|Home | About | Journals | Submit | Contact Us | Français|
GPR4 is a G protein-coupled receptor expressed in the vasculature, lung, kidney, and other tissues. In vitro ectopic overexpression studies implicated GPR4 in sensing extracellular pH changes leading to cyclic AMP (cAMP) production. To investigate its biological roles in vivo, we generated GPR4-deficient mice by homologous recombination. Whereas GPR4-null adult mice appeared phenotypically normal, neonates showed a higher frequency of perinatal mortality. The average litter size from GPR4−/− intercrosses was ~30% smaller than that from GPR4+/+ intercrosses on N3 and N5 C57BL/6 genetic backgrounds. A fraction of knockout embryos and neonates had spontaneous hemorrhages, dilated and tortuous subcutaneous blood vessels, and defective vascular smooth muscle cell coverage. Mesangial cells in kidney glomeruli were also significantly reduced in GPR4-null neonates. Some neonates exhibited respiratory distress with airway lining cell metaplasia. To examine whether GPR4 is functionally involved in vascular pH sensing, an ex vivo aortic ring assay was used under defined pH conditions. Compared to wild-type aortas, microvessel outgrowth from GPR4-null aortas was less inhibited by acidic extracellular pH. Treatment with an analog of cAMP, a downstream effector of GPR4, abolished microvessel outgrowth bypassing the GPR4-knockout phenotype. These results suggest that GPR4 deficiency leads to partially penetrant vascular abnormalities during development and that this receptor functions in blood vessel pH sensing.
Blood vessels in mammalian embryos develop through vasculogenesis and angiogenesis. During the vasculogenesis phase, endothelial progenitor cells cluster to form a vascular plexus, which then undergoes angiogenesis by sprouting and remodeling to form a complex vascular network. Subsequently, recruitment of smooth muscle cells and pericytes to nascent blood vessels is important for the stability and maturation of the vasculature (6). At the molecular level, a myriad of studies demonstrate that cell surface receptors and their ligands, such as vascular endothelial growth factor (VEGF) receptors and VEGF, are critical for vascular formation and function (9). Deficiency in these regulators may lead to various degrees of blood vessel defects. For instance, deletion of VEGF receptors or Tie receptor tyrosine kinases causes abnormal vasculogenesis and angiogenesis at different stages of mouse development (16, 39, 45); lack of platelet-derived growth factor receptor β (PDGFRβ) or PDGF-B results in defective vascular maturation with reduced smooth muscle cell coverage in a subset of blood vessels and deficient mesangial cell recruitment to kidney glomeruli (18, 32, 33, 47). Although protein growth factors and their receptors have been a major focus in angiogenesis studies, many different factors play a positive or negative role in maintaining a delicate balance in vascular formation (7, 38, 52).
Tissue pH is one of the factors that can regulate blood vessel formation and function (4, 10). Under various physiological and pathological conditions including exercise, growth of solid tumors, inflammation, ischemia, diabetic ketoacidosis, and renal and respiratory failure, blood vessels form and/or function in a local environment with an acidic pH. Notably, glycolysis and the production of lactic acid by tumor cells in a hypoxic microenvironment lead to a significant reduction of extracellular pH in tumors (17). Studies show that acidic extracellular pH has pleiotropic effects on angiogenesis, including the inhibition of endothelial cell migration, tube formation, and in vitro vascular growth (4, 10). Most previous research has focused on the effects of acidic pH on angiogenic factors. Acidosis has been shown to modulate the expression and activity of both angiogenic and angiostatic factors, which in turn regulate the angiogenesis process (4, 14, 46, 55). Therefore, further understanding of vascular pH responses may help in designing methods to control blood vessel formation and function in various pathologies such as cancer and ischemia. However, little is known about how vascular cells per se directly sense acidic extracellular pH.
GPR4 and several related G protein-coupled receptors (GPCRs) have recently been identified as novel proton-sensing receptors (20, 34, 37, 44, 56). Studies have shown that GPR4 is expressed in vascular endothelial and smooth muscle cells, lung, kidney, heart, liver, and other tissues (1, 23, 35, 36, 51). The in vivo function of GPR4 is not well understood, although recent studies indicate that GPR4-related genes in Xenopus laevis are involved in the regulation of embryonic gastrulation (8, 49). Previous in vitro analyses suggest that GPR4 might mediate the sphingosylphosphorylcholine-induced endothelial tube formation and lysophosphatidylcholine-induced impairment of endothelial barrier function (23, 41), but the publication proposing the receptor-ligand relationship between GPR4 and sphingosylphosphorylcholine and lysophosphatidylcholine has been withdrawn (62).
