Address correspondence and reprint requests to Dr Jane L. Ko, Department of Biological Sciences, Seton Hall University, 400 South Orange Ave., South Orange, NJ 07079, USA. E-mail: email@example.com
Poly C binding protein 1 (PCBP1) is an expressional regulator of the mu-opioid receptor (MOR) gene. We hypothesized the existence of a PCBP1 co-regulator modifying human MOR gene expression by protein–protein interaction with PCBP1. A human brain cDNA library was screened using the two-hybrid system with PCBP1 as the bait. Receptor for activated protein kinase C (RACK1) protein, containing seven WD domains, was identified. PCBP1-RACK1 interaction was confirmed via in vivo validation using the two-hybrid system, and by co-immunoprecipitation with anti-PCBP1 antibody and human neuronal NMB cell lysate, endogenously expressing PCBP1 and RACK1. Further co-immunoprecipitation suggested that RACK1-PCBP1 interaction occurred in cytosol alone. Single and serial WD domain deletion analyses demonstrated that WD7 of RACK1 is the key domain interacting with PCBP1. RACK1 over-expression resulted in a dose-dependent decrease of MOR promoter activity using p357 plasmid containing human MOR promoter and luciferase reporter gene. Knock-down analysis showed that RACK1 siRNA decreased the endogenous RACK1 mRNA level in NMB, and elevated MOR mRNA level as indicated by RT-PCR. Likewise, a decrease of RACK1 resulted in an increase of MOR proteins, verified by 3H-diprenorphine binding assay. Collectively, this study reports a novel role of RACK1, physically interacting with PCBP1 and participating in the regulation of human MOR gene expression in neuronal NMB cells.
PCBP1, the mu-opioid receptor (MOR) gene regulator, belongs to the K homology (KH) domain superfamily, which is consisting of two subsets in mammalian cells: hnRNPs K/J and the αCP proteins (Makeyev and Liebhaber 2002). The αCP proteins are also known as PCBPs, including PCBP1, 2, 3, and 4, with additional isoforms generated via alternative splicing. Within the KH domain superfamily, hnRNPK is considered the prototype and well characterized. PCBP1 and PCBP2 are the major forms of PCBPs expressing in mammalian cells, while the expressions of PCBP3 and PCBP4 are comparatively less (Perera et al. 2007).
PCBP proteins are known to be involved in various biological processes, including but not limited to, mRNA stabilization, transcriptional and translational regulation (Kiledjian et al. 1997; Thakur et al. 2003; Malik et al. 2006; Rivera-Gines et al. 2006). For example, PCBP2 has been shown to regulate viral genome replication and translation (Walter et al. 2002; Zhang et al. 2007). PCBP3 can function as a repressor (Kang et al. 2012). PCBP4, also known as MCG10, can suppress cell proliferation by inducing apoptosis and cell cycle arrest (Zhu and Chen 2000). In addition, studies have shown that PCBP1 participates in regulations of the androgen receptor gene (Cloke et al. 2010), eIF4E gene (Meng et al. 2007), and the MOR gene (Ko and Loh 2005). The relationship between PCBP1 and MOR gene was originally discovered in a yeast one-hybrid study, in which PCBP1 was identified as a mouse MOR gene regulator by screening a mouse brain cDNA library with the mouse MOR single-stranded (ss) cis-element as the target (Ko and Loh 2005). Naturally, various cis-elements and factors are also involved in the MOR gene regulation (Choe et al. 1998; Ko et al. 1998, 2003; Ko and Loh 2001; Choi et al. 2005; Kim et al. 2005). The PCBP1 not only participates in mouse MOR gene regulation but also involves in human MOR gene in a positive manner, reflecting a high homology of the MOR ssDNA element between mouse and human, and the PCBP1 protein is conserved among mammals (Cook et al. 2010). In addition, PCBP1 not only can act as a positive regulator but also as a negative regulator (Zhang et al. 2010). Taken together, these data suggest that PCBP1 possesses divergent roles in regulating gene expressions.
The different functional roles of a protein can be affected by its microenvironment in a context-dependent manner, which may be mediated through variable protein–protein interactions. Therefore, we tested a hypothesis of the presence of a PCBP1-interacting protein, which can impact human MOR gene regulation via its physical interaction with PCBP1. In this study, a bacteria two-hybrid screening study was carried out using PCBP1 as the bait protein. We identified one of the PCBP1-interacting proteins as RACK1, a known tryptophan-aspartic acid (WD) containing protein (McCahill et al. 2002; Tcherkasowa et al. 2002; Imai et al. 2009). RACK1-PCBP1 interaction and the interaction domain as well as the effects of RACK1, via over-expression and siRNA knock-down, on human MOR gene expression were investigated using the human neuronal NMB cell model system.
