CCR2 splice variant expression patterns
The ORFs for most chemokine receptor genes are located within a single exon; however, some chemokine receptor genes feature alternative splicing that combines the ORF across two exons, resulting in the expression of multiple isoforms. CCR2 has been cloned as two isoforms: CCR2A and CCR2B. The ORF for CCR2B is located within a single exon, similar to the gene structure for most chemokine receptors, whereas the ORF for CCR2A is separated across two exons, joined through alternative splicing, resulting in two CCR2 isoforms that feature different C-terminal intracellular regions (Fig. 1A). To explore the expression pattern of these two CCR2 isoforms, reverse transcriptase–polymerase chain reaction (RT-PCR) was performed using isoform-specific primer pairs in several cell lines. THP-1 cells, a human acute monocytic leukemia cell line, expressed both isoforms, with CCR2B mRNA detected at higher levels than CCR2A mRNA, which is similar to previous reports [14, 16]. All anchored cell lines tested here only expressed CCR2A mRNA (Fig. 1B), suggesting that the splicing machinery for this gene is highly active in these cells. Both isoforms resulted in identical amino acid sequences from the N-terminus to the seventh transmembrane domain, with the only differences observed in the C-terminal intracellular tail region.
Interaction with β-arrestins internalized CCR2B but not CCR2A
In addition to G protein activation, most chemokine receptors interact with β-arrestins, which induce internalization to downregulate their chemokine signaling and promote β-arrestin-mediated signaling. Using a structural complementation analysis based on NanoBiT technology, the interactions between β-arrestins and CCR2A or CCR2B were investigated. Multiple combinations of NanoBiT constructs featuring the receptor isoforms and β-arrestins were expressed in HEK293 cells treated with MCP-1, and luciferase activity was measured using Nano-Glo Live Cell Reagent and a luminometer. Cells expressing any NanoBiT construct combination of CCR2A and β-arrestin showed no increase in luciferase activity following MCP-1 treatment (Fig. 2A), whereas MCP-1 treatment induced significant luciferase activation for the combination of CCR2B-LgBiT and SmBiT-β-arrestin 1/2 (Fig. 2B). Maximum luciferase activity was observed for the interaction between CCR2B and β-arrestin 1, which was approximately twice as high as that observed for β-arrestin 2, implying that CCR2B has a higher affinity for β-arrestin 1.
Many GPCRs require GPCR kinase (GRK)-dependent Ser/Thr phosphorylation of the third intracellular loop and the C-terminal region to interact with β-arrestin. Of the ten Ser/Thr residues identified in the C-terminal region of CCR2B, some, especially in Ser/Thr-rich sequence, may potentially act as target sites for GRK phosphorylation, whereas none of the five Ser/Thr residues in the C-terminal region of CCR2A are likely to be phosphorylated by GRKs (Fig. 2C). Although a few common Ser/Thr residues were identified in the third intracellular loop between these two variants, these may be neither phosphorylated by GRKs nor sufficient for the phosphorylation-dependent recruitment of β-arrestin.
CCR2A is 14 amino acids longer in the C-terminal region than CCR2B, which might contribute to differences in their biological roles (Fig. 2C). The interaction with β-arrestins facilitates the internalization of GPCRs [17]. To examine the internalization behavior of the CCR2 variants, SmBiT-Clathrin and b-arrestin1-LgBiT were co-expressed with CCR2 variants in HEK293 cells, and MCP-1-stimulated luciferase activity was measured (Fig. 2D). Furthermore, NanoBiT constructs of the receptors and the FYVE domain from early endosome antigen 1 (EEA1), an early endosomal marker, were expressed in HEK293 cells, and MCP-1-stimulated luciferase activity was measured (Fig. 2E). Both data showed that MCP-1-dependent luciferase activity was increased in CCR2B-expressing cells but not in CCR2A-expressing cells, suggesting that only CCR2B is internalized to endosome in a β-arrestin-dependent manner.
Both CCR2 variants mediate the MCP-1-stimulated interaction between Gβ1 and GRKs
Following ligand binding, CCR2 interacts with and activates heterotrimeric G proteins, resulting in the dissociation of Gα and Gβγ subunits, which activate distinct signaling pathways. However, the interaction between the CCR2 isoforms and either Gα or Gβγ could not be confirmed using the NanoBiT assay, despite the examination of all possible combinations (Fig. 3A, and data not shown). β-arrestins are recruited to CCR2B through interactions with GRK-phosphorylated regions. Seven GRK isoforms have been identified and can be divided into three subfamilies based on their functions, structures, and expression patterns [18]. GRK2/3 and GRK5/6 are ubiquitously expressed in mammalian tissues and, therefore, are considered likely to regulate the activity of most GPCRs [19]. To confirm involvement of GRK2/3 in β-arrestin recruitment to CCR2B as previously described [17, 20], cells expressing NanoBiT constructs for CCR2B and β-arrestin 1 were pretreated with a GRK2/3-specific kinase inhibitor, Cmpd101, and MCP-1-dependent luciferase activity was measured. As shown in Fig. 3B, the luciferase activity stimulated by MCP-1 declined in a Cmpd101 dose–dependent manner. However, treatment with 50 µM Cmpd101, which was sufficient to complete inhibition of GRK2/3 activity [21], did not inhibit all luciferase activity, suggesting that other GRK subfamilies may contribute to CCR2B phosphorylation.
