Src Regulates Constitutive Internalization and Rapid Resensitization of a Cholecystokinin 2 Receptor Splice Variant *

The third intracellular loop domain of G protein-coupled recep-torsregulatestheirdesensitization,internalization,andresensitiza- tion. Colorectal and pancreatic cancers, but not the nonmalignant tissue, express a splice variant of the cholecystokinin 2 receptor (CCK 2 R) called CCK 2i4sv R that, because of intron 4 retention, contains an additional 69 amino acids within its third intracellular loop domain. This structural alteration is associated with agonist-inde-pendent activation of Src kinase (Olszewska-Pazdrak, B., Townsend, C. M., Jr., and Hellmich, M. R. (2004) J. Biol. Chem. 279, 40400–40404).The purpose of the study was to determine the roles of intron 4 retention and Src kinase on CCK 2i4sv R desensitization, internalization, and resensitization. Gastrin1–17 (G17) binds to bothCCK 2 RandCCK 2i4sv RandinducesintracellularCa 2 (cid:1) ([Ca 2 (cid:1) ] i ) increases. Agonist-induced increases in [Ca 2 (cid:1) ] i were used to assess receptor activity. Src kinase activity was inhibited by transducing cells with a retrovirus containing

Receptor desensitization and resensitization are essential mechanisms for maintaining physiologically appropriate cellular responses to extracellular stimuli. For G protein-coupled receptors (GPCR 2 (s)), these processes are regulated, in part, by the intracellular domains of the receptors. Within seconds to minutes of agonist binding, the activated receptor is desensitized when specific amino acid residues within the 3il and/or C-terminal tail domain of the receptor are phosphorylated, creating binding sites for proteins, such as ␤-arrestins and Src kinase, which block further receptor interactions with heterotrimeric G proteins and initiate receptor internalization (1)(2)(3). Resensitization involves the dephosphorylation of the intracellular domains and recycling of the intact receptor back to the plasma membrane where it can once again bind ligand (4,5). Dysregulation of these processes can lead to disease (6,7). CCK 2 R is a GPCR that exists as various splice variant isoforms (8). Previously, we reported the identification and cloning of a variant, called CCK 2i4sv R, which, as a result of intron 4 retention, contains an additional 69 amino acid residues in its 3il domain (9). In contrast to CCK 2 R, which is widely expressed in normal tissues of the gastrointestinal tract and nervous system, CCK 2i4sv R appears to be the predominant variant expressed by some hyperplastic polyps, colorectal and pancreatic cancers, and pancreatic cancer-derived cell lines (9 -12). Several lines of evidence support a role for CCK 2i4sv R in the progression of these cancers. First, CCK 2i4sv R, but not CCK 2 R, coimmunoprecipitates with activated Src kinase in an agonist-independent manner (13). Increased Src activity is associated with the development and progression of breast, brain, pancreas, and colon cancers (14,15). Second, expression of CCK 2i4sv R stimulates cell growth in vitro (9) and HEK293 tumor growth in vivo in a Src-dependent fashion (16). Third, inhibition of CCK 2i4sv R expression, using antisense oligonucleotides, slowed agonist (gastrin1-17 (G17))-stimulated growth of BxPC3 pancreatic cancer cells in vitro (17). Finally, treatment with G17 promotes proliferation, motility, and invasion in various in vitro and in vivo cancer models in pancreatic and colorectal cancer cells (18 -21). Together, these studies suggest that CCK 2i4sv R is involved in the pathophysiology of pancreatic and colorectal cancers through both agonist-independent and -dependent mechanisms.
Because the 3il domain of GPCRs plays a critical role in regulating agonist-dependent receptor activities, including desensitization, internalization, and resensitization, the purpose of this study was to determine the effects of intron 4 retention on these processes. We report that both CCK 2 R splice variants desensitize at a similar rate. However, CCK 2i4sv R resensitizes significantly faster than CCK 2 R. The rapid resensitization rate of CCK 2i4sv R is a consequence of its constitutive, Src-dependent internalization, which creates an intracellular pool of receptors that rapidly recycles back to the plasma membrane. These studies suggest an additional mechanism by which Src can contribute to the cancer phenotype: the rapid resensitization of CCK 2i4sv R to the growth-promoting agonist, G17. * This work was supported by Grants R01 DK48345 and R01 DK58119 from the National Institutes of Health. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. 1

EXPERIMENTAL PROCEDURES
Plasmid Expression Constructs and Transfected Cell Lines-Receptor constructs containing a C-terminal enhanced green fluorescent protein (EGFP), tag, designated CCK 2 R-EGFP and CCK 2i4sv R-EGFP were constructed and transfected into HEK293 cells as described previously (13).