The expression of GPR4 in vascular cells and its biochemical function of sensing pH changes have led us to hypothesize that GPR4 is one sensor that blood vessels may use to respond to extracellular acidic pH. Previous cell line overexpression studies demonstrate that acidic pH activates GPR4 to induce the production of cyclic AMP (cAMP) (34, 44). In this respect, cAMP and its downstream kinase protein kinase A have been shown to inhibit angiogenesis (11, 15, 24, 25, 53).
The objectives of the present study were twofold: (i) to investigate the in vivo biological role of GPR4 and (ii) to identify whether GPR4 is a functional sensor for blood vessels in response to pH changes. To assess these, we generated GPR4-knockout mice by homologous recombination. Spontaneous hemorrhaging was observed in approximately 17% of GPR4-null embryos and neonates. These mice showed dilated small blood vessels with a decrease of smooth muscle cell coverage and significantly reduced mesangial cells in kidney glomeruli. Direct tissue explants from knockout mice revealed that GPR4 is responsive to acidic extracellular pH to regulate microvessel outgrowth.
A clone harboring the GPR4 gene was obtained by screening the 129 mouse genomic library (Incyte Genomics). The bacterial artificial chromosome clone was then digested with EcoRI, and an 8-kb DNA fragment containing the GPR4 gene was subcloned, from which a 1.5-kb XhoI-XmaI DNA fragment containing the GPR4 coding region was replaced with a 3.4-kb internal ribosome entry site (IRES)-green fluorescent protein (GFP) and phosphoglycerokinase (PGK)-neo cassette. The targeting vector with a 3.3-kb right arm and a 3.2-kb left arm was linearized by NotI digestion and electroporated into 129SvJ embryonic stem (ES) cells (Genomes System). Neomycin-resistant ES clones were screened for homologous recombination by Southern blotting. Genomic DNA from ES cells was digested with XbaI and hybridized with a right external probe which detected the wild-type and targeted alleles as 7.7- and 9.7-kb bands, respectively (see Fig. Fig.1B).1B). ES cells from positive clones were injected into C57BL/6 blastocysts (UCLA transgenic facility) to generate chimeric male mice, which were crossed with C57BL/6 females to produce agouti GPR4 heterozygous progeny. The genotype of the pups (about 2 weeks old) was detected by Southern blotting and PCR of tail genomic DNA. The genotype of the embryos was determined by PCR of yolk sac or tail genomic DNA. The three PCR primer sets were primer 1 (5′-TCCCGCCCCCCGTGGCCCTG-3′), primer 2 (5′-AAGTTCATCAGGTAGACGCC-3′), and primer 3 (5′-TGCCTGCAAAGGGTCGCTAC-3′), which amplified the wild-type allele as a 250-bp band and the knockout allele as a 427-bp band (Fig. (Fig.1C).1C). The PCR conditions were 35 cycles of 94°C for 15 s, 58°C for 30 s, and 72°C for 1 min.
All mice were housed in the Thoren caging system (Hazelton, PA) in an air-conditioned room with a 12-h light/dark cycle at the UCLA animal facility, following current institutional regulations. The studies reported here used GPR4+/+, GPR4+/−, and GPR4−/− mice backcrossed for three or five generations (N3 or N5) to the C57BL/6 genetic background. Mendelian ratio calculations were done with N5 heterozygote intercrosses. Litter sizes and perinatal mortality rates of GPR4+/+ and GPR4−/− intercrosses were determined on both N3 and N5 genetic backgrounds.
Mouse tissues and embryos were fixed in 10% buffered formalin overnight at 4°C and embedded in paraffin. Serial tissue sections were stained with hematoxylin and eosin. For immunohistochemistry, antigens were retrieved by boiling tissue sections in acidic citrate buffer. Endogenous peroxidase was blocked with 0.3% H2O2 in methanol. Tissue sections were incubated with the primary rabbit polyclonal anti-α-smooth muscle actin (αSMA) antibody (Abcam Inc., Cambridge, MA) and then with the secondary horseradish peroxidase-labeled polymer-conjugated goat anti-rabbit immunoglobulin G antibody (DakoCytomation). Signals were detected through the reaction with diaminobenzidine chromogen (DakoCytomation) as brown staining, and the slides were counterstained with hematoxylin.
