Potential G protein-coupled receptor kinase (GRK) and protein kinase A (PKA) mediation of homologous desensitization of corticotropin-releasing factor type 1 (CRF1) receptors was investigated in human retinoblastoma Y-79 cells. Inhibition of PKA activity by PKI5–22 or H-89 failed to attenuate homologous desensitization of CRF1 receptors, and direct activation of PKA by forskolin or dibutyryl cAMP failed to desensitize CRF-induced cAMP accumulation. However, treatment of permeabilized Y-79 cells with heparin, a nonselective GRK inhibitor, reduced homologous desensitization of CRF1 receptors by ∼35%. Furthermore, Y-79 cell uptake of a GRK3 antisense oligonucleotide (ODN), but not of a random or mismatched ODN, reduced GRK3 mRNA expression by ∼50% without altering GRK2 mRNA expression and inhibited homologous desensitization of CRF1 receptors by ∼55%. Finally, Y-79 cells transfected with a GRK3 antisense cDNA construct exhibited an ∼50% reduction in GRK3 protein expression and an ∼65% reduction in homologous desensitization of CRF1 receptors. We conclude that GRK3 contributes importantly to the homologous desensitization of CRF1 receptors in Y-79 cells, a brain-derived cell line.
- G protein-coupled receptor kinase
- corticotropin-releasing factor type 1 receptor regulation
- homologous and heterologous desensitization of the CRF1 receptor
corticotropin-releasing factor (CRF) regulates neuroendocrine, autonomic, and behavioral responses to stress by acting at two high-affinity CRF receptors: CRF1 and CRF2 (2, 14,38). Because CRF1 receptor expression is especially abundant in the amygdala and its extended neurocircuits, as well as in ACTH-secreting corticotropes of the anterior pituitary (7,14, 38), it has been hypothesized that the CRF1receptor is the most important CRF receptor subtype activating neuronal responses to stress.
CRF receptors, like other receptors belonging to the G protein-coupled receptor (GPCR) superfamily, respond to agonist binding by coupling to Gs, the stimulatory heterotrimeric GTP binding protein (9, 14, 38). Gs-coupled GPCRs signal via the cAMP pathway by initiating the adenylyl cyclase-mediated conversion of ATP into cAMP (6, 20, 26, 36, 39). In the continuing presence of high agonist concentrations, GPCR signaling is rapidly terminated via a process termed homologous desensitization (6,20, 26, 36, 39). During the initial phase of homologous desensitization, a G protein receptor kinase (GRK) phosphorylates serines and/or threonines in the activated receptor's COOH terminus or third intracellular loop (4-6, 20, 26, 36, 39). The binding of β-arrestin to the phosphorylated GPCR-GRK complex uncouples the phosphoreceptor from its G protein, terminating signaling (6, 26, 36, 39).
To date, six GRKs have been cloned and sequenced: GRK1 (rhodopsin kinase), GRK2 (β-adrenergic receptor kinase or β-ARK-1), GRK3 (β-ARK-2), GRK4, GRK5, and GRK6 (6, 20, 26, 36, 39). Overexpression experiments performed in cell transfection systems suggest that each GPCR is selectively phosphorylated and desensitized by one or more GRKs (6, 20, 26, 36,39). For example, a recent study has demonstrated that overexpression of GRK2, but not GRK5, augments the homologous desensitization of endothelin receptors (12).
A number of methods have been used to inhibit homologous desensitization of GPCRs. Overexpression of the catalytically inactive dominant negative mutant β-ARK-K220R has been shown to block the action of GRK2 and thereby reduce phosphorylation and desensitization of a number of different GPCRs (6, 21, 26, 36, 39). Recently, GRK antisense cDNA constructs and GRK antisense oligonucleotides (ODNs) have been designed to selectively block the action of a single GRK. For example, two studies demonstrated that cellular uptake of a GRK2 antisense ODN inhibited agonist-induced β2-adrenergic and histaminergic H2 receptor desensitization (31, 45). Similarly, cotransfection of cells with a GRK2 or a GRK5 antisense cDNA construct has been shown to block homologous desensitization of adrenergic and thyroid stimulating hormone receptors, respectively (30, 44). The present study is the first to investigate GRK mediation of CRF1 receptor desensitization.
Our previous experiments revealed that while CRF1 receptors exposed to increasing physiological range concentrations of CRF were rapidly phosphorylated and desensitized, CRF1 mRNA expression did not change (15, 17). The present study was primarily undertaken to investigate the hypothesis that GRK3 homologously desensitizes CRF1 receptors endogenously expressed in human Y-79 retinoblastoma cells. Second, we sought to determine whether protein kinase A (PKA) mediates homologous and/or heterologous desensitization of CRF1 receptors in Y-79 cells. Studies have shown that PKA phosphorylates β-adrenergic receptors during both homologous and heterologous desensitization (36, 39, 40). Because CRF receptors and β-adrenergic receptors both signal via a cAMP-dependent mechanism, we hypothesized that PKA may also play a role in retinoblastoma CRF1receptor desensitization.
MATERIAL AND METHODS
CRF peptides and reagents.
All cell culture reagents were purchased from GIBCO, except for aprotinin (Trasylol; Roche Diagnostics, Mannheim, Germany). Ovine CRF (oCRF) purchased from Bachem (Bubendorf, Switzerland; purity >98%) was used to desensitize CRF receptors and/or stimulate cAMP accumulation in all experiments. Other commercially purchased reagents included the following: 1) BSA (fraction V), heparin, and forskolin (Sigma, Munich, Germany); 2) H-89 and dibutyryl cAMP (DBcAMP) (Calbiochem, La Jolla, CA); and 3) cAMP-dependent PKA inhibitor amide, PKI15–22 (Saxon Biochemicals, Hanover, Germany). The RT-PCR primers used in this study were purchased from Eurogentec (Seraing, Belgium).