Retroviral Expression Construct-The dominant-negative (dn)Src cDNA was cloned into a bicistronic packaging murine oncoretroviral vector based on pFB (Stratagene) using SalI and EcoRI sites. The vector contains the murine leukemia retrovirus packaging sequence and a multiple cloning site, flanked by the murine leukemia retrovirus long terminal repeat regions. The 5Ј-long terminal repeat functions as a strong promoter upon chromosomal integration of proviral DNA. The pFB plasmid was modified to contain a cassette comprising an ECMV internal ribosome entry site followed by a gene encoding ␤-galactosidase (modified with a nuclear localization signal), which enabled retrovirus titer and transcript expression levels to be determined by staining for ␤-galactosidase (22). Retroviruses were made by simultaneous transfection of HEK293FT cells (Invitrogen) with the dnSrc expression plasmid and plasmids encoding murine leukemia retrovirus gag-pol and vesicular stomatitis virus envelope protein. Cell supernatants were collected and filtered through a 0.45-m filter to remove cell debris prior to infection.
Quantification of Receptor Expression-The level of receptor expression for each clonal line was quantified using 125 I-labeled agonist binding to isolated cell membranes (23). Receptor expression levels by the various clones ranged from ϳ0.43 to 1.8 pmol receptor/mg of membrane protein. The smallest difference in level of receptor expression between EGFP-tagged clones was ϳ2-fold. In all cases, cells transfected with CCK 2 R-EGFP expressed higher levels of receptor than cells transfected with CCK 2i4sv R-EGFP. For the studies presented, we used CCK 2i4sv R-EGFP clonal lines that expressed between 2 and 3.5-fold fewer receptors than the cell line expressing CCK 2 R-EGFP.
Receptor Desensitization and Resensitization-The rates of receptor desensitization and resensitization were compared using agonist-induced increases in [Ca 2ϩ ] i as a measure of receptor activity. The change in the concentration to [Ca 2ϩ ] i was determined using the Ca 2ϩ -binding dye, Fura 2/AM (Molecular Probes, Eugene, OR) as described previously (24). Briefly, cells were cultured on 25-mm glass coverslips in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum, washed with a physiological medium (KRH) containing NaCl (125 mM), KCl (5 mM Laser Scanning Confocal Microscopy (LSCM) and Image Analysis-LSCM was performed on both live and 4% paraformaldehyde-fixed cells. For live cells, the HEK293 cells expressing either CCK 2 R-EGFP or CCK 2i4sv R-EGFP were grown on laminin-coated (15 g/ml) (Trevigen Inc., Gaithersburg, MD) coverslips for 24 h. The cells were then rinsed and placed into a heated stage incubator containing KRH solution (pH 7.4) at 37°C. The EGFP fluorescence distribution was observed using a Zeiss LSM510 META laser scanning confocal microscope (63ϫ objective with oil immersion; excitation 488 nm, emission 505-530 nm). Sequential confocal images were collected at 0.8 m sections throughout the cells. 17-22 optical sections were collected for each cell scanned. After acquiring the initial Z-series scan, the cells were treated with agonist (100 nM), and additional Z-series images were acquired at various time points.
For fixed cells, the cells were plated on laminin-coated glass coverslips, treated as described in the figure legends and fixed in 4% paraformaldehyde for 30 min. The plasma membrane of cells was stained with 2.5 g/ml of 1,1Ј-dioctadecyl-3,3,3Ј,3Ј-tetramethylindocarbocyanine perchlorate (DiI) (Molecular Probes) for 8 min at 37°C. The coverslips were mounted with Vectashield mounting medium on microscope slides, and confocal images were collected as described above. Image processing and colocalization analysis were performed using Metamorph 5.05 software (Universal Imaging, West Chester, PA). The percentage of plasma membrane associated receptor was defined as the percentage of EGFP (green) pixels that colocalized with plasma membrane dye DiI (red) pixels. At least 20 individual cells/condition were analyzed for each experiment, and each experiment was repeated three times.

RESULTS AND DISCUSSION
Both CCK 2 R Variants Exhibit Similar Rates of Desensitization; However, CCK 2i4sv R Resensitizes Faster-Because the 3il domain of GPCRs plays a critical role in receptor desensitization, endocytosis, recycling, and resensitization, it is reasonable to hypothesize that these processes are differentially regulated by the two CCK 2 receptor splice variants. To test this hypothesis, we compared the rates of receptor desensitization and resensitization using agonist-induced increases in [Ca 2ϩ ] i as a measure of receptor activity.