Whole-mount staining was performed as previously described (29). Briefly, neonatal skin or vibratome-sectioned tissues were fixed with 4% paraformaldehyde in phosphate-buffered saline overnight, blocked with 5% donkey serum, and 0.3% Triton in phosphate-buffered saline for 2 h and incubated with the primary rat anti-mouse CD31 antibody (BD Pharmingen) followed by the secondary Cy3-conjugated donkey anti-rat immunoglobulin G antibody (Jackson Immunoresearch). For CD31 and α-smooth muscle actin double staining, tissues were further stained with the fluorescein isothiocyanate-conjugated mouse monoclonal anti-α-smooth muscle actin antibody (Sigma). Fluorescent signals were visualized with a confocal microscope (model MRC-1024; Bio-Rad Laboratories).
All primary cells were cultured in a tissue culture incubator with 5% CO2 at 37°C. Human umbilical vein endothelial cells (HUVEC) were purchased from Cambrex (Walkersville, MD) and grown in the endothelial cell culture medium EGM-2, which contains EBM-2 basal medium and supplements of 2% fetal bovine serum, hydrocortisone, human fibroblast growth factor B, VEGF, R3-insulin-like growth factor 1, human EGF, ascorbic acid, heparin, gentamicin, and amphotericin B (Cambrex). To overexpress murine GPR4, HUVEC were transduced with MSCV-GPR4-GFP retrovirus, and GFP-positive cells were purified by fluorescence-activated cell sorting.
For all pH experiments, culture media were buffered as previously described (44) with 7.5 mM HEPES, 7.5 mM EPPS [N-(2-hydroxyethyl)-piperazine-N′-3-propanesulfonic acid], and 7.5 mM 2-(4-morpholino)ethanesulfonic acid (HEM). The production of cAMP in HUVEC under different pHs was measured using the cAMP enzyme immunoassay kit (Amersham Pharmacia Biotech, Piscataway, NJ) as previously described (44).
The HUVEC wound healing assay was performed by growing the cells to confluence in EGM-2 medium and generating a wound with a pipette tip. The process of wound closure was recorded using a Nikon invert microscope with a 4× objective lens.
Mice were heparinized by being injected with 500 μl of heparin (1,000 U/ml) intraperitoneally to prevent blood clots. Aortas isolated from mice were rinsed with EGM-2 medium, cleaned of periadventitial fat and connective tissues, and cut into ~1- to 1.5-mm-long rings. The aortic ring assay was performed as previously described (48) with a few modifications. One hundred microliters of Matrigel (BD Biosciences, Bedford, MA) was added to each well of a 48-well tissue culture plate (Costar) and solidified for 20 min at 37°C. Aortic rings were placed on the Matrigel layer, and another 100 μl of Matrigel was added to embed the rings. After the Matrigel was solidified, 1 ml of HEM-buffered EGM-2 medium was added to promote microvessel outgrowth at different pHs in a tissue culture incubator with 3% CO2 at 37°C. EGM-2 medium was changed once a day to minimize pH changes during culture. The cAMP analog 8-bromo-cAMP was purchased from Calbiochem and used at 500 μM in the assay. Microvessel outgrowth from aortas on day 4 was photographed with a digital camera linked to a Nikon invert microscope with a 4× objective lens. After images were acquired, the outgrowth area and the aortic ring area were delineated and quantified with the ImageJ software (http://rsb.info.nih.gov/ij/). The area of microvessel outgrowth was determined by subtracting the ring area from the microvessel area plus the ring area (2).
Wound clot formation was used as a functional assay to examine blood coagulation in GPR4+/+ and GPR4−/− mice as previously described (13). Young pups or young adult mice were anesthetized with ketamine-xylazine, and approximately 6 to 8 mm of distal tail was amputated. Clot formation was observed, and bleeding times were recorded.
RNA was isolated from cells and tissues using the TRIzol reagent (Invitrogen) and treated with RNase-free DNase I (Ambion, Austin, TX) to remove genomic DNA contamination. cDNA was synthesized using the oligo(dT) primer and the reverse transcriptase Superscript II (Invitrogen), and reverse transcription-PCR (RT-PCR) was performed as previously described (60). PCR primers for murine GPR4 were 5′-CAAGACCCACTTGGACCACA-3′ and 5′-TGTCCTGGGCCTCCTTTCTA-3′; PCR primers for GFP were 5′-GTGACCACCCTGACCTACGG-3′ and 5′-CGGACTGGGTGCTCAGGTAG-3′; PCR primers for β-actin (both human and murine) and human GPR4, G2A, TDAG8, and OGR1 were previously described (60, 61). The PCR conditions were 35 cycles of 94°C for 30 s, 58°C for 30 s, and 72°C for 1 min.
