, Ca⁺²-calmodulin, and apoptosis pathways. An anti-apoptotic effect of vasopressin was confirmed through demonstration of a decrease in the proteolytic fragments of caspase-3 and -7 following vasopressin exposure in independent experiments. Our data also indicate that vasopressin regulates the phosphorylation of many proteins associated with structural components of the collecting duct cell including the actin cytoskeleton, microtubules, and various tight junction and adherens junction proteins. Although the signaling network that we have described shows a high degree of connectivity, the basis of the connectivity is not yet clear. It is important to note that many of the phosphorylation sites that were regulated by vasopressin in this study were not previously known to be regulated by any physiological stimulus. In addition, many of the phosphorylated proteins within a given pathway demonstrated coordinate regulation. An important question for future studies is whether the findings of this work can be generalized to other G_s-coupled receptors.

The resource produced by this study is an online database for temporal regulation of protein phosphorylation in a physiological setting. Along with the quantitative phosphoproteomic data, this study provides a large amount of relevant metadata including gene ontology-based classification, identification of conserved protein domains, identification of kinase phosphorylation motifs, and pathway analyses. These new data provide a systems view of GPCR signaling that can lead to a deeper understanding of disorders associated with perturbed vasopressin signaling, including the water retention seen in congestive heart failure, water wasting in nephrogenic diabetes insipidus, hyponatremia seen in many forms of cancer caused by the syndrome of inappropriate antidiuretic hormone hypersecretion, and autosomal dominant polycystic kidney disease.