Degenerate PCR amplification and subcloning of cDNA constructs.
Poly(A)+ RNA was isolated from Y-79 cells using the Micro Fast Track kit (Invitrogen, San Diego, CA). Human retina Poly(A)+ RNA was purchased from Clonetech (Palo Alto, CA). First-strand cDNA was synthesized from 500 ng retinoblastoma or retinal Poly(A)+ RNA, as described previously (9). Degenerate primers BAR5 and BAR3 targeted to residues 1–6(Met-Ala-Asp-Leu-Glu-Ala; BAR5) and residues 175–180 (Cys-Gln-Trp-Lys-Asn-Val; BAR3) of GRK2 and GRK3 were used. Nucleotide sequences (5′- to 3′- with N = G,A,T, or C; Y =T or C; R = A or G) were as follows: BAR5, ATGGCNGAYYCNGARGC; BAR3, ACRTTYTTCAYTGRCA. GRK2 and GRK3 cDNA were amplified in the following manner: 1) 10 ng Poly(A)+ RNA, 50 pmol of each primer and 1 U Taq DNA polymerase (Roche Diagnostics, Mannheim, Germany) were combined and 2) the PCR reaction was run for 35 cycles (94°C for 20 s, 55°C for 20 s, and 72°C for 1 min). PCR products were then subcloned into pBluescript SK+(Stratagene). Degenerate primers for GRK4–6 were designed based on two conserved regions of the three enzymes, with region 1corresponding to residues 173–178(Gln-Trp-Lys-Trp-Leu-Glu; GRKforw) and region 2corresponding to residues 349–355(Val-Gly-Tyr-Met-Ala-Pro; GRKrev). Nucleotide sequences were as follows: GRKforw 5′-CARTGGAARTGGYTNGA-3′ and GRKrev 5′-GGNGCCATRTANCCNAG-3′. The partial human GRK3 cDNA encoding amino acids 1–180 was excised from pBluescript SK+ with Ecl136 II and Sal I and then subcloned in its antisense orientation into the Eco RV/Xho I cut pcDNA 3 vector (Invitrogen). The human GRK3 partial fragment encodingresidues 1–180 was generated in its sense orientation and subcloned into pcDNA3 using restriction enzymes Hind III and Xbal. The GRK3 antisense construct was designed to target the first 180 amino acids of the GRK3 peptide's NH2 terminus, because this site determines binding selectivity and specificity (6, 26, 36, 39). Sequencing was performed using an ABI 310 DNA sequencer; the GCG software package (Madison, WI) was used for DNA sequence analysis.
Synthesis of GRK3 antisense, random, and mismatched ODNs.
To ensure that the inhibitory action of our antisense ODN was specific, we adhered to the following established methods (22, 29). First, to ensure sequence-specific inhibition of gene expression, we used a GRK3 antisense oligomer 18 bases in length at a final concentration of 1 μM in Y-79 cell culture media. Antisense oligomers of this size, which can selectively discriminate between two gene products that differ by the mutation of a single base, target discrete sites to form specific DNA:RNA hybrids (29). Second, to confirm the specificity of target hybridization, we used a mismatched ODN with 10 base substitutions as a control. Third, to demonstrate that an oligomer identical to our antisense ODN in base composition, but divergent from it in sequence, did not inhibit CRF1receptor desensitization, we used a random (scrambled) ODN as a control. GRK3 antisense, random, and mismatched ODNs were designed and provided by Biognostik (Goettingen, Germany). The sequence for each ODN is as follows: 1) GRK3 antisense ODN (5′-TCCAGTGTCTGCTTTCCT-3′), which corresponds to nucleotides 625–642 of human GRK3 (EMBL accession #X69117);2) random control (5′-CCTTCGTACCCTTTTTCC-3′); or3) mismatched control (5′-TCGACTGTGGACACTGGA-3′).
Cell culture, GRK3 cDNA transfection, and GRK3 antisense ODN treatment.
Y-79 cells endogenously express high-affinity CRF1receptors coupled via Gs to adenylyl cyclase (15), GRK3 mRNA (assessed by Northern blot), and β-arrestin (Dautzenberg and Hauger, unpublished data). In PCR experiments, whereas the human CRF1 receptor was strongly amplified from Y-79 cells, only very weak amplification of a partial human CRF2 receptor fragment was obtained using retinoblastoma DNA (8). Furthermore, mRNA for the human CRF1, but not CRF2, receptor was detected in Northern blots (8). Consequently, we used this brain-derived cell line as a well-controlled model of CRF1receptor regulation (15). Suspension cultures of human retinoblastoma Y-79 cell line (American Type Cell Culture, Rockville, MD) were grown at a density of 2 × 107 cells in 50-ml Falcon culture bottles containing RPMI 1640 (GIBCO) (growth medium) and used between passages 15 and 25 as previously described (15). With the use of the Superfect reagent (Qiagen, Hilden, Germany) as described previously (9), Y-79 cells were transiently transfected with 20 μg plasmid DNA encoding either the empty pcDNA3 vector or the partial GRK3 construct (corresponding to residues 1–180) cloned into pcDNA3 in sense (partial GRK3 sense) or antisense (GRK3 antisense) orientation. For oligodeoxynucleotide experiments, Y-79 cells (2 × 107 cells/group) were exposed to a vehicle (10 mM Tris · HCl, 1 mM EDTA, pH 7.4) or to the GRK3 antisense, random, or mismatched ODN at a final concentration of 1 μM. ODN uptake was then allowed to proceed while cells were incubated in growth medium at 37°C for 60 h before the beginning of a desensitization experiment.