Data were analyzed by the GraphPad Prism 4 software (GraphPad Software, Inc.). P values were calculated using the unpaired t test, and the values of <0.05 were considered statistically significant. Deviation from the Mendelian ratio was calculated by the χ2 test in Microsoft Excel.
To disrupt the GPR4 gene in 129 mouse ES cells, its open reading frame was replaced with the IRES-GFP and PGK-neo cassette by homologous recombination (Fig. (Fig.1A).1A). ES clones resistant to G418 were screened by Southern blotting. Genomic DNA with the GPR4 gene recombined allele showed a 9.7-kb XbaI band (Fig. (Fig.1B).1B). Targeted ES clones were injected into C57BL/6 mouse blastocysts to generate chimeric mice, and the targeted allele was transmitted into the germ line. GPR4-null and heterozygous progeny were produced and confirmed by Southern blotting and PCR analysis (Fig. 1B and C). The absence of GPR4 RNA transcripts in knockout mice was verified by RT-PCR in the lung, kidney, and heart. Consistently, the presence of GFP transcripts was detected in GPR4-knockout mice but not in wild-type mice (Fig. (Fig.1D1D).
GPR4-null mice were obtained from the heterozygote intercrosses. Adult GPR4-null mice were viable and fertile, without gross phenotypic difference from their wild-type littermates on both N3 and N5 backcrossed C57BL/6 backgrounds. To systematically evaluate the mice, histology with hematoxylin and eosin staining was performed to examine tissues from GPR4+/+ and GPR4−/− male and female littermates at 8 weeks of age. No significant histological difference was detected in major organs including the brain, heart, lung, thyroid, trachea, thymus, spleen, liver, kidney, skin, testis, uterus, bladder, stomach, and intestine (data not shown). Several GPR4-null mice up to 14 months of age did not show a significant difference in longevity from wild-type controls. Routine hematological tests failed to identify abnormalities in the numbers of red or white blood cells and platelets and the level of hemoglobin between adult GPR4−/− and GPR4+/+ mice (data not shown).
We examined the effects of GPR4 deficiency during mouse fetal development. Intercrosses of heterozygous mice on an N5 backcrossed C57BL/6 genetic background generated GPR4-null mice at a 16.8% frequency, less than the expected 25% Mendelian ratio (Table (Table1),1), suggesting that some GPR4-null pups died before genotyping (2 weeks after birth). The major fatal events likely occurred after birth, as the ratio of GPR4-null embryos by 18.5 days postcoitum (dpc) was not significantly reduced. Further examination of GPR4+/+ × GPR4+/+ and GPR4−/− × GPR4−/− mouse intercrosses on an N3 backcrossed genetic background revealed that average litter sizes on the first day after birth were 8.1 ± 2.8 and 5.2 ± 2.1 (mean ± standard deviation, P < 0.001), respectively (Fig. (Fig.2A).2A). Similarly, the average litter size from GPR4−/− × GPR4−/− mice on an N5 background was also smaller than that from GPR4+/+ × GPR4+/+ mice (5.9 ± 2.0 versus 8.3 ± 1.7, respectively; P < 0.001) (Fig. (Fig.2B).2B). The decrease of litter size is unlikely to be due to reproductive defects of the parents because the litter size from the intercross of GPR4−/− females with GPR4+/+ males or of GPR4+/+ females with GPR4−/− males was not significantly different from the litter size from the GPR4+/+ intercross (Fig. (Fig.2A).2A). We observed that, compared to wild-type neonates from GPR4+/+ intercrosses, GPR4-null neonates from GPR4−/− intercrosses exhibited a higher perinatal mortality rate. These animals usually died within a week after birth (most in the first 2 days). As shown in Table Table2,2, the mortality rates of newborns from GPR4+/+ × GPR4+/+ and GPR4−/− × GPR4−/− were 15.3% and 38.1%, respectively.