Gβγ separated from the GTP form of Gα may binds to GRKs to make it easier to recognize and phosphorylate ligand bound GPCRs [22]. To examine the interaction between Gβ1 and GRKs, a real-time luciferase assay was performed in HEK293 cells expressing NanoBiT constructs for both proteins. In the presence of CCR2A and CCR2B, MCP-1 stimulated luciferase activities through binding of SmBiT-Gβ1 with either of GRK2-LgBiT or GRK5-LgBiT (Fig. 3C), suggesting that the interaction between GRK2 and Gβ1 differs from that of GRK5 and Gβ1. These findings suggest that MCP-1 binding to both CCR2A and CCR2B triggers the dissociation of Gα and Gβγ, resulting in the subsequent interaction between Gβ and GRKs.
To investigate whether the chemokine receptors interact with GRKs, cells expressing different combinations of NanoBiT constructs were treated with MCP-1 and subjected to a real-time luciferase assay. The ligand-dependent luminescence increase was observed in cells expressing CCR2B-LgBiT and GRK2-SmBiT, suggesting that GRK2 may interact with CCR2B through Gβγ. However, no signal change was observed for any combination of CCR2B and GRK5, suggesting that this assay system may not be suitable for examining the interaction between CCR2B and GRK5. No ligand-stimulated luciferase activity was observed for any combination between CCR2A and either GRK2 or GRK5 (Fig. 3D, Additional file 1: Fig. S1), providing additional evidence that the C-terminal region of CCR2A may not interact with GRKs or support GRK-mediated phosphorylation.
The C-terminal tail region may play an essential role in the membrane localization of CCR2
The detection of endogenous CCR2 proteins using biochemical methods can be difficult because of protein amount, difficulty of 7-TM protein preparation, and lack of appropriate antibodies for detection; therefore, plasmids expressing epitope-tagged CCR2 genes were generated. CCR2A and CCR2B were expressed in HEK293 cells with N-terminal FLAG-tags or C-terminal HA-tags, and cell lysates were subjected to western blot analysis using appropriate antibodies. As shown in Fig. 4A, regardless of the epitope position, both isoforms were expressed at similar levels. However, confocal images of C-terminal GFP-tagged receptors demonstrated that CCR2A was primarily localized to the cytosol, whereas CCR2B signals were detected in the plasma membrane (Fig. 4B), suggesting that the C-terminal region determines the membrane localization of chemokine receptors. The membrane localization of chemokine receptors was further determined by the HiBiT assay. Luminescence signals in cells expressing N-terminal SmBiT-tagged receptors became stronger depending on the transfected plasmid amount. Intriguingly, SmBiT-CCR2B signals were much stronger than those for SmBiT-CCR2A, which was consistent with the plasma membrane localization patterns observed using GFP-tagged receptors, as shown in Fig. 4B. Comparing the sequences of the C-terminal region revealed that fewer basic amino acids were located in the membrane-proximal region of CCR2A than in a similar region of CCR2B (Fig. 4C). Other chemokine receptors (e.g., CCR1) also contain a relatively higher number of basic amino acids in similar regions. To further explore the specific roles played by basic amino acids in determining the membrane localization of the receptor, an SmBiT-fusion CCR2 chimera was generated, in which the C-terminal region was replaced with the C-terminal region from CCR1 (CCR2C1). In the HiBiT assay, luciferase activity induced by the chimera was similar to that induced by SmBiT-CCR2B (Fig. 4D). Furthermore, both CCR2B and CCR2C1 recruited β-arrestin 1 following MCP-1 stimulation with similar efficiencies (Fig. 4E). These results suggest that the C-terminal region may play a pivotal role in determining the plasma membrane localization of chemokine receptors and β-arrestin recruitment and that the C-terminal CCR2A region may be insufficient for membrane localization. Many GPCRs are expressed at the cell surface as homodimers or heterodimers. In previous studies using NanoBiT technology, homodimerization but not heterodimerization was observed for CXCR4 and CXCR7. However, the NanoBiT assay performed using cells expressing CCR2A and CCR2B, as SmBiT or LgBiT constructs, showed no increases in luminescence signals, implying that CCR2 isoforms are not able to form homodimeric receptor complexes (Fig. 4F).