CRF receptor desensitization protocol.
Retinoblastoma cells were centrifuged at 150 g for 10 min at room temperature and then resuspended at a density of 2 × 106 cells/ml. For PKA experiments, Y-79 cells were pretreated for 30 min at 37°C with the PKA inhibitor H-89 at a final concentration of 3–20 μM. For GRK3 antisense experiments, cells were resuspended in a similar manner. Because human CRF binds to the CRF binding protein and has a shorter half-life than oCRF, oCRF (which does not bind to the CRF binding protein) was used to desensitize retinoblastoma CRF receptors (14). oCRF at a final concentration of 100 nM was added to the cell suspension, and the treated cells were incubated for 60 min at 37°C using our previously published protocol (15). At the end of the incubation period, cells were washed three times with 40 volumes of growth medium and resuspended in stimulation media [growth media containing 1 mg/ml BSA, 5 mM 3-isobutyl-1-methylxanthine (IBMX), and 0.05 mg/ml ascorbate, pH 7.4 (basal)] or stimulation media with 100 nM CRF (to determine the maximum response for CRF-stimulated cAMP accumulation).
Retinoblastoma cell permeabilization experiments.
In experiments determining the effect of heparin (1 or 5 μM) and the PKA inhibitor PKI5–22 (1 μM) on CRF receptor desensitization, Y-79 cells (∼2 × 106 cells) were resuspended in 33 mM PIPES buffer containing (in mM) 2 MgCl2, 2 EGTA, 5.4 KCl, 0.44 KH2PO4, 137 NaCl, 5.6 glucose, 1.3% polyethylene glycol (molecular weight 8,000), and 1 ATP, pH 7.4. With the use of previously published methods (1, 25), Y-79 cells were permeabilized by being incubated with 0.005% digitonin or 0.5 μg/ml streptolysin O (SLO) for 1 min. Immediately afterward, permeabilized cells (determined by trypan blue uptake) were pretreated for 1 min with heparin (1 μM) or PKI5–22 (1 μM) before the addition of 10 or 100 nM CRF for 30 min. A longer period of CRF exposure was not possible because of the toxicity resulting from cell permeabilization. Basal and CRF-stimulated cAMP accumulation was measured in the presence of stimulation media containing 10 μM GTP and 100 μM ATP.
After extensive cell washing, intracellular cAMP levels were measured in ether-extracted and acetylated cell lysates using a double-antibody RIA kit (cAMP 125I-assay system, RPA 509; Amersham International, Little Chalfont, UK), as previously described (9,15).
Semiquantitative RT-PCR and dot blot analyses.
Using our previously published method for semiquantitative RT-PCR (34), we measured either the relative expression of GRK2 and GRK3 in human retina and Y-79 cells or the effect of ODN uptake on mRNA levels of GRK2 and GRK3. GRK2 and GRK3 were amplified from human retina and Y-79 cells using a common primer GRKco [5′-ATGGCGGACCTGGAGGC-3′, nucleotides 108–126 of human GRK2 (EMBL accession #X61157)] and nucleotides 1–17 of human GRK3 (EMBL accession #X69117)] and specific primers for GRK2 (5′-TGGAAGAGATCCGGAGG-3′, nucleotides 547–531 of human GRK2) and for GRK3 (5′-GTGCACTTCTGCAAAGG-3′, nucleotides 319–303 of human GRK3). Amplification was performed with cDNA corresponding to 10 ng Poly(A)+ RNA and 1 U Taq DNA-Polymerase (Roche Diagnostics) for 25 cycles (94°C for 10 s, 55°C for 20 s, 72°C for 1 min). PCR products were separated on a 2% agarose gel and transferred to Nylon membranes (Amersham). Hybridization was performed overnight at 37°C in a solution consisting of 6× sodium chloride-sodium citrate (SSC), 5× Denhardts solution, 100 mg/ml yeast tRNA, and 0.1% SDS with a 32P-labeled GRKco primer. The blot was washed at room temperature with a solution of 2× SSC and 0.1% SDS followed by a final wash at 37°C with a solution of 2× SSC and 0.1% SDS containing a 32P-labeled GRKco primer. Next, the blot was washed twice for 15 min each with a solution of 2× SSC and 0.1% SDS, once at room temperature and a final time at 37°C. To quantitate the antisense ODN treatments, GRK2 DNA was amplified using the specific primers BAR15 (5′-sequence-3′, position 1388–1411 of human GRK2) and BAR13 (5′-sequence-3′, position 1828–1804 of human GRK2), whereas GRK3 DNA was amplified using the primers BAR25 (5′-sequence-3′, position 1344–1369 of human GRK3) and BAR23 (5′-sequence-3′, position 1721–1696 of human GRK3). Human β-actin, providing an internal control, was amplified using primers act1 and act2, as described previously (15, 34). For each amplification, 25 pmol of the GRK primer set and 1 pmol of the actin primer set were combined with 1 U Taq DNA Polymerase (Roche Diagnostics), and PCR was performed for 25 cycles (94°C for 15 s, 65°C for 15 s, and 72°C for 75 s). These conditions result in the amplification of transcripts in the linear range (<35% or the maximal amplification). After completion of the amplification, 5 μl of the PCR reaction was run on a 1.5% agarose gel as a control for successful amplification. GRK2, GRK3, and β-actin cDNA expression were quantitated by immobilizing 2 μl of each reaction in quadruplicate on Hybond N Nylon membranes (Amersham). The blots were hybridized to32P-labeled specific ODNs coding for GRK2 (5′-ATCAGGAGCTCTACCGC-3′), GRK3 (5′-AAAGTACCCACCACCCT-3′), or β-actin (5′-GTGATGGACTCCGGTGA-3′). Hybridization and washing conditions were performed overnight at 37°C in a solution consisting of 6× SSC, 5× Denhardts solution, 100 mg/ml yeast tRNA, and 0.1% SDS. Afterward, hybridized blots were washed at room temperature with 2× SSC and 0.1% SDS, followed by an incubation in 2× SSC and 0.1% SDS for 15 min at 37°C. Squares, 1 cm2 in size, which contained the radioactive sample, were cut out and counted using a β-counter (Hewlett Packard).