Embryos at different gestational stages (10.5 to 18.5 dpc) were examined for potential defects. Various degrees of hemorrhaging were observed in a fraction of GPR4-null embryos. Macroscopic superficial hemorrhages were detected in 11 of 66 GPR4-null embryos, 4 of 122 heterozygous embryos, and 0 of 62 wild-type embryos (Table (Table11 and Fig. Fig.3).3). Twenty-two resorbed embryos were also observed, but the genotypes could not be determined. Hemorrhagic sites were visible in the head, limbs, or other regions along the body wall (Fig. 3B, D, and F). Petechial hemorrhages were also identified in line with the track of blood vessels under the skin (Fig. (Fig.3F),3F), suggesting abnormalities of small blood vessels. Some GPR4-null neonates died perinatally and were frequently cannibalized by the parents. Postmortem examination of several dead or dying pups identified blood leakage in the peritoneal cavity; hemorrhages in the lung, kidney, or skin; or anomalies of mesenteric blood vessels (Fig. 3H, J, L, N, and P). Histological analyses of GPR4-null embryos revealed the subcutaneous and mesenchymal accumulation of fetal blood cells at the hemorrhagic sites (Fig. (Fig.4B).4B). In some GPR4-null neonates, hemorrhages were detected in the parenchyma of the lung and the glomeruli and parenchyma of the kidney (Fig. 4D and F).
To assess potential defects in blood coagulation, another important cause of bleeding, a functional test of tail bleeding was performed. The bleeding times in GPR4+/+ and GPR4−/− young pups at 2 weeks of age were 26.2 min (n = 5) and 24.6 min (n = 5, P = 0.7), respectively (see Fig. S1A in the supplemental material). The bleeding times of GPR4+/+ and GPR4−/− young adult mice were also very similar (data not shown). The platelet numbers in peripheral blood were normal in GPR4+/+ and GPR4−/− young adult mice (see Fig. S1B in the supplemental material). These combined results suggest that the hemorrhagic diathesis is likely due to defects in blood vessels but not blood coagulation.
In addition to hemorrhaging, some GPR4-deficient neonates had difficulty in breathing, became cyanotic, and died shortly after delivery from GPR4−/− intercrosses (Fig. (Fig.5A),5A), indicating severe respiratory distress. As shown by hematoxylin and eosin staining of neonatal lung paraffin sections, wild-type bronchioles had a single layer of epithelial cells with normal morphology lining the airway lumen (Fig. 5B and C). In contrast, the bronchiolar epithelium from GPR4-null neonates with respiratory distress consisted of abnormal cells with enlarged clear cytoplasms. In some regions, these cells expanded to form multiple layers (Fig. 5D and E), representing airway lining metaplasia that extended peripherally to respiratory and terminal bronchioles. The severity of the metaplasia was variable among GPR4-null neonates and was milder in the neonates without obvious respiratory distress (Fig. 5F and G). The lung epithelial metaplasia appeared to be transient and reversible because the clear-cytoplasm cells were not detected in surviving GPR4−/− adult mice (Fig. 5H and I). Furthermore, compared to the wild type (Fig. (Fig.4C),4C), the sacculi of the lungs of several GPR4-null mice were poorly expanded (Fig. (Fig.4D).4D). These lung anomalies could contribute to respiratory difficulties that manifest as labored breathing in the GPR4-deficient neonates and might have a role in perinatal mortality.
To further define the cause of the observed macroscopic hemorrhages, embryos and neonates were stained for CD31 (platelet endothelial cell adhesion molecule) and αSMA to delineate vascular endothelial and smooth muscle cells, respectively. In contrast to the well-organized vasculature in the wild-type neonatal skin (Fig. (Fig.6A),6A), dilated, tortuous, and poorly organized blood vessels were observed in the hemorrhagic GPR4-null mouse skin (Fig. (Fig.6B).6B). Double staining for CD31 and αSMA revealed that these dilated small arteries and veins were poorly enveloped by smooth muscle cells (Fig. (Fig.6D),6D), which could lead to compromised vascular stability and hemorrhaging. Compared to wild-type littermates (Fig. (Fig.6E),6E), αSMA staining of GPR4-null embryos with petechial hemorrhages (Fig. (Fig.3F)3F) also showed that some small blood vessels under the dermis had poor smooth muscle cell coverage (Fig. (Fig.6F).6F). Disrupted vessels with blood cell leakage could be seen in GPR4-null embryos (Fig. (Fig.6H).6H). However, smooth muscle cell coverage of larger blood vessels was not significantly different in GPR4−/− and GPR4+/+ embryos (Fig. 6I and J), suggesting that the reduction of smooth muscle cell coating is selective in different vascular walls.