Comparison of CCR2 splice variant-mediated G protein signaling
To explore the cellular responses mediated by CCR2 splice variants, downstream signaling events were evaluated. Recently, engineered G proteins, called mini-G proteins (mG), were developed to study the biophysical interactions between active GPCRs and each subtype of G proteins [23, 24]. We constructed mG proteins in NanoBiT vectors, as described in previous reports. The analysis of different combinations of LgBiT- and SmBiT-tagged proteins revealed that N-terminal-tagged LgBiT constructs for each mG protein are likely to bind with CCR2-SmBiT constructs (data not shown). In general, chemokine receptors activate Gαi/o, Gα16, or Gα12/13. MCP-1 induced luminescence in cells expressing mGsi and each of the CCR2 variants, particularly CCR2B (Fig. 5A). Unfortunately, luminescence signals in the presence of either mG16 or mG12 did not increase with MCP-1 stimulation (data not shown), indicating that these mini-G proteins may not provide an adequate structure to interact with activated receptors, as shown in previous reports [23, 24]. The luminescence signals were increased by interaction between mGsq and both CCR2 variants in the presence of MCP-1, which confirms CCR2-mediated activation of both Gαi and Gαq pathways [25, 26].
Extracellular signal-regulated kinase (ERK) phosphorylation is a representative event in GPCR-mediated signaling pathways [27]. MCP-1-stimulated ERK phosphorylation was detected in HEK293 cells without the expression of exogenous CCR2 variants, and according to RT-PCR analysis, CCR2A is endogenously expressed in HEK293 cells (Fig. 1B). To determine the effects of each CCR2 variant on MCP-1-stimulated ERK phosphorylation, CCR2 knockout (KO) cells were developed using the CRISPR-Cas9 system with subsequent cloning and genomic DNA analysis. ERK was weakly phosphorylated following MCP-1 stimulation in CCR2 KO cells compared to parental cells (Fig. 5B). Although CCR2 is the only GPCR known to bind MCP-1, MCP-1 has been reported to interact with glycosaminoglycans (GAGs), a group of cell surface proteins [23, 24] which may be responsible for the MCP-1-stimulated ERK phosphorylation observed in the absence of CCR2. MCP-1-stimulated ERK phosphorylation was enhanced by the expression of exogenous CCR2A and CCR2B in CCR2 KO cells, suggesting that both variants have equal impacts on ERK phosphorylation. The observed time-dependent phosphorylation pattern was similar in cells expressing wild-type CCR2, CCR2 KO cells, and cells expressing exogenous CCR2A or CCR2B, appearing at 5 min, declining at 15 min, and disappearing at 30 min.
MCP-1 appears to activate Gαq as well as Gαi through CCR2 variants (Fig. 5A), and an increase in intracellular calcium was determined by NanoBiT technology using LgBiT-MYLK2s (calmodulin-binding motif in myosin light-chain kinase 2) and calmodulin-SmBiT, which were previously developed to measure real-time intracellular calcium change in our laboratory [28]. MCP-1-stimulated luminescence increased slightly in cells expressing CCR2A and CCR2B, with a stronger signal in cells expressing CCR2A than in cells expressing CCR2B (Fig. 5C). Since chemokine receptors mainly activate the Gαi/o pathway often activate calcium signaling, cells expressing the chimeric G protein Gαqi which converts Gαi to Gαq pathway were subjected to the NanoBiT calcium assay to determine activation of the pathway. The luminescence induced by MCP-1 treatment was much higher than that observed in the absence of Gαqi (Fig. 5D), and the CCR2A-mediated signals were stronger than those mediated by CCR2B, which increased in an MCP-1 dose-dependent manner (half-maximal excitatory concentration: 4.24 × 10−9 for CCR2A vs. 1.01 × 10−8 for CCR2B; Fig. 5E, Additional file 1: Fig. S2). Downstream signals of ERK phosphorylation and calcium upregulation can stimulate transcription factor binding to the serum response element (SRE); therefore, SRE-reporter gene assay was performed in cells expressing CCR2A and CCR2B. Figure 5F shows that CCR2A mediated stronger luciferase activity than CCR2B, which was consistent with the results of the calcium assay. These results indicate that CCR2A mediates MCP-1-stimulated G protein signaling with relatively high potency compared with CCR2B.
CCR2A is essential for cancer cell proliferation and migration
To explore the functional effects of CCR2 signaling on cellular behavior, CCR2 KO HeLa cells were generated using the CRISPR-Cas9 system, and their growth was compared with that of parent cells. Under normal culture conditions containing 10% fetal bovine serum (FBS), the growth rate of the KO cells was slightly slower than that of the parent cells (Fig. 6A, upper graph). To examine the effects of MCP-1 treatment on cell growth, the cells were treated with 100 ng/ml MCP-1 in the presence of 1% FBS because HeLa cells die easily in the absence of serum. CCR2 KO cells grew slowly compared with their parent cells, regardless of MCP-1 treatment (Fig. 6A lower graph). RT-PCR using specific primer pairs revealed similar levels of endogenous MCP-1 expression in both cell groups (Fig. 6B), suggesting that the lack of response to exogenous MCP-1 application might be due to the strong endogenous expression of MCP-1 in HeLa cells. In a migration assay, CCR2 KO HeLa cells did not migrate to the lower chamber containing MCP-1, suggesting that the chemotactic activity of HeLa cells toward MCP-1 was impaired in the absence of CCR2. The spontaneous motility of CCR2 KO cells was also decreased significantly in media containing 1% FBS (Fig. 6C).