Western blot quantitation of GRK protein expression.
GRK2 and GRK3 are endogenously synthesized in cells as phosphoproteins (36, 39). To avoid potential proteolysis and dephosphorylation that may occur when measuring endogenously phosphorylated GRKs (36, 39), the following protease and phosphatase inhibitors were added to the lysis buffer (20 mM Tris · HCl, pH 7.5; 150 mM NaCl, 10 mM EDTA, 1% Triton-X100, 1% wt/vol sodium deoxycholate): 1 mM PMSF, 5 μg/ml aprotinin, 1 μg/ml pepstatin, 10 μg/ml soybean trypsin inhibitor, 20 μM leupeptin, 10 μg/ml benzamide, 1 mM NaF, and 1 mM sodium orthovanadate. Retinoblastoma cells (∼2 × 107 cells per transfection group) were lysed by incubating them in 200–300 μl of the above lysis buffer for 1 h at 4°C. Afterward, insoluble cell fractions were removed by centrifugation in a microfuge at 14,000 rpm for 10 min at 4°C. Protein concentrations of supernatants were determined using the BioRad DC protein assay. Cell lysates were used immediately after lysis or after storage at −70°C for no more than 1 wk. SDS-PAGE and GRK immunoblotting were performed according to the protocol of Oppermann et al. (33). After lysates were boiled for 3 min to denature cellular proteins, they were typically loaded at a concentration of 40 μg per lane on to 10% Tris-glycine gels (Invitrogen-NOVEX, Carlsbad, CA) that had been placed in a NOVEX Xcell II Mini-Cell System. For each gel, separate lanes were also loaded with molecular weight markers (Amersham RPN800) and GRK2 and GRK3 protein standards (standards were purified from SF9-infected insect cells and kindly provided by Dr. R. Lefkowitz, HHMI, Duke University). After the addition of Tris-glycine SDS Running Buffer (Invitrogen-NOVEX, #LC2675) to the cathode and anode chambers, proteins were resolved using SDS-PAGE under reducing conditions (Invitrogen-NOVEX Tris-glycine SDS sample buffer containing 2.5% mercaptoethanol) at a fixed 125 V (current 35–40 A) for 90 min according to the method of Laemmli (15). Western transfer of resolved retinoblastoma proteins on to polyvinylidene difluoride (PVDF) membranes (Invitrogen-NOVEX) was accomplished by exposing the gel/membrane transfer sandwich inside the NOVEX XCell II Mini-Blot Module to 25 V (current 100 mA) for 2 h under reducing conditions. After completion of the Western transfer, transfer efficiency and equal protein loading of lanes were confirmed by staining PVDF membranes with BLOT-FastStain (Geno Technology). After PVDF membranes were destained, blots were blocked for 1 h in a solution of TTBS (20 mM Tris, pH 7.5, 500 mM NaCl, 0.2% Tween 20) with 5% nonfat dried milk (5 g/100 ml) (TTBS-Blotto) with constant shaking at room temperature. Blots were then washed with TTBS and immunoprobed with a mouse GRK2–3 monoclonal antibody (C5/1; kindly provided by Dr. R. Lefkowitz, HHMI, Duke University) at a dilution of 1:1,000 overnight (∼18 h) at 4°C with constant shaking. The C5/1 monoclonal recognizes a COOH-terminal epitope common to GRK3 and GRK2 (33). In certain experiments, cell lysate proteins were resolved a second time by SDS-PAGE, and the resultant Western blots were immunoprobed with one of the following polyclonal antibodies targeted to COOH-terminal amino acid sequences unique to GRK3 (Santa Cruz Biotechnology, Santa Cruz, CA): 1) a rabbit polyclonal GRK3 IgG (C14, sc-563) at a concentration of 1:200 or2) a goat GRK3 IgG (E15, sc-9306) at a concentration of 1:100. After the membranes were washed in TTBS (6 × 10-min washes), blots were incubated for 1 h at room temperature with constant shaking with one of the following antibodies (in TTBS-Blotto):1) sheep anti-mouse IgG-HRP (NA931, 1:10,000; Amersham Pharmacia Biotech, Piscataway, NJ), 2) sheep anti-rabbit IgG-horseradish peroxidase (HRP) (NA934, 1:5,000; Amersham Pharmacia Biotech), or 3) donkey anti-goat IgG-HRP (sc-2033, 1:5,000; Santa Cruz Biotechnology). After membranes were washed extensively in TTBS (6 × 20-min washes), chemiluminescent detection of Western blots was performed using ECL+ Plus (Amersham Pharmacia Biotech).
Data reduction and statistical analyses.
CRF receptor desensitization data were calculated as percent control or percent desensitization values, as previously described (15). Data reduction for the cAMP RIA was performed using the StatLIA log-logit program (Brandon). ANOVAs across experimental groups were performed on a MacIntosh personal computer using StatView Student (Abacus Concepts, Berkeley, CA) and PRISM, version 2.0 (GraphPad Software, San Diego, CA). If the one-way ANOVA was statistically significant, planned post hoc analyses were performed using both Tukey's and Bonferroni's multiple-comparison tests to determine individual group differences. Quantitative densitometric analyses of DNA-DNA hybridization Southern blots were performed by developing and scanning the film followed by analysis using a BAS reader (Phosphorimager with WinCam2.2 software). Immunoreactive GRK2–3 protein bands on Western blots were quantitated and analyzed on the STORM imager using ImageQuant software (Molecular Dynamics). Figures of Western blots (*tif files) were prepared for publication using Adobe Photoshop (version 5.0) and Illustrator (version 8.0) software programs for MacIntosh personal computers.