Mesangial cells are a type of smooth muscle-like cell supporting the complex of capillary loops and epithelial podocytes in kidney glomeruli. We stained kidney sections of GPR4+/+ and GPR4−/− neonates for αSMA, a marker to identify mesangial cells, smooth muscle cells, and interstitial myofibroblasts in neonatal kidneys (5, 12, 33). In GPR4+/+ neonatal kidneys, αSMA antibodies clearly labeled mesangial cells in the core of mature glomeruli, smooth muscle cells around small arteries and afferent/efferent arterioles, and peritubular interstitial cells (Fig. 7A and C). In GPR4−/− neonatal kidneys, αSMA staining of smooth muscle cells covering small arteries and arterioles was detected at a similar level. However, mesangial cells were significantly decreased in the majority of GPR4-null glomeruli (Fig. 7B and D). Weak staining was identified in some mutant glomeruli (Fig. (Fig.7B,7B, right arrow). αSMA-positive peritubular interstitial cells also appeared to be reduced (Fig. 7B and D). To further examine the interaction between endothelial cells and mesangial cells/smooth muscle cells, neonatal kidneys were double stained for CD31 and αSMA. Glomerular capillary loops and peritubular capillaries could be detected by CD31 staining in both GPR4+/+ and GPR4−/− kidneys (Fig. 7E and F). However, whereas close contact between capillary loops and mesangial cells was evident in wild-type glomeruli (Fig. (Fig.7E),7E), GPR4-null glomerular capillaries were defective in mesangial cell connection (Fig. (Fig.7F).7F). No obvious difference in αSMA staining of small arteries was observed between GPR4+/+ and GPR4−/− kidneys (Fig. 7E and F).
Taken together, these results suggest that GPR4 plays a functional role in blood vessel maturation during development. In the absence of GPR4, smooth muscle coverage of small blood vessels and mesangial cell coverage of glomeruli are defective, which correlates with blood vessel dilation and hemorrhaging observed in some GPR4-null embryos and neonates as a partially penetrant developmental phenotype.
Ectopic overexpression studies have shown biochemically that GPR4 can be activated by acidic extracellular pH to produce cAMP in 293T cells (34, 44). Our mouse phenotype analyses have not established the direct molecular connection between the proton-sensing function of GPR4 and the in vivo hemorrhage and vascular instability. To assess whether GPR4 is functional in the pH sensing of vascular cells, we took advantage of in vitro assay systems in which we could evaluate biological responses under precisely controlled pH conditions. HUVEC were transduced with a retroviral vector overexpressing the murine GPR4 gene with a GFP marker attached at its C terminus (44). Microscopic examination identified GPR4-GFP expression in the HUVEC, and the DiI-Ac-low-density lipoprotein uptake assay demonstrated that the cells are functional endothelial cells (see Fig. S2A in the supplemental material). We also detected endogenous GPR4 RNA transcripts in HUVEC, while the expression of related GPCRs including G2A, TDAG8, and OGR1 was barely detectable in these cells by RT-PCR (see Fig. S2B in the supplemental material). In the pH-induced cAMP assay, pH 5.8 treatment resulted in a 1.6-fold increase of cAMP level over that at pH 7.8 in HUVEC. Retroviral overexpression of GPR4 in these cells significantly augmented the cAMP production to approximately 10-fold at low pH (see Fig. S2C in the supplemental material), demonstrating that GPR4 can be activated by acidic pH in endothelial cells. To examine the functional response, the HUVEC wound healing assay showed that low pH delayed the wound closure, and overexpression of GPR4 significantly enhanced the inhibitory effects of acidic pH (see Fig. S2D in the supplemental material), suggesting that the acidic pH/GPR4 signaling negatively regulates endothelial cell migration.
To further evaluate the pH-sensing function of GPR4, in addition to the two-dimensional cell culture system, we used a three-dimensional aortic ring assay to more closely approximate angiogenesis under defined pHs which are difficult to precisely control in animals. In the aortic ring culture with Matrigel and EGM-2 endothelial growth medium containing fetal bovine serum, VEGF, fibroblast growth factor, insulin-like growth factor, and other factors (48), similar amounts of microvessels were detected from both the wild-type and GPR4-null aortic rings cultured at pH 7.5 (Fig. (Fig.8).8). Consistent with previous studies (4), angiogenesis was significantly reduced at acidic pH; however, approximately fourfold-more microvessel outgrowth was observed from GPR4-null aortic rings than from wild-type rings at pH 6.9. When the aortic ring cultures were treated with the cAMP analog 8-bromo-cAMP, angiogenesis was significantly inhibited from both GPR4+/+ and GPR4−/− aortas (Fig. (Fig.8),8), indicating that the elevation of cAMP can bypass the effects of the GPR4 genotype. These data suggest that acidic pH activates GPR4 to negatively regulate microvessel growth, consistent with cAMP production. This agrees with previous results showing that cAMP can attenuate angiogenesis in other systems (15, 25, 53).