Materials and methods
Human neuroblastoma NMB cells (Baumhaker et al. 1994), grown in RPMI medium (Invitrogen, Carlsbad, CA, USA) supplemented with 10% fetal calf serum (Invitrogen), were incubated at 37°C with 5% CO2.
Bacteria two-hybrid system
The pTRG plasmid, containing a human brain cDNA library (Stratagene, La Jolla, CA, USA), RACK1 or truncated RACK1, was co-transformed with the plasmid pBT-PCBP1, according to manufacturer's protocol (Stratagene). The resulting transformants were grown on selective media containing chloramphenicol, tetracycline, 3-amino-1,2,4-triazole (3-AT), and streptomycin, based on the manufacture's protocol. The surviving clones were further verified using X-gal screening.
Cells were harvested and then lysed with the lysis buffer, containing 1% sodium dodecyl sulfate (SDS), phenylmethylsulfonyl fluoride (PMSF) 1 mM, pepstatin 10 μg/mL, aprotinin 1 μg/mL, and sodium vanadate at 1 μM, along with sonication. Cell mixtures were microcentrifuged at the top speed at 4°C. The supernatant was collected and subjected to Lowry assay for determining the protein concentration. The nuclear extract was prepared as described previously (Lin et al. 2008). For co-immunoprecipitation, the same amount of proteins from each lysate or nuclear extract was used. First, the lysate or extract was pre-cleaned with IgG beads. The pre-cleaned lysates or extracts were then incubated with anti-PCBP antibody or non-specific IgG antibody (as the negative control) at 4°C. The IgG sepharose beads were then added to the mixtures, which were then incubated for 2 h on a rotatory platform at 4°C. The beads were then washed three times with washing buffer [containing 1% NP-40 and 0.5% sodium deoxycholate in phosphate buffered saline (PBS)]. The bound proteins were eluted by adding the polyacrylamide gel electrophoresis (SDS-PAGE) treatment buffer.
SDS-PAGE and western blot analysis
Samples were subjected to a 12% SDS-PAGE and then transferred to the polyvinylidene difluoride membrane (GE healthcare, Waukesha, WI, USA). The membrane was then incubated with the milk containing blocking solution and sequentially washed with 0.1% and 0.3% TTBS (containing 0.1% or 0.3% Tween-20 in Tris-buffered saline). The membrane was probed with the anti-RACK1 or anti-PCBP1 antibody, as well as anti-β-actin or anti-β-tubulin antibodies (Santa Cruz biotechnology, Santa Cruz, CA, USA). Signals were detected using an enhanced ECL detection system (GE healthcare) with a Storm phosphoimager (Molecular Dynamics, Sunnyvale, CA, USA).
Cell fixation and immunofluorescence staining
Cells, fixed with 4% formaldehyde, were permeabilized using Triton-X 100 (Sigma, St. Louis, MO, USA). Cells were then incubated with the blocking solution containing 2% bovine serum albumin (BSA) and 0.3% Triton-X 100 in PBS. Next, cells were incubated with anti-RACK1 antibody (Santa Cruz) at 4°C, and further incubated with a Cy3-conjugated anti-IgG secondary antibody (Jackson Immunoresearch, West Grove, PA, USA). These cells were counter-stained with 4′,6-diamidino-2-phenylindole (DAPI) (Invitrogen), and the images were viewed using Fluoview 1000 confocal microscope (Olympus, Center Valley, PA, USA).