Effect of PKA on the desensitization of retinoblastoma CRF1 receptors.
PKA has been shown to homologously and/or heterologously desensitize a number of GPCRs by phosphorylating consensus sites on the receptors' intracellular loops (36, 39, 40). We began our study by investigating potential PKA mediation of retinoblastoma CRF1 receptor desensitization. Pretreating Y-79 cells with 50 μM forskolin or 4 mM DBcAMP for up to 3 h to maximally stimulate PKA activity produced no measurable CRF1 receptor desensitization (Fig. 1). However, 62% desensitization (P < 0.0001) of CRF-stimulated cAMP accumulation was observed in Y-79 cells exposed for 3 h to 10 nM CRF (Fig. 1). The magnitude of forskolin-stimulated cAMP accumulation was not significantly decreased by pretreating Y-79 cells with CRF, forskolin, or DBcAMP (Fig. 1). Seven additional experiments showed that CRF1 receptor desensitization was unaltered in Y-79 cells preincubated for 30 min with 3 or 20 μM H-89, a cell-permeable PKA inhibitor (37, 40), and then exposed to either 10 nM CRF for 1–3 h or 100 nM CRF for 1 h (Fig.2, A and B). Although H-89 has been shown to act as both a PKA inhibitor and β-adrenergic receptor antagonist (37), it did not attenuate CRF-stimulated cAMP accumulation in a control group of Y-79 cells (Fig. 2 B).
Effects of heparin and PKI5–22 pretreatment on CRF1 receptor desensitization in permeabilized Y-79 cells.
Preliminary experiments showed that homologous desensitization of CRF1 receptors in Y-79 cells permeabilized with SLO was inhibited 34% (P < 0.01) by pretreating cells with 1 μM heparin, a nonselective GRK inhibitor (25, 26, 39), before exposure to 100 nM CRF for 30 min (Fig.3). Similarly, heparin pretreatment (1 μM) inhibited CRF1 receptor desensitization in Y-79 cells permeabilized with 0.005% digitonin and then exposed to 10 nM CRF for 30 min (data not shown). However, pretreatment of Y-79 cells with 1 μM PKI5–22, a potent PKA inhibitor (42), failed to block CRF1 receptor desensitization (Fig. 3). Because heparin, but not PKI5–22, significantly attenuated homologous desensitization of CRF1 receptors, we proceeded to investigate GRK-mediated regulation of CRF1 receptor signaling.
GRK expression in Y-79 cells and human retina.
The expression of GRK2 and GRK3 mRNA in Y-79 cells was quantified using a degenerate RT-PCR approach. With the use of BAR5 and BAR3 primers, cDNAs for GRK2 and GRK3 were amplified in both Y-79 cells and human retina (data not shown). Unlike human GRK2 mRNA, human GRK3 mRNA undergoes extensive alternative splicing (5, 35). In keeping with this observation, our Northern blot analyses revealed the presence of several GRK3-specific mRNA bands in Y-79 cells (data not shown). Semiquantitative RT-PCR performed using the common primer GRKco in combination with specific primers for GRK2 and GRK3, revealed that GRK3 mRNA expression was only slightly less than GRK2 mRNA expression in retinoblastoma cells (Fig.4 A). In contrast, GRK2 mRNA expression was markedly greater than GRK3 mRNA expression in retina (Fig. 4 A). Southern blot measurement of GRK3 cDNA revealed that this kinase accounted for 39.7 ± 2.6% of total GRK cDNA in retinoblastoma cells (Fig. 4 B). In human retinal tissue, on the other hand, GRK3 cDNA accounted for only 12.7 ± 1.6% of total GRK cDNA (Fig. 4 B).
Having observed that GRK3 mRNA expression was particularly abundant in Y-79 cells (Fig. 4) and increased about threefold during homologous desensitization of CRF1 receptors (13), we undertook a series of experiments to determine the effect of GRK3 antisense on CRF1 receptor desensitization by measuring changes in the cellular expression of GRK2 and GRK3 following GRK3 antisense ODN uptake or GRK3 antisense cDNA transfection. Semiquantitative RT-PCR and dot blot measurements revealed that GRK3 antisense ODN pretreatment significantly decreased GRK3 mRNA expression by 48.4 ± 1.8% (P < 0.0001) (Fig.5) without altering GRK2 expression (Fig.5).
Immunoblotting of retinoblastoma cell lysates probed with a mouse monoclonal antibody that recognizes both GRK2 and GRK3 (C5/1) confirmed the presence in Y-79 cells of the GRK3 protein (33). Blots were run with purified protein standards for GRK3 and GRK2 that migrated to ∼78 and ∼80 kDa, respectively, the known molecular weights of these kinases (36, 39) (Fig.6). Well-defined cell lysate bands that migrated to a position parallel to the GRK3 standard were identified as GRK3 protein (Fig. 6). However, no immunoreactive bands were detected at the position of the GRK2 standard. Densitometric quantitation of Western blots revealed that GRK3 protein levels decreased 51.2 ± 7.4% in Y-79 cells transfected with the GRK3 antisense cDNA construct (P = 0.0001; Fig. 6). Additional immunoblotting experiments were performed using one of two polyclonal antibodies (E15/#sc-9306 or C14/#sc-563) that detect GRK3, but not GRK2. Typically, the GRK3 polyclonal antibodies detected faint bands migrating to ∼78 kDa (the molecular weight of GRK3), as well as numerous darker bands migrating to different molecular mass positions. Nevertheless, Western blots immunoprobed with the polyclonal antibodies E15/#sc-9306 or C14/#sc-563 revealed a ∼50% reduction in GRK3 protein expression in GRK3 antisense-transfected cells (data not shown).