Collectively, the results from the aortic ring assay and wound healing assay indicate that GPR4 can serve as a sensor for blood vessels to detect pH changes, and this pH-sensing function may be related to vascular growth and other processes. Compared to global regulators such as R-Ras that suppress angiogenesis (26), GPR4 negatively modulates blood vessel growth under acidic pH conditions.
Blood vessels consist of two major types of cells: endothelial cells and mural cells. Endothelial cells form vascular channels, and mural cells including smooth muscle cells and pericytes contribute to the stability and maturation of blood vessels (6). The molecular interaction between endothelial cells and mural cells is essential for vascular development and homeostasis. Signals from endothelial cells are important for the induction, migration, and proliferation of mural cells from mesenchymal progenitors, which is elegantly illustrated in mice lacking PDGF-B or PDGFRβ (3). Deficiency of either PDGF-B or PDGFRβ leads to similar phenotypes with hemorrhages and abnormal mural and mesangial cell coverage (18, 32, 33, 47), demonstrating that both signaling molecules from endothelial cells (e.g., PDGF-B) and receptors on smooth muscle cells (e.g., PDGFRβ) are important for maintaining vascular stability.
In the present study, we observed a partially penetrant phenotype of vascular abnormalities in GPR4-null mice, which shares some similarities with the phenotype of other knockout mice which lack PDGF-B, PDGFRβ, or S1P G protein-coupled receptors. GPR4-null mice had a smaller litter size and an increased perinatal mortality rate. The partial lethality in GPR4-null mice is reminiscent of the phenotype in the mice deficient for both S1P2 and S1P3 receptors (19, 27), in which a bleeding phenotype is detected in a fraction of knockout fetuses. Similar to S1P2−/− S1P3−/− mice, surviving GPR4-null adult mice did not show obvious abnormalities under standard housing conditions, suggesting that certain compensatory mechanisms may come into play during postnatal development. In GPR4-deficient embryos and neonates with hemorrhages, immunohistological analyses revealed dilated and tortuous small blood vessels, decreased smooth muscle cell coating, and reduced mesangial cells in glomeruli, which were also identified in PDGF-B- and PDGFRβ-knockout mice (18, 32, 33, 47). Interestingly, the smooth muscle cell coverage of larger blood vessels was not significantly affected in GPR4-null mice. Reduction of mural cells in selective blood vessels was also observed in PDGF-B- and PDGFRβ-knockout mice (18, 47), suggesting that deficiency of the receptors may differentially influence different vascular walls and signaling redundancy may exist. Nevertheless, our results indicate that GPR4 is functionally involved in maintaining adequate coverage of mural cells and mesangial cells to support vascular stability.
Tissue pH in the body is tightly controlled near pH 7.4. Respiratory excretion of CO2, renal secretion, and bone buffering of acid and base play regulatory roles in maintaining pH homeostasis (22, 28, 31). However, certain pathologies can lead to tissue acidosis. In many diseases such as tumor and ischemia, acidosis and hypoxia coexist and neovascular formation in the acidic and hypoxic environment is important for disease progression (17, 50, 52). Compared to our understanding about the vascular response to hypoxia, less is known about how blood vessels detect pH changes in the microenvironment. Recent identification of GPR4, OGR1, TDAG8, and possibly G2A as proton-sensing GPCRs (20, 34, 37, 44, 56) may provide a novel molecular mechanism for vascular pH sensing. In this study we demonstrated that GPR4 modulates the response of blood vessels to acidic extracellular pH. In the aortic ring assay, microvessel outgrowth from GPR4-null aortas was less inhibited by low pH. In the cell wound healing assay, acidic pH delayed the HUVEC wound closure and overexpression of GPR4 synergistically inhibited this process, consistent with the production of cAMP and its inhibitory effects on angiogenesis (15, 25, 53). Our results suggest that GPR4 functions as an acidic pH sensor for vascular cells, which may provide new insights into the pH-sensing mechanisms of blood vessels. Conceivably, further identification of GPR4 agonists and antagonists may provide a pharmacological approach to regulate vascular response to acidosis in various diseases.