Generations of RACK1 deletion constructs
Various lengths of RACK1 fragments were generated by PCR using a specific pair of primers for the intended WD regions (McCahill et al. 2002). PCR products were cloned into pCR2.1 vector (Invitrogen) and then subjected to DNA sequencing. PCR fragments with correct sequences were then subcloned into pTRG vector (Agilent, Santa Clara, CA). The resulting plasmids were then verified using the restriction enzyme digestion as well as DNA sequencing. Primers used were as follows: WD1 5′-ACTGAGCAGATGACCCTT-3′; WD2-7, 5′-GAGACCAACTATGGA-3′; WD3-7, 5′-ACCACCACGGGCGA-3′; WD4-7, 5′-GTGCAAATACACTGT-3′; WD5-7, 5′-TGCAAGCTGAAGACC-3′; WD6-7 5′-AAACACCTTTACACGCTA-3′; WD7, 5′-AAGCAAGAAG-3′; WD1R, 5′-ACCTTTGACTGGTCCCTAATC-3′; WD2R, 5′-GCGGAGACCCTAGAGTGTATC; WD4R, 5′-CATACCTTGGACCGAATC-3′; WD5R, 5′-CGGTACAATACCCTAGAGATC; WD6R, 5′-TTCATCTACAATGATCTTTCCCTCATC-3′; WD7R, 5′-TGGTAACCGTGTGCGATCCATAGTTTATC-3′.
RNA isolation and RT-PCR
Total RNA from cells was extracted using Tri-Reagent (Molecular Research Center, Cincinnati, OH, USA). First strand cDNA was synthesized using random primers with the reverse transcriptase (Invitrogen) at 37°C for 50 min. PCR amplification was performed at 95°C for 1 min, 68°C for 35 s, and 72°C for 35 s using a specific pair of primers. Primer sets for human MOR and β-actin primers were described previously (Cook et al. 2010). RACK1 primers are as follows: 5′-AAACACCTTTACACGCTA-3′; 5′-TGGTAACCGTGTGCGATCCATAGTTTATC-3′. The PCR products were separated using 2% agarose gel, and then were quantified using Image Quant software (Molecular Dynamics, Sunnyvale, CA, USA).
Cells, seeded in the six-well plates, were transfected with RACK1 siRNA (Invitrogen) or the scrambled siRNA (Invitrogen) by lipofectamine RNAimax (Invitrogen) according to the manufacture's protocol. Forty-eight hours or seventy-two hours after transfection (as indicated in each experiment), cells were harvested.
Transient transfection and luciferase reporter assay
Cells were transfected using lipofectamine (Invitrogen), according to manufacturer's protocol, with the pGL3-basic luciferase reporter containing hMOR promoter (p357 plasmid) along with the pcDNA3-RACK1 expression plasmid and an internal control, pCH110 plasmid containing the β-galactosidase gene. Forty-eight hours after transfection, cells were harvested and lysed using the reporter lysis buffer (Promega, Madison, WI, USA). Cell lysates were subjected to luciferase assay using a Luciferase Assay system (Promega), and the activity was quantified as relative light units (RLU) with a Berthold Luminometer (Perkin Elmer, Waltham, MA, USA).
Competitive opioid binding assay
Cells were incubated with 2 nM [3H]-diprenorphine (Perkin Elmer) in the presence or absence of 1 μM unlabeled mu-selective antagonist, CATP (Sigma), in 25 mM HEPES, pH 7.4, for 10 min at 22–25°C. To block the delta and kappa receptor bindings, 1 μM of DADLE [D-Ala2, D-Leu5]-enkephalin and U50488 (Sigma) were also included in all reactions. The reaction was terminated by filtration on GF/B filters and washed with 7% PEG in HEPES buffer. The filters were then incubated with Scintiverse-BD (Fisher Scientific, Waltham, MA, USA) for 1 h at 22–25°C before counting in a beta-counter (Perkin Elmer). Data were normalized by the amount of protein present in control and transfected cells.
Values are reported as mean ± SE. Statistical significance was determined using the student's t-test.
Identification of RACK1 as a PCBP1-interacting protein via the two-hybrid screening
To identify PCBP1 interacting proteins, a human cDNA library was screened using bacteria two-hybrid system with PCBP1 as a bait protein, which was cloned into pBT vector (pBT-PCBP1). One of the positive clones survived under the highest stringent selections, containing 3-AT, streptomycin, chloramphenicol, and tetracycline as well as with the X-gal color screening. This clone contained approximately 1 Kb cDNA insert, encoding a full length of RACK1 sequence as determined by DNA sequencing and NCBI blast analysis (accession number NM 006098). To validate RACK1 as a PCBP1 interacting protein, the individual plasmid containing RACK1 (pTRG-RACK1) was co-transformed with PCBP1 containing plasmid (pBT-PCBP1) into bacteria, and the survival clones under the highest stringent selection were reproduced and verified. These results suggested that RACK1 interacts with PCBP1 directly.