Effect of GRK3 antisense ODN uptake or GRK3 antisense cDNA transfection on homologous desensitization of retinoblastoma CRF1 receptors.
Next, we investigated changes in CRF1 receptor desensitization in Y-79 cells exposed to a GRK3 antisense ODN or transfected with a GRK3 antisense cDNA construct. Homologous desensitization of CRF1 receptors was markedly less in Y-79 cells pretreated with an 18-mer GRK3 antisense ODN and exposed to 100 nM CRF for 1 h (17.7 ± 1.9% desensitization,P < 0.001) than in Y-79 cells pretreated with a mismatched ODN (34.9 ± 2.3%), a random ODN (32.6 ± 2.6%), or buffer (39.2 ± 2.1%) and exposed to 100 nM CRF for 1 h (Fig. 7).
Next, a human GRK3 cDNA encoding amino acids 1–180 (i.e., nucleotides 1–540 of the GRK3 gene) was cloned into a pcDNA3 vector in both sense (GRK3 sense) and antisense (GRK3 antisense) orientations. Each of three groups of Y-79 cells was then transiently transfected with one of the following constructs: 1) the pcDNA3 vector (control group), 2) the GRK3 sense construct, or 3) the GRK3 antisense construct. Two days after transfection, each group of Y-79 cells was exposed to 100 nM CRF for 1 h. The magnitude of CRF1 receptor desensitization was similar in the group of Y-79 cells transfected with the GRK3 sense construct (48.3 ± 3.4% desensitization) and in the control group (41.7 ± 3.2% desensitization) (Fig.8). In contrast, homologous desensitization of CRF1 receptors decreased by ∼65% (15.0 ± 2.1% desensitization, P < 0.0001) in Y-79 cells transfected with the GRK3 antisense construct (Figs. 8 and9). Furthermore, the maximum level of CRF-stimulated cAMP accumulation was 1.8-fold greater in Y-79 cells transfected with the GRK3 antisense construct than in Y-79 cells transfected with the empty vector (Fig.9). It is important to note that inhibition of CRF1 receptor desensitization in Y-79 cells was greater when cells were transfected with GRK3 antisense cDNA than when they were exposed to GRK3 antisense ODN (Figs. 7-9). The greater inhibition of CRF1 receptor desensitization observed in GRK3 antisense cDNA-transfected cells compared with GRK3 antisense ODN-exposed cells, may be attributable to more efficient cellular accumulation and nuclear localization of antisense via transfection than via uptake (29). For example, studies conducted in A431 epidermoid carcinoma cells revealed that GRK2 antisense cDNA transfection more effectively inhibited homologous desensitization of β2-adrenergic receptors than did GRK2 ODN uptake (44, 45).
Although one group has reported that basal cAMP levels increased twofold in follicle-stimulating hormone receptor cDNA-transfected Ltk cells after GRK2 or GRK3 antisense cotransfection (47), other studies have shown that basal cAMP levels remained constant in native cells endogenously expressing GPCRs after GRK2 antisense cDNA transfection or GRK2 antisense ODN uptake (6,18, 21, 27, 36, 49). Similar to these studies, we found that basal levels of cAMP remained unaltered in retinoblastoma cells transfected with our GRK3 antisense cDNA (Figs. 9 and 10).
We also determined that phosphodiesterase (PDE) activity was not altered by transfecting Y-79 cells with the GRK3 antisense construct. Retinoblastoma cells not treated with IBMX exhibited five- and sevenfold lower (P < 0.0001) levels of basal and CRF-stimulated cAMP accumulation, respectively, than did control or GRK3 antisense-transfected cells incubated with 5 mM IBMX (Fig.10). The magnitude of CRF-stimulated cAMP accumulation was the same in control and GRK3 antisense-transfected cells incubated without IBMX (Fig. 10). However, in the presence of IBMX, CRF-stimulated cAMP accumulation was significantly greater (25.0 ± 2.7% increase, P< 0.001) in GRK3 antisense-transfected Y-79 cells than in control cells (Figs. 9 and 10). This finding may be attributable to the development of GRK3 antisense transfection-induced inhibition of CRF receptor desensitization during the 30-min stimulation of Y-79 cells with 100 nM CRF.
Collectively, our data suggest that GRK3 plays an important role in the homologous desensitization of CRF1 receptors endogenously expressed in Y-79 cells. Heparin, a non-selective GRK inhibitor (25, 26, 39), produced a small but significant reduction (34%) in CRF-induced retinoblastoma CRF1receptor desensitization. Cellular uptake of a GRK3 antisense ODN, but not of a random or mismatched ODN, decreased GRK3 mRNA expression by ∼50% without altering GRK2 mRNA expression, and inhibited homologous desensitization of CRF1 receptors by ∼55%. Transient transfection of Y-79 cells with a GRK3 antisense cDNA construct decreased GRK3 protein expression by ∼50% and CRF1receptor desensitization by ∼65%, whereas the same procedure performed using an empty vector or a partial GRK3 sense construct failed to alter CRF1 receptor desensitization. These findings, coupled with our recently published data showing that the human CRF1 receptor undergoes rapid phosphorylation in response to agonist exposure but not in response to PKA or calcium/calmodulin-dependent kinase activation (17), suggest that GRK3-mediated phosphorylation contributes importantly to the homologous desensitization of retinoblastoma CRF1receptors.