Our studies indicate that GPR4 plays a role in both vascular growth and vascular stability, two commonly related yet distinct processes. With respect to vascular stability, it remains to be determined how the proton-sensing function of GPR4 is connected to the in vivo hemorrhage phenotype in some GPR4-null mice. Since GPR4 is expressed in both endothelial cells and smooth muscle cells (references 23, 35, and 51 and data not shown), its genetic deficiency may possibly affect signaling in both cell types that contributes to vascular stability. Further studies will be directed to understand whether acidosis and GPR4 can modulate the pathways for smooth muscle cell induction, migration, proliferation, and survival. The pH-sensing function of GPR4 may have biological relevance in vascular development as developing embryos have extensive hypoxic regions (30), which can presumably lead to local acidosis. However, our study does not exclude other possible mechanisms for the knockout phenotype. For instance, in addition to protons, other endogenous ligands for GPR4 could also exist. Furthermore, the potential molecular interaction between GPCRs and receptor tyrosine kinases including PDGFR or the heterodimerization between GPR4 and other GPCRs may provide another level of complexity to regulate the stability and growth of blood vessels (40, 54).
GPR4 is highly conserved in vertebrate species during evolution. Amino acid sequence alignment reveals that mouse GPR4 shares >90% homology with orthologs in rat, human, chimpanzee, dog, bovine, and pig genomes and >70% homology with orthologs in zebrafish and puffer fish (L. V. Yang and O. N. Witte, unpublished data). Recent genetic analyses demonstrated that the disruption of gene expression of two GPR4-related genes (>50% homology), either XGPCR4 or XFlop, abolishes cell movement during gastrulation in Xenopus laevis, resulting in early embryonic defects such as reduction or absence of head structures, distorted axes, unclosed blastopores, and severe truncations (8, 49). In comparison, the phenotype that we observed in GPR4-null mice is milder. Adult GPR4-null mice could be produced, and the gastrulation and morphogenesis of knockout embryos appeared to be normal. Instead, we observed hemorrhages in some GPR4-null embryos and neonates and an increase of perinatal mortality. The incompletely penetrant phenotype might be due to local variations in fetal pH and functional redundancy.
Studies show that multiple GPCRs are involved in proton sensing (20, 34, 37, 44, 56). Acidic extracellular pH can also modulate activities of other GPCRs such as the P2Y receptors and the calcium-sensing receptor CaR (42, 57). It should also be noted that intracellular acidic pH may regulate blood vessel function. Studies have shown that ATP-sensitive potassium channels are involved in acidosis-induced coronary arteriolar dilation, and these channels are activated by intracellular acidosis (21, 58). Among the proton-sensing GPCRs, OGR1, which shares the highest sequence homology to GPR4, is responsive to acidic extracellular pH in smooth muscle cell cultures (51). It will be of interest to investigate how these two receptors function coordinately in the pH sensing of blood vessels. We recently reported that mice deficient for TDAG8, another proton-sensing GPCR predominantly expressed in immune cells, exhibit normal immune development, suggesting possible functional redundancy among these receptors (43). Although GPR4 and TDAG8 have distinct expression patterns, they could mediate potential interactions between vascular cells and immune cells under certain acidic conditions such as inflammation. It is frequently noted that GPCR subfamily members responding to the same ligand, such as the S1P and lysophosphatidic acid receptors, may function redundantly (59). Thus, compound knockout of GPR4 and related receptors may provide further insights into the pH-sensing mechanisms of blood vessels and other tissues.
We thank Shirley Quan, LaKeisha Perkins, Donghui Chen, James Johnson, and Mireille Riedinger for expert technical assistance; Alfonso Luque, Sanaz Memarzadeh, Shuling Guo, and Gregory Ferl for helpful discussion; Nathan Lee and Jen Hofmann for help with the confocal microscope; Antoni Ribas and Roger Lo for critically reading the manuscript; the Division of Laboratory Animal Medicine at UCLA for mouse histological evaluation and blood cell counting; and Barbara Anderson for help with manuscript preparation.
L.V.Y. is an Associate and O.N.W. is an Investigator of the Howard Hughes Medical Institute. L.V.Y. is partially supported by NCI/NIH award no. T32 CA009056.
Published ahead of print on 4 December 2006.
†Supplemental material for this article may be found at http://mcb.asm.org/.