Demonstration of the physical interaction of RACK1 and PCBP1 using co-immunoprecipitation
To determine whether the direct interaction between RACK1 and PCBP1 is not only detectable using the two-hybrid system but also occurs in the mammalian system, the co-immunoprecipitation assay was performed using the lysates from human neuronal NMB cells, endogenously expressing PCBP1 (Lin et al. 2008).
First, to determine if the endogenous RACK1 is present in NMB cells, western blot analysis using anti-RACK1 antibody was performed with NMB cell lysate. As shown in Fig. 1a, lane 2, the result demonstrated that RACK1 protein with approximately 32 kDa M.W. was observed (indicated by the arrow). This same blot was first probed with anti-PCBP1 antibody (lane 1) to confirm the presence of endogenous expression of PCBP1, which corroborated with our previous data (Lin et al. 2008).
The cellular distribution of RACK1 in NMB neuronal cells was then determined using confocal microscopy with anti-RACK1 antibody as the primary antibody and Cy3-conjugated anti-IgG antibody as the secondary antibody. As shown in Fig. 1b, RACK1 was mainly located in the cytosol with less in the nucleus. The nuclear DNA was counter-stained using DAPI. This result was further determined by western blot analysis with anti-RACK1 antibody (Fig. 1c) using the same amount of proteins from either whole cell lysates or nuclear extracts. The quantitative data are shown in Fig. 1d with the signal of RACK1 in whole cell lysates is arbitrarily defined as 100%. Taken together, these results demonstrated that RACK1 is found in both cytosol and nucleus, with the major distribution of RACK1 in the cytosol of NMB cells.
Next, the in vivo physical interaction between RACK1 and PCBP1 was investigated using co-immunoprecipitation assay with anti-PCBP1 antibody and NMB cell lysates to isolate PBCP1, and the presence of RACK1 as the poly C binding protein 1 (PCBP1) co-immunoprecipitant was tested via Western blot analysis using anti-RACK1 antibody. As shown in Fig. 2a, PCBP1 was successfully pulled down by anti-PCBP1 antibody (lane 2; indicated by an arrow), whereas the PCBP1 band was not observed using a non-specific antibody (indicated as control in lane 1). In lane 3, input lane was loaded with cell lysates only. The same blot, without stripping the PCBP1 signal, was then probed with the anti-RACK1 antibody. As shown in Fig. 2b, the presence of RACK1 was also observed in lane 2 containing PCBP1, demonstrating that RACK1 was co-immunoprecipitated with PCBP1, whereas RACK1 was not observed in control (lane 1). These results demonstrated the direct physical interaction of RACK1 and PCBP1 in human neuronal cells.
Cellular location for RACK1-PCBP1 interaction
As RACK1 and PCBP1 are present in both cytosol and nucleus, we next examined if the RACK1-PCBP1 interaction takes place in the cytosol or nucleus, or both locations. The co-immunoprecipitation was further carried out using anti-PCBP1 antibody to pull-down PCBP1, or a non-specific antibody as control with NMB nuclear extracts. Then, RACK1 as the co-immunoprecipitant was examined using western blot analysis with anti-RACK1 antibody.
As shown in Fig. 2c, the anti-PCBP1 antibody successfully pulled down PCBP1 (indicated by an arrow) using either nuclear extract (indicated as NE) or whole cell lysates (indicated as cell lysates), whereas no PCBP1 was detected in the non-specific antibody lane (control Ab). Importantly, the RACK1 signal (indicated by the arrow) was only found in the cell lysate, but not in the nuclear extract lane (NE). This result therefore suggested that the physical interaction of RACK1-PCBP1 took place in the cytosol, but not in the nucleus of NMB cell.
Mapping the domain of RACK1 interacting with PCBP1
RACK1 contains seven WD domains (as illustrated in Fig. 3) (Tcherkasowa et al. 2002), which may play a significant role in protein–protein interaction (McCahill et al. 2002; Tcherkasowa et al. 2002; Imai et al. 2009). Therefore, to determine which domain of RACK1 is responsible for the interaction with PCBP1, the serial WD domain deletion analysis was performed (Fig. 3a), starting from the first WD domain located at the N terminus (WD1 domain) of RACK1. The WD domain deletion construct was generated by PCR with the sense primer containing the start codon (ATG) and the antisense primer containing a stop codon. Each PCR product was cloned into pCR2.1 vector, and was verified by DNA sequencing. The insert with the correct sequence was then subcloned into pTRG vector. Six serial WD domain deletion constructs were generated, resulting in the plasmids of WD2-7, WD3-7, WD4-7, WD5-7, WD6-7, and WD7. The ability of each truncated RACK1 protein to interact with the full-length PCBP1 (pBT-PCBP1) was examined using the bacteria two-hybrid system. The blank vector pTRG (indicated as empty vector) was also utilized as a negative control. A stronger interaction between PCBP1 and truncated RACK1 resulted in the stronger cell growth under the stringent selection, including 3-AT, chloramphenicol, tetracycline, and streptomycin and therefore the ability of cell growth was evaluated.