To determine the rates of receptor desensitization, HEK293 cells expressing either recombinant CCK 2i4sv R or CCK 2 R were repeatedly stimulated with G17 (10 nM) over a time course. The initial change in [Ca 2ϩ ] i induced by a 15-s exposure to 10 mM agonist at 22°C was defined as the maximum response. Following this first stimulation, cells were washed with KRH and rechallenged with 10 nM G17 for another 15-s period. This process was repeated 10 times at 2.5-min intervals over a 25-min time course. Both receptors rapidly desensitized with first-order decay kinetics. Analyses of the desensitization curves revealed t 1/2 maximum Ca 2ϩ response values of 2.9 min (95% confidence interval, 2.6 -3.3 min, R 2 ϭ 0.99) and 2.6 min (95% confidence interval, 2.2-3.1 min, R 2 ϭ 0.99) for CCK 2i4sv R and CCK 2 R, respectively ( Fig. 1A). At the end of the time course, treatment with thapsigargin confirmed that the decrease in the G17-induced Ca 2ϩ responses were not because of depletion of the intracellular Ca 2ϩ stores (data not shown).
In contrast to their similar rates of desensitization, the two receptor variants differ significantly in their rates of resensitization. Cells expressing either CCK 2i4sv R or CCK 2 R were first treated with 100 nM G17 and 10 M cycloheximide for 1 h to completely desensitize the receptors and block new protein synthesis, respectively. Treatment with 10 M cycloheximide for up to 3 h did not cause any measurable cell damage. After the cells were washed with KRH supplemented with cycloheximide, they were restimulated at various time points with 10 nM G17 at 22°C, and the changes in [Ca 2ϩ ] i were recorded. A plot of percent maximum Ca 2ϩ response versus time showed that the G17-induced Ca 2ϩ response recovered up to 6.5 times faster in cells expressing CCK 2i4sv R compared with cells expressing CCK 2 R (Fig. 1B). The averaged data from three experiments indicated that CCK 2i4sv R resensitized ϳ5 times faster at 22°C.
The Two Receptor Variants Exhibit Different Subcellular Distributions at Steady State-GPCR desensitization typically involves a rapid and reversible agonist-dependent phosphorylation of specific serine and/or threonine residues within the 3il and C-terminal domains of the receptor by one or more kinases, including second messenger-dependent kinases, such protein kinases A and C, G protein-coupled receptor kinases, and/or casein kinase 1␣ (1, 2). Subsequently, resensitization of the receptor-mediated response involves the internalization of the phosphorylated receptor, dephosphorylation of the intracellular domains, and recycling of the intact receptor back to the plasma membrane (4). To assess the role of receptor trafficking in the rapid resensitization of CCK 2i4sv R, we fused the full-length receptor cDNA to the N terminus of EGFP and examined its subcellular distribution before and after agonist-stimulation using LSCM.
At steady state (before agonist stimulation) the subcellular distributions of CCK 2i4sv R-EGFP and CCK 2 R-EGFP were distinctly different. Cells expressing CCK 2 R-EGFP exhibited predominantly plasma mem-   brane-associated fluorescence ( Fig. 2A), whereas the EGFP fluorescence in cells expressing CCK 2i4sv R-EGFP was located primarily in an intracellular membranous compartment concentrated around, but not in, the nucleus ( Fig. 2A). To quantify the amount of plasma membraneassociated receptor for each variant, we determined the percentage of total cellular EGFP fluorescence that colocalized with plasma membrane dye, DiI, using LSCM. At steady state, only (20 Ϯ 1.7%) of the total cellular EGFP was associated with the plasma membrane in CCK 2i4sv R-EGFP-expressing cells, whereas (80 Ϯ 1.3%) of the fluorescence was plasma membrane-associated in cells expressing CCK 2 R-EGFP (Fig. 2B).