Data generated by the second set of experiments described in this paper suggest that PKA plays no role in either the homologous or heterologous desensitization of retinoblastoma CRF1 receptors. We observed no appreciable heterologous desensitization of CRF1 receptors in Y-79 cells exposed to forskolin or DBcAMP, although both of these substances maximally activate PKA. Furthermore, exposure of Y-79 cells to two PKA inhibitors, H-89 and PKI5–22 (40, 42), failed to block homologous desensitization of CRF1 receptors. Our finding that PKA does not contribute to CRF1 receptor desensitization in Y-79 cells is surprising for two reasons. First, PKA-induced phosphorylation is known to regulate a spectrum of processes in postsynaptic brain neurons, including receptor signaling, ion channel activity, and gene transcription (32). Second, consistent with previous data demonstrating that PKA phosphorylates and desensitizes β-adrenergic receptors (26, 39), our preliminary experiments revealed that PKA desensitized β-adrenergic receptors in both Y-79 and AtT-20 cells, a mouse pituitary tumor cell line expressing CRF1 receptors (Hauger and Dautzenberg, unpublished data). Thus, although PKA does not homologously or heterologously desensitize CRF1 receptors in Y-79 and AtT-20 cells, this cAMP-dependent kinase does regulate β-adrenergic receptor signaling in these cell lines.
Several recent studies suggest that PDEs play a role in the desensitization of GPCRs expressed in peripheral cells. For example, a study performed using Jurkat T cells found that β2-adrenergic receptor desensitization increased approximately twofold when PDE expression was upregulated by chronic PKA activation (43). However, we conclude that PDE upregulation plays no role in CRF1 receptor desensitization in retinoblastoma cells for the following reasons. First, our data indicate that PKA is not involved in either the homologous or heterologous desensitization of CRF1 receptors endogenously expressed in Y-79 cells. Second, GRK-mediated desensitization of the CRF1 receptor occurs over the first 60 min of agonist exposure (Figs. 6-9), whereas PDE activity has been found to increase only after cells have been exposed to agonist for a prolonged period of time (i.e., 2–24 h) (23).
We chose to employ an antisense strategy to inhibit GRK activity because antisense appears to inhibit GRK action more selectively and completely than other methods. However, the use of antisense is somewhat controversial because the cellular events mediating its effects are not well understood (22, 29). Several investigators have proposed that antisense exerts its effects in peripheral cells by interacting (through complementary Watson-Crick base pairing) with specific, mature mRNA sequences, thereby reducing ribosomal read-through via the RNase H mechanism (22, 29). Recent data suggest that the RNase H mechanism is not operative in the central nervous system and that other processes such as splicing inhibition and translational arrest may mediate antisense-induced decreases in the expression of targeted cellular proteins in the brain (22).
Because cell-permeable GRK inhibitors are not yet available, in addition to antisense, the following methods have been used to selectively or nonselectively attenuate the activity of GRKs. Heparin has been used to inhibit GRK-mediated GPCR phosphorylation and desensitization. However, this polyanion has two disadvantages. First, it nonselectively blocks the action of GRK2, GRK3, GRK5, and GRK6 (26). Second, brain cell lines poorly tolerate permeabilization, a process that is required for intracellular heparin entry. The dominant negative mutant GRK2-K220R has been used to block GRK2-mediated homologous desensitization of GPCRs in both cell overexpression systems and native cells (6, 18, 21, 24, 26, 27,36, 39). Because GRK2-K220R substitutes an arginine for the lysine normally occupying position 220 of the GRK2 central catalytic domain (the locus of the phosphotransfer reaction) (6, 26, 36,39), it specifically but incompletely blocks the phosphorylating activity of GRK2. However, to date it has not been determined whether GRK2-K220R also inhibits the actions of GRKs other than GRK2. Although specific monoclonal antibodies and antagonist peptides targeting GRK2 or GRK3 have been used to inhibit the homologous desensitization of adrenergic, angiotensin AT1A, and odorant receptors (10, 11, 26, 39, 41), this method of inhibiting GRK activity requires either cell permeabilization or injection of the antibody into the cytosol, procedures that are poorly tolerated by Y-79 cells. The use of antisense allowed us to inhibit GRK3 action without damaging retinoblastoma cell integrity.
Overexpression of GRKs in heterologous, artificial cell systems frequently results in nonselective phosphorylation and desensitization of GPCRs. Thus antisense strategies are being increasingly used to investigate the specificity of GRK-GPCR interactions in native cell lines endogenously expressing signaling proteins in a physiologically correct setting (6, 24, 30, 31, 44, 45, 47-49). Antisense studies performed in native cell lines have provided compelling new evidence that GRK2 and GRK5 selectively desensitize specific GPCRs. For example, one study reports significant inhibition of TSH receptor desensitization and GRK5 expression in rat thyroid FRTL5 cells transfected with an antisense cDNA construct targeted to the catalytic domain of GRK5 (30). Another recent study found that cellular uptake of a GRK2, but not a GRK6, antisense ODN decreased homologous desensitization of histaminergic H2 receptors endogenously expressed in human gastric carcinoma MKN-45 cells (31). In addition, strong inhibition of endogenous α2c, but not endogenous 5-hydroxytryptamine IB, receptor desensitization has been observed in opossum kidney cells transfected with a GRK2 antisense construct (24). Finally, Willets et al. (49) obtained a marked decrease in GRK2, but not GRK3, expression and strong attenuation of agonist-induced adenosine A2A receptor desensitization by transfecting NG108–15 rodent neuroblastoma-glioma cells with an antisense cDNA targeting the GRK2 COOH terminus (49). This group also found that secretin and IP-prostanoid receptor desensitization remained unaltered in GRK2 antisense-transfected neuroblastoma-glioma cells (49). Thus GRK2 has been shown to selectively desensitize adenosine A2A receptors in the NG108–15 cell line, which expresses GRK2 protein at a fivefold higher level than GRK3 protein (28, 49). Selective targeting of GPCRs by GRKs has also been observed in GRK knockout and GRK-overexpressing transgenic mice (20).