As shown in Fig. 3a, the full length of RACK1 possessing the strongest interaction ability with PCBP1 showed the strongest cell growth. The WD1 domain deletion (WD2-7) resulted in no significant loss of growth ability as compared with the full-length PCBP1. Deletion of WD1 and 2 domains (WD3-7) resulted in a slight decrease of interaction, and further deletion (WD4-7) displayed a significant decrease of interacting ability as compared to WD3-7, implicating that WD4 domain may have a negative effect on the interaction ability. Deletion of WD1-5 domains (WD6-7) demonstrated no further change of PCBP1 interacting ability as compared to that of WD5-7. Most significant, the WD7 domain alone possessed the ability to interact with PBCP1, similar to the full-length PCBP1. These results demonstrated that the WD7 of RACK1 is the key domain, which is responsible for the interaction with PCBP1.
To further examine if the other individual WD domains also processed the PCBP1 interacting ability, various plasmids containing each of different WD domains of RACK1 were constructed. Using PCR with the sense primer containing the ATG and the antisense primer containing the stop codon, each generated WD fragment was cloned into pCR2.1 vector and was then sequenced. The correct insert was subcloned into pTRG plasmid, resulting in WD1, WD2, WD3, WD4, WD5, and WD6 constructs, respectively (Fig. 3b). Using the two-hybrid system, each plasmid was co-transformed with pBT-PCBP1 plasmid and screened under the high stringent selection, as described above. As shown in Fig. 3b, in addition to the WD7 domain, the WD2 and WD3 domains of RACK1 also displayed some degree of ability (approximately 1/3) to interact with PCBP1, but with a weaker capability as compared to that of the WD7 domain. The WD6 domain had a marginal ability, while the other WD domains (WD1, 4 and 5) possessed no PCBP1 interacting ability.
Taken together, these results demonstrated that the WD7 domain of RACK1 is the key domain and contributes to the strong interaction with PCBP1, with WD2 and WD3 domains contributing to a weak interaction.
RACK 1 regulates the human mu-opioid (MOR) promoter activity
The data above demonstrated the physical interaction between PCBP1 and RACK1 and therefore, the functional effect of this interaction was next investigated. Because of PCBP1 participating in the regulation of MOR promoter activity (Malik et al. 2006; Rivera-Gines et al. 2006; Cook et al. 2010), the effect of RACK1 on human MOR (hMOR) promoter activity was examined using NMB cells. RACK1 cDNA was cloned into a mammalian pcDNA3 expression vector, resulting in the pcDNA3-RACK1 plasmid. Luciferase reporter assay was performed by co-transfection of 0.2, 0.5, and 1 μg of pcDNA3-RACK1 plasmid or blank vector pcDNA3 as a control, along with the p357 plasmid, a luciferase reporter plasmid containing the hMOR promoter (Cook et al. 2010). In addition, the pCH110 plasmid containing β-galactosidase was also included in every transfection for normalization use. The normalized activity obtained from cells transfected with the blank vector was designated as 100%.
As shown in Fig. 4a, the reporter activity decreased as the amount of pcDNA3-RACK1 plasmid increased (indicated by the gray bar), demonstrating that RACK1 over-expression can negatively regulate hMOR promoter activity. This result suggested that a functional effect of RACK1 and PCBP1 interaction can modulate the hMOR promoter activity.
Effect of RACK1 knockdown on MOR mRNA level
To confirm the regulatory effect observed in Fig. 4a, RACK1 siRNA knock-down experiment was performed. Various concentrations (50, 100, and 200 nM) of RACK1 siRNA were transfected into NMB cells, and the scrambled siRNA as a control (C) was also used. Forty-eight hours after transfection, RNAs from transfected cells were extracted and subjected to RT-PCR analysis. As shown in Fig. 4b, the increasing amounts of RACK1 siRNA significantly decreased RACK1 mRNA levels incrementally. The mRNA level of β-actin was used as an internal reference for normalization, and the amount of endogenous RACK1 mRNA from the control cells was defined as 100%. The quantitative data are shown in Fig. 4c. These results demonstrated that RACK1 siRNA successfully knocked down the endogenous RACK1 mRNA level in NMB cells.
The effects of various amounts of RACK1 siRNA on the hMOR mRNA level were then determined using the same RNA preparations. As shown in Fig. 4d, in a dose-dependent manner, a significant increase of the endogenous hMOR mRNA level was detected via RT-PCR analysis as RACK1 mRNA level decreased. The quantitative data are shown in Fig. 4e, using β-actin as an internal control for normalization. The amount of endogenous hMOR mRNA from control cells (C) was arbitrarily defined as 100%. Collectively, these results confirmed that RACK1 negatively regulates the hMOR gene expression.
Effect of RACK1 knockdown on MOR protein level
To investigate if the alteration of MOR mRNA level also resulted in a change of MOR protein level, the endogenous level of RACK1 protein in NMB cells transfected with 200 nM RACK1 siRNA was examined first using western blot analysis. No significant change of RACK1 protein level was observed at 48 h after RACK1 siRNA transfection (data not shown). Therefore, the changes of mRNA and protein levels were further examined at 72 h time point after the siRNA transfection. As shown in Fig. 5a, RACK1 mRNA level remained low at 72 h after transfection, as compared to the control (C), using RT-PCR analysis with β-actin as an internal control. As shown in Fig. 5b, using western blot analysis with anti-RACK1 antibody, the endogenous RACK1 protein level was significantly decreased (approximately 50%) at 72 h, as compared to that of control (defined as 100%). The β-actin band was used as an internal standard for normalization, and the quantitative data is shown in the bar graph. These results showed that the endogenous mRNA of RACK1 protein remained low at 72 h after siRNA transfection, and the significant decrease of RACK1 protein levels was detectable at 72 h but not 48 h after transfection.
The endogenous level of PCBP1 in the RACK1 siRNA treated cells was further tested. There was no significant change of PCBP1 mRNA level observed (Fig. 5c), as compared with the control (C). In addition, no significant changes of PBCP1 protein level were detected (Fig. 5d), as compared with control, using western blot analysis with anti-PCBP1 antibody. The β-tubulin was used as the internal standard here, because of the close proximity of β-actin band and PCBP1 band. The quantitative data are shown using the bar graph. These results suggested no significant change of the endogenous PCBP1 level in cells transfected with RACK1 siRNA.
We next determined the endogenous hMOR mRNA and hMOR protein levels at 72 h after RACK1 siRNA transection. As shown in Fig. 6a, in cells with RACK1 knock-down, the endogenous level of hMOR mRNA was still elevated as compared to the control at 72 h. The hMOR protein was then examined.
The hMOR protein is a receptor, which can be bound by a specific opioid ligand, such as diprenorphine. Therefore, the hMOR protein level was determined using 3H-diprenorphine receptor binding assay with 1 μM CTAP as the competitive ligand, a specific mu-opioid antagonist. As NMB cells also contain delta and kappa opioid receptors (Baumhaker et al. 1994), 1 μM of DADLE (delta) and U50488 (kappa) were used in all experiments to block the binding to these receptors. As shown in Fig. 6b, a significant increase of hMOR receptor binding was observed in cells receiving RACK1 siRNA transfection, as compared with that of control cells. The data demonstrated that RACK1 knock-down caused an increase of hMOR mRNA level, and also resulted in a significant increase of hMOR protein level at 72 h after transfection.
In conclusion, these results showed that RACK1 knock-down lowered the endogenous level of RACK1 protein, which in turn increased the endogenous MOR protein level in NMB cells.
In this study, RACK1 was identified as a PCBP1-interacting protein via screening a human cDNA library using the bacteria two-hybrid system and by the co-immunoprecipitation assay. Effect of the functional RACK1-PCBP1 interaction on the human MOR (hMOR) gene expression was also revealed using the human NMB neuronal cell model system.
PCBP1 is known to participate in the hMOR gene regulation (Cook et al. 2010); therefore, physical interaction (Figs 2, 3) between PCBP1 and RACK1 in turn can bestow RACK1 modulating the hMOR gene expression is reasonable. This RACK1 functional role is strongly supported by the observations that (i) an increase of RACK1 by over-expression decreased the hMOR gene expression as determined using the reporter gene assay (Fig. 4a) and (ii) a decrease of RACK1 by siRNA knock-down increased the hMOR expression (Figs 4d–e and 6a, b). Interestingly, the changes of RACK1 and hMOR mRNA levels were evident at 48 and 72 h after RACK1 siRNA transfection (Figs 4b–e, 5a, and 6a), but the detectable changes at the protein level required 72 h (Figs 5b and 6b), implicating that approximately a 24-h lag is required to effectively translate the information of altered mRNA amounts into the detectable change of protein level using western blot analysis, in addition to the turnover rate of MOR protein. However, this discrepancy could also reflect the differences in detection limits and/or sensitivities between western blot and RT-PCR analyses.
RACK1 was originally identified as a receptor for activated protein kinase C (Ron et al. 1995). It contains seven Trp-Asp repeats (WD domains), which can mediate protein–protein interaction (McCahill et al. 2002; Tcherkasowa et al. 2002; Imai et al. 2009). From this study, the key interaction region of RACK1-PCBP1 is mapped at the C-terminus WD7 of RACK1 protein (Fig. 3a–b), for WD7 alone can provide the interactive capability similar to that of the full-length RACK1. In addition, the WD2 or WD3 domain alone confers a minor capability, and WD6 possesses a marginal interacting ability, whereas WD1, 4 and 5 domains contain no interactive capability. These conclusions were also supported by the serial WD deletion results, for WD1 deletion (WD2-7) showed no significant change on the cell growth as compared with the full-length RACK1 (WD1-7), supporting that WD1 contributed no PCBP1-interacting ability. Deletion of WD2 and WD3 resulted in a partial decrease of interactive ability, corroborating that there is a degree of interaction ability attributing from WD2 and WD3 domains. Interestingly, the construct containing WD4-7 displayed a further decrease of interactive ability, and the removal of WD4 (WD5-7) improved the capability to interact. These results implicates that WD4 domain may confer a negative effect, possibly a steric hindrance, on the PCBP1-RACK1 interaction ability. Particularly, WD4 itself possesses no PCBP1 interaction capability. In addition, it is also possible that WD4 may interact with a negative regulator, which can inhibit or interfere the PCBP1-RACK1 interaction. There was no significant difference between WD5-7 and WD6-7 constructs, suggesting that WD5 domain contributes no direct effect to the PBCP1-RACK1 interaction, which is collaborated with the WD5 alone possessing no interaction ability. Although WD6 alone possesses a slight interaction capability, yet the combination of WD6 and WD7 domains resulted in a decrease of interaction ability as compared to WD 7 alone, implicating a possible conformation interaction or a degree of steric hindrance between WD6 and 7. Lastly and critically, WD7 alone is sufficient for the full interactive ability with PCBP1.
The WD motif has the bladed propeller structure (Buensuceso et al. 2001; Sondek and Siderovski 2001), and can form hydrogen bonding between the β strands of the propeller (Sklan et al. 2006). In addition, the sequences of RACK1 WD domains are well conserved in various species, such as plants (Nakashima et al. 2008), Drosophila, C. elegans (Julie et al. 2007), mammals and humans (Guillemot et al. 1989). However, on the basis of the two-hybrid results using the constructs containing each single RACK1 WD domain, our data clearly suggested that the individual WD domains of RACK1 are not functionally equivalent, though each WD domain has the bladed propeller structure. There are at least two possibilities, which contribute to the uniqueness of each WD motif: (i) Each WD domain possesses different sequences (Fig. 3c), though each domain contains the WD repeat with the bladed propeller structure. (ii) The region connecting two sequential WD domains, called the loop of RACK1 (Garcia-Higuera et al. 1998), may differ in size and sequence, contributing to the unique features of each WD motif.
The WD motif may also provide a docking platform for various protein–protein interactions (Chen et al. 2004). Therefore, via the different WD domains, RACK1 may coordinate various protein interactions. RACK1 has been reported to be able to interact with several proteins, such as protein kinase C (PKC), src kinase and insulin receptor/IGF1R. The interaction site for RACK1-PKC was mapped at WD3 and/or WD5 and 6 domains (Buensuceso et al. 2001; McCahill et al. 2002). WD6 domain mediates RACK1-src interaction (McCahill et al. 2002), and WD 1-4 domains are responsible for RACK1-insulin receptor/IGF1R interaction (Zhang et al. 2006). This study provides additional evidence that RACK1 interacts with PCBP1 via WD7 domain predominantly. Taken together, RACK1 possesses capability to interact with various proteins via different WD domains. Furthermore, proteins may also interact with RACK1 at the sides, below or at its border of the propeller (Sklan et al. 2006).
There are several reports related to the functional roles of RACK1 and its interacting proteins. For example, RACK1 has been suggested to serve as a scaffold protein for PKC isoforms, and it enables translocation and stabilization of PKC isoforms (Mochly-Rosen and Gordon 1998). The reduction of RACK 1 levels was shown to correlate with malfunctioning of PKC translocation in aging rat brain (Battaini et al. 1997). It has also been shown to impair insulin-induced kinase translocation, Xenopus oocyte maturation (Ron et al. 1995) and regulation of calcium channels in cardiomyocytes (Zhang et al. 2006). A study of RACK1-src kinase interaction showed that RACK1 over-expression can decrease src activity and cell growth rates in NIH 3T3 cells (Chang et al. 1998). Collectively, these reports and our data demonstrated that RACK1 plays various functional roles depending on the particular proteins it interacts with.
Thus, based on this study using human neuronal NMB cell system, a functional role of RACK1-PCBP1 interaction and its mechanism governing the MOR gene expression are proposed in Fig. 6c. A simplified model depicts a center surrounding the dynamic interaction/equilibrium between free forms of RACK1 and PCBP1 (on left-hand side) as well as the RACK1-PCBP1 complex (on the right-hand side of the equilibrium): (i) The free form of PCBP1 can enter the nucleus and is known to act as transcriptional regulator of MOR gene by increasing MOR transcription (Ko and Loh 2005; Malik et al. 2006; Cook et al. 2010). (ii) When an elevated amount of RACK1 is achieved via over-expression, RACK1 interacts with PCBP1, shifting the equilibrium toward the right-hand side to generate more RACK1-PCBP1 complexes, reducing the available amount of free PCBP1, and then resulting in a decrease of MOR gene expression. (iii) Conversely, when a decrease of endogenous RACK1 level is achieved via RACK1 siRNA knock-down, the equilibrium is then driven toward the left-hand side, resulting in more free form of PCBP1, increasing MOR mRNA expression (iv), which then in turn increases the MOR protein level via post-transcriptional events (v) as well as translation and post-translational events (vi).
In addition, this study also found that RACK1 was mainly distributed in the cytosol with a few hot spots in the nuclei of NMB cells (Fig. 1b, indicated by arrows). The hot spots in the nucleus may suggest a potential functional role of RACK1 in the nucleus, though its functional role in the nucleus is unknown for now. Furthermore, PCBP1 is also expressed in the entire cell with many hot spots found in the nucleus of neuronal cells (Berry et al. 2006). Although these two proteins are distributed throughout the entire cell, importantly, PCBP1-RACK1 interaction takes place in the cytosol, but not in the nucleus of NMB cells (Fig. 2c). The actual reason is unknown; however, there are several potential explanations. For example, the change of a protein's phosphorylation/dephosphorylation status may influence its protein–protein interaction ability. It has been reported that PCBP1 nuclear retention is enhanced via its phosphorylation (Meng et al. 2007). Therefore, the protein–protein interaction ability of phosphorylated PCBP1 may be compromised in the nucleus. In addition, the phosphorylated/dephosphorylated RACK1 may also possess different protein–protein interaction ability. A second possibility is protein compartmentalization, with the protein in a particular microenvironment not available for the protein–protein interaction. RACK1 demonstrates the punctate staining in the nucleus (Fig. 1b), suggesting that RACK1 may be compartmentalized and thus not available for interacting with PCBP1 in the nucleus. Still another possibility is that RACK1 may directly or indirectly regulate MOR gene expression via an unidentified factor in the nucleus. These possibilities will need to be further investigated.
In summary, this study shows that RACK1 is a PCBP1-interacting protein and can modulate hMOR gene regulation as determined using the human NMB neuronal cell model system. This new functional role of PCBP1-RACK1 interaction provides some insight into MOR regulation and may be helpful to develop potential strategies of alternating MOR gene expression.
We thank Drs. Hsien-Ching Liu, Andrew P. Smith, and Linda Hsu for editing the manuscript. Fluoview 1000 confocal microscope was supported by a MRI grant of NSF. This research was supported by NIH research grant DA016673 and Biological Research Fund in Seton Hall University. The authors declare no conflict of interest.