Intracellular Localization of CCK 2i4sv R Is due to Constitutive Receptor Activity-Previously, we and others have shown that CCK 2i4sv R exhibits ligand-independent (constitutive) activation of intracellular signaling pathways (9,13,25). To determine whether the intracellular distribution of CCK 2i4sv R-EGFP was because of its constitutive activity, we treated the cells with the inverse agonist, (R)-1-[2,3-dihydro-1-(2Јmethylphenacyl)-2oxo-5-phenyl-1H-1,4-benzodiazepin-3-yl]-3-(3methylphenyl)urea (YM022). Although initially classified as a CCK 2 Rselective antagonist (26), Beinborn et al. (27) demonstrated the inverse agonist activity of YM022 using an engineered mutant of CCK 2 R that induced increased basal levels of inositol 1,4,5-triphosphate production. YM022 treatment of COS cells expressing the mutated receptor resulted in a decrease in the basal levels of inositol 1,4,5-triphosphate when compared with cells expressing the wild-type receptor (27). Furthermore, Harris et al. (25) showed that YM022 decreased the elevated basal level of phospho-Akt associated with CCK 2i4sv R expression in the esophageal cancer cell line, OE-18. When we treated HEK293 cells expressing CCK 2i4sv R-EGFP with YM022 (100 nM for 2 h), we observed an accumulation of EGFP fluorescence on the plasma membrane ( Fig.  2A). Compared with either untreated or vehicle (0.001% Me 2 SO) treated cells in which only 20% of the total EGFP was associated with the plasma membrane, YM022-treated cells exhibited 77 Ϯ 1.4% of their total cellular EGFP fluorescence on the plasma membrane (Fig. 2B), suggesting that the intracellular location of CCK 2i4sv R is because of constitutive internalization.
Inhibition of Src Kinase Causes the Redistribution of CCK 2i4sv R from the Intracellular Compartment to the Plasma Membrane-CCK 2i4sv R coimmunoprecipitates with activated Src kinase in an agonist-independent manner (13). Because Src regulates GPCR endocytosis (28 -32), we  assessed its role in regulating the subcellular distribution of CCK 2i4sv R. Receptor-expressing cells were infected with a retroviral expression construct containing either empty vector or a kinase-dead Src mutant (A430V) that functions as a dominant-negative (33). In the presence of dnSrc, CCK 2i4sv R was located primarily on the plasma membrane (Fig.  3A). Approximately 74 Ϯ 1.5% of the total cellular EGFP fluorescence was associated with the plasma membrane in cells expressing dnSrc compared with just 19 Ϯ 1.7% in cells expressing a control retroviral vector (Fig. 3A). Coexpression of dnSrc did not affect the distribution pattern of CCK 2 R-EGFP (Fig. 3A). In the presence of dnSrc, 82 Ϯ 1.3% of the EGFP fluorescence was associated with the plasma membrane compared with 80 Ϯ 1.3% in cells transduced with the control retrovirus (Fig.  3B). Together, these data suggest that the constitutive activation of Src kinase by CCK 2i4sv R leads to its agonist-independent internalization.
Expression of dnSrc Does Not Inhibit Agonist-induced Receptor Internalization-A simple model for the constitutive internalization of CCK 2i4sv R is that the additional 69 amino acid residues in the 3il domain cause the receptor to transition into an activated conformation that mimics the conformation of the agonist-bound receptor, activating a common internalization mechanism. Because Src has been shown to regulate agonist-induced receptor internalization through both clathrin-and caveolin-dependent mechanisms (28,29,32,34), we questioned whether dnSrc could also block agonist-induced receptor internalization. To address this question, cells expressing either dnSrc or a control retroviral vector were stimulated with G17 (10 nM), and the subcellular distribution of the EGFP-tagged receptors was monitored in live cells over a 40-min time course. Treatment with G17 induced a time-dependent redistribution of the EGFP-tagged receptors from the plasma membrane to intracellular membrane vesicles, which eventually concentrated in a perinuclear compartment by 40 min (Fig. 4). Thus, unlike constitutive CCK 2i4sv R internalization, expression of dnSrc did not inhibit the agonist-induced internalization of either CCK 2i4sv R or CCK 2 R (Fig. 4, dnSrc), suggesting that different mechanisms mediate the agonist-dependent and -independent receptor internalization.
Expression of dnSrc Does Not Affect the Rates of Receptor Desensitization but Significantly Slows the Rate of CCK 2i4sv R Resensitization-Consistent with the hypothesis that agonist-induced and constitutive receptor internalization involves distinct mechanisms, expression of dnSrc had no effect on the rates of G17-induced desensitization of either receptor (Fig. 5A). Similar to the data shown in Fig. 1A, in the presence of dnSrc, both receptors rapidly desensitize in accordance with firstorder decay kinetics. Analyses of the desensitization curves showed t 1/2 maximum Ca 2ϩ response values of 2.7 min (95% confidence interval, 2.3-3.2 min, R 2 ϭ 0.99) and 2.9 min (95% confidence interval, 2.6 -3.2 min, R 2 ϭ 0.99) for CCK 2i4sv R and CCK 2 R, respectively (Fig. 5A). In contrast to desensitization, expression of dnSrc significantly slowed the rate of CCK 2i4sv R resensitization (Fig. 5B). A plot of percent maximum Ca 2ϩ response versus time showed that the G17-induced Ca 2ϩ response in cells expressing CCK 2i4sv R and dnSrc recovered significantly slower than cells expressing CCK 2i4sv R and control retroviral vector (Fig. 5B). In fact, the rate of resensitization of CCK 2i4sv R in the presence of dnSrc was the same as rates of resensitization of CCK 2 R-expressing cells with or without dnSrc expression (Fig. 5B). . Model of CCK 2i4sv R internalization and rapid resensitization. CCK 2i4sv R is internalized via two distinct pathways: 1) agonist-independent (constitutive), which requires Src, and 2) agonist-dependent, which does not require Src. The agonist-independent (constitutive) Srcdependent internalization creates an intracellular pool of ready receptors that recycles back to the plasma membrane at a rate similar to the rate of CCK 2i4sv R resensitization. The agonist-dependent pathway is shared by both CCK 2 R splice variants.