To date, several studies have investigated GRK3-mediated desensitization of GPCRs in neuronal cell systems. Diverse-Pierluissi et al. (11) showed that an inhibitory peptide targeted to the COOH-terminal Gβγ-binding site of GRK3 significantly reduced desensitization of α2-adrenergic receptors regulating a neuronal voltage-dependent calcium channel (11). Two other noteworthy studies demonstrated that a monoclonal antibody targeted to GRK3, but not one targeted to GRK2, markedly reduced odorant-induced phosphorylation and desensitization of olfactory receptors in nasal cilial neurons (10, 41). This finding is consistent with the observation that GRK3 expression is approximately eightfold greater than GRK2 expression in olfactory neuroepithelial cells (10,41).
The present study demonstrates for the first time that GRK3 contributes importantly to the homologous desensitization of the human CRF1 receptor endogenously expressed in human retinoblastoma Y-79 cells. Our preliminary data indicated that GRK3 expression increased significantly in Y-79 cells during prolonged activation of CRF1 receptors (13). Six additional experiments showed that CRF1 receptor desensitization decreased 64.1 ± 5.4% (P < 0.0001) in GRK3 antisense-transfected Y-79 cells (Figs. 8 and 9). Western blots immunoprobed with the C5/1 monoclonal antibody revealed a decrease of 51.2 ± 7.4% (n = 11;P = 0.0001) in GRK3 protein expression in GRK3 antisense-transfected Y-79 cells (Fig. 6). In addition, Western blots immunoprobed with either of two GRK3 polyclonal antibodies (E15/#sc-9306 or C14/#sc-563) revealed an ∼50% decrease in GRK3 protein levels in GRK3 antisense-transfected Y-79 cells. We observed these strong biological effects despite having used transient transfection, a method that is generally less efficient than stable transfection. Although GRK3 appears to play a dominant role in the homologous desensitization of retinoblastoma CRF1receptors, we cannot rule out the possibility that other GRKs regulate CRF1 receptor signaling in different cell systems. Because GRK2 and GRK3 share ∼80% sequence homology and GRK2 is abundantly expressed in the central nervous system, it is very likely that GRK2 also contributes to the homologous desensitization of brain CRF receptors. To test this hypothesis, we are using the GRK2 antisense construct developed by Willets et al. (49) to investigate GRK2-mediated CRF1 receptor desensitization in brain-derived cell lines endogenously expressing CRF1receptors and GRK2 protein.
The highly localized and concentrated presence of GRK2, GRK3, β-arrestins, and regulators of G-protein signaling (RGS proteins) at neuronal synapses in brain regions abundantly expressing CRF1 receptors suggests that these molecules contribute importantly to the regulation of CRF1 receptor signaling during stress (3, 7, 14, 38). We posit that a genetically or developmentally induced deficit in the expression of GRK3 that impairs homologous desensitization of CRF1receptors may prolong and/or intensify CRF neurotransmission during stress. Stressful events often precipitate episodes of bipolar I disorder and schizophrenia, and overactivation of the stress axis is a prominent feature of mania (16). High cortisol levels during mania appear to correlate with the development of comorbid anxiety, irritability, dysphoria, and depression (16). Tiberi et al. (46) demonstrated that GRK3 homologously desensitizes the dopamine D1 receptor, which is known to mediate the mania- and psychosis-like effects produced by amphetamine. It is widely believed that increased dopaminergic neurotransmission plays a major role in the pathogenesis of mania and psychosis (16). Thus a GRK3 deficit may constitute a vulnerability factor for bipolar I disorder and schizophrenia by intensifying and/or prolonging both CRF and dopaminergic neurotransmission. Our hypothesis that a GRK3 deficit contributes to the pathophysiology of mania, bipolar mixed states, and psychosis is supported by the observation that the GRK3 gene maps to chromosome 22q11 (5, 26, 36), a region of the human genome that has been identified as a potential locus for genetic abnormalities involved in both bipolar illness and schizophrenia (19).
The authors are greatly indebted to Dr. R. Lefkowitz and C. Stone, who kindly provided the mouse monoclonal C5/1 antibody, GRK2 and GRK3 protein standards, the Western immunoblotting protocol, and invaluable advice. In addition, the authors thank Drs. M. Caron, K. Catt, M. Lohse, and P. Insel for advice. We also acknowledge A. Turken and S. Wille for expert technical assistance. Dr. S. Bhakdi (Institute for Medical Microbiology, Johannes-Gutenberg University, Mainz, Germany) kindly provided highly purified streptolysin O. Finally, completion of this manuscript would not have been possible without the superb editorial assistance of S. Shew.
R. L. Hauger received support from a Veterans Affairs Merit Review grant; the VA VISN 22 Mental Illness Research, Education and Clinical Center (MIRECC); and the National Institute of Mental Health Mental Health Clinical Research Center (PHS MH-20914–14). A portion of this research received financial support from the Max Planck Institute for Experimental Medicine, Department of Molecular Neuroendocrinology (Director: Dr. Joachim Spiess), Goettingen, Germany.
Address for reprint requests and other correspondence: R. L. Hauger, VA Medical Center and Dept. of Psychiatry, Univ. of California, San Diego, 9500 Gilman Drive, La Jolla, CA 92093-0603 (E-mail:).
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2001 the American Physiological Society