Receptor for activated C kinase 1 promotes hepatocellular carcinoma growth by enhancing mitogen-activated protein kinase kinase 7 activity

Authors


  • Potential conflict of interest: Nothing to report.

  • This work was supported by grants from the National Natural Science Foundation of China (grant nos. 30973547 and 31100544), the Key Natural Science Program of Beijing (grant no.: 7101008), and the National Key Basic Research Program of China (grant no.: 2010CB911904).

Abstract

c-Jun N-terminal protein kinase (JNK) is a member of the mitogen-activated protein kinase (MAPK) superfamily. The activation of JNK is mediated by sequential protein phosphorylation through a MAPK module, namely, MAPK kinase kinase (MAP3K or MEKK) → MAPK kinase (MAP2K or MKK) → MAPK. Elevated levels of JNK activity have been frequently observed in hepatocellular carcinoma (HCC) and have been demonstrated to contribute to HCC growth by promoting HCC cell proliferation and resistance to tumor necrosis factor–related apoptosis-inducing ligand (TRAIL)- or Fas-mediated apoptosis. Chronic inflammation contributes to the up-regulation of JNK activity in HCC. However, it remains unknown whether aberrant JNK activity also results from some cell intrinsic defect(s). Here, we show that receptor for activated C kinase 1 (RACK1), an adaptor protein implicated in the regulation of multiple signaling pathways, could engage in a direct interaction with MKK7, the JNK-specific MAP2K, in human HCC cells. Levels of RACK1 protein show correlation with the activity of the JNK pathway in human HCC tissues and cell lines. RACK1 loss-of-function or gain-of-function analyses indicate that RACK1 enhances MKK7/JNK activity in human HCC cells. Further exploration reveals that the interaction of RACK1 with MKK7 is required for the enhancement of MKK7/JNK activity by RACK1. RACK1/MKK7 interaction facilitates the association of MKK7 with MAP3Ks, thereby enhancing MKK7 activity and promoting in vitro HCC cell proliferation and resistance to TRAIL- or Fas-mediated apoptosis as well as in vivo tumor growth. Conclusion: Overexpressed RACK1 augments JNK activity and thereby promotes HCC growth through directly binding to MKK7 and enhancing MKK7 activity. (HEPATOLOGY 2013)

Hepatocellular carcinoma (HCC) is the third-most common cause of cancer death worldwide, particularly in Africa and Asia.1 Even though extensive studies have shown that chronic inflammation associated with persistent viral infections and/or persistent exposure to hepatotoxic agents is clearly the primary inducer of HCC, the molecular events controlling the development and progression of HCC remain elusive.1

Recently, c-Jun N terminal protein kinase (JNK) has been implicated in regulating liver tumorigenesis. JNK is a member of the mitogen-activated protein kinase (MAPK) superfamily, which also includes extracellular signal-regulated kinase (ERK) and p38 family of kinases.2-5 JNK has two ubiquitously expressed isoforms, JNK1 and JNK2, and a tissue-specific isoform (JNK3), with different splicing forms (p54 and p46).3, 4 The activation of JNK is mediated by sequential protein phosphorylation through a MAPK module, namely, MAPK kinase kinase (MAP3K or MEKK) → MAPK kinase (MAP2K or MKK) → MAPK, in response to various cytokines and environmental factors.2-5 Two MAP2Ks (MKK4 and MKK7) have been identified to play a nonredundant role in the dual phosphorylation of JNK at Thr183 and Tyr185 (P-JNK).2-5 MKK4 also activates p38, whereas MKK7 specifically activates JNK.3, 5 Elevated levels of P-JNK have been frequently observed in HCC.6-8 Pharmacologic inhibition of JNK activity suppresses the growth of both xenografted human HCC cells and chemically induced mouse/rat liver cancers.6, 9 Mice lacking JNK1 display impaired liver carcinogenesis as a result of decreased tumor cell proliferation, whereas mice lacking p38α or IκB kinase (IKK)-β in hepatocytes exhibit enhanced liver cancers because of augmented JNK1 activity.6, 10, 11 Thus, JNK activity, especially JNK1 activity, is essential for the proliferation of liver cancer cells.6, 11 JNK activity might also contribute to liver tumorigenesis by rendering HCC cells resistant to tumor necrosis factor–related apoptosis inducing ligand (TRAIL)- or Fas-mediated apoptosis.12, 13 Apparently, chronic inflammation contributes to the augmented levels of P-JNK in HCC. However, it remains unknown whether aberrant JNK activity also results from some cell-intrinsic defect(s).6