The Rate of CCK 2i4sv R Recycling Is Similar to the Rate of Its Resensitization-Because dnSrc expression caused a steady-state redistribution of CCK 2i4sv R from its intracellular location to the plasma membrane and slowed its rate of resensitization, we hypothesized that intracellular CCK 2i4sv R serves as a recycling receptor pool for the rapid resensitization of the agonist-induced Ca 2ϩ response. To test this hypothesis, we determined the recycling rate of CCK 2i4sv R by using the inverse agonist, YM022. Because an inverse agonist "locks" its cognate receptor in an inactive conformation, we reasoned that treatment with YM022 would block the constitutive reinternalization of the recycled receptor via the Src-dependent pathway, resulting in a time-dependent accumulation of receptor on the plasma membrane. The rate at which EGFP-tagged receptor translocates from the intracellular compartment to the plasma membrane would therefore reflect the rate at which the receptor recycles from the intracellular compartment.
Cells expressing CCK 2i4sv R were treated with YM022 and cycloheximide over a time course. The redistribution of intracellular EGFPtagged CCK 2i4sv R to the plasma membrane was monitored by LSCM. Consistent with our hypothesis, YM022 treatment caused a time-dependent redistribution of CCK 2i4sv R from its intracellular compartment to the plasma membrane (Fig. 6). The rate at which EGFP-tagged CCK 2i4sv R accumulated on the plasma membrane was found to be similar to the rate of receptor resensitization (Figs. 1B and 5B), suggesting that recycling of receptor from the intracellular pool is responsible for the rapid recovery of the agonist-induced Ca 2ϩ response. Interestingly, the CCK 2 R antagonist, JB93182, did not affect CCK 2i4sv R subcellular distribution, confirming that YM022 acts an inverse agonist and suggesting that the CCK 2i4sv R conformation and mechanism involved in its constitutive internalization are distinct from those involved in agonistinduced internalization.
Model for the Rapid Resensitization of CCK 2i4sv R-In comparing the desensitization, internalization, and resensitization of the CCK 2 receptor splice variants, we have shown that intron 4 retention and the resulting addition of 69 amino acid residues to the 3il domain of CCK 2i4sv R affects both its subcellular distribution and rate of agonist-dependent resensitization. Based on the data presented, we developed the model shown in Fig. 7. CCK 2i4sv R is internalized through two mechanisms. The first involves constitutive receptor internalization through a Src kinasedependent pathway (Fig. 7). Internalization of CCK 2i4sv R through the agonist-independent pathway is blocked both by the inverse agonist, YM022, and expression of dnSrc. The second mechanism of receptor internalization, induced by agonist binding, is common to both receptor variants and does not require Src kinase activity. The rapid resensitization of CCK 2i4sv R, in this two-compartment model, results from the recycling of constitutively internalized receptor back to the plasma membrane in a process that is driven largely by the amount of receptor in the intracellular pool (4). As predicted by this model, decreasing the amount of receptor in the intracellular recycling compartment by inhibiting Src-mediated receptor internalization slowed the rate of CCK 2i4sv R resensitization.
Biological Significance-The rapid resensitization of CCK 2i4sv R may contribute to the potential role of the receptor in regulating the progression and/or spread of colorectal and pancreatic cancers, where it is ectopically expressed. It is well established that gastrin promotes tumor cell proliferation, motility, and invasion of these cancers (18 -21). Because many colorectal cancer cells also express gastrin, and experi-mental evidence suggests that the peptide hormone acts in an autocrine and/or paracrine manner to stimulate tumor growth, the hypothesis that rapid resensitization of CCK 2i4sv R contributes to the agonist-mediated tumor promoting properties will be the focus of future studies.