Numerous intracellular signaling molecules, including receptor for activated C kinase 1 (RACK1), have been implicated in regulating the activity of the JNK pathway.14, 15 RACK1 was originally identified on the basis of its ability to anchor the activated form of protein kinase C (PKC) and is currently recognized as a multifunctional scaffold protein.14, 15 It has been reported that RACK1 can associate with both PKC and JNK, which enables PKC to phosphorylate JNK at Ser129 and thereby facilitates the basal and inducible dual phosphorylation of JNK by MKK4/7 in human melanoma cells.14, 16 However, the interaction of RACK1 with JNK was not detected by another group in COS7 African green monkey kidney cells. Instead, the binding of RACK1 to MEKK4 has been revealed to be essential, but not sufficient, for MEKK4-mediated JNK activation in this cell model.15 Thus, the molecular mechanism by which RACK1 regulates the JNK pathway may be cell context dependent. Because elevated levels of RACK1 messenger RNA have been independently observed in clinical HCC samples,17, 18 it is of importance to explore how RACK1 is involved in hepatic carcinogenesis. Here, we report that overexpressed RACK1 augments JNK activity and thereby promotes HCC growth through directly binding to MKK7 and enhancing MKK7 activity.

Abbreviations

Ab, antibody; Co-IP, coimmunoprecipitation; ERK, extracellular signal-regulated kinase; GFP, green fluorescent protein; GST, glutathione S-transferase; HCC, hepatocellular carcinoma; IB, immunoblotting; IgG, immunoglobulin G; IKK, IκB kinase; IP, immunoprecipitation; JNK, c-Jun N terminal protein kinase; MAPK, mitogen-activated protein kinase; MAP3K or MEKK, MAPK kinase kinase; MAP2K or MKK, MAPK kinase; P-JNK, dual phosphorylation of JNK at Thr183 and Tyr185; P-MKK7, dual phosphorylation of MKK7 at Ser271 and Thr275; PH, partial hepatectomy; PKC, protein kinase C; RACK1, receptor for activated C kinase 1; shRNA, short hairpin RNA; siRNA, small interfering RNA; TRAIL, tumor necrosis factor–related apoptosis-inducing ligand; WT, wild type.

Materials and Methods

Clinical HCC Samples.

Formaldehyde-fixed and paraffin-embedded clinical HCC samples were examined for P-JNK (Cell Signaling Technology, Beverly, MA) and RACK1 (BD Biosciences, San Jose, CA) staining on tissue microarray slides (US Biomax, Inc., Rockville, MD; see detailed clinicopathological features in Supporting Table 2). Patients' consent and approval by the local ethics committee were obtained for the use of the clinical materials in research.

In Vivo Tumor Growth.

Male athymic BALB/c nude mice were purchased from the Institutes of Experimental Animals, Academy of Chinese Medical Sciences (Beijing, China), and maintained under specific pathogen-free conditions. All experiments were performed in accord with institutional guidelines for animal care. Six- to eight-week-old nude mice were subcutaneously inoculated with human HCC cells (1 × 106/0.2 mL of phosphate-buffered saline; n = 10). Lengths and widths of tumors were measured with a caliper at the indicated time points.

A full description of additional Materials and Methods are given in the Supporting Materials.

Results

Interaction of RACK1 With MKK7 In Vitro and In Vivo.

RACK1 was shown to bind MKK7 in the yeast two-hybrid system (Supporting Table 1). The interaction of RACK1 with MKK7 was confirmed by coimmunoprecipitation (Co-IP) analysis: Green fluorescent protein (GFP)-MKK7 coprecipitated with coexpressed FLAG-RACK1, and FLAG-RACK1 coprecipitated with coexpressed GFP-MKK7 in human embryonic kidney 293T cells (Fig. 1A,B). To test whether RACK1 interacts directly with MKK7, we performed in vitro glutathione S-transferase (GST) pull-down assays. As expected, substantial FLAG-RACK1 and endogenous RACK1 in lysates of 293T cells was precipitated specifically by GST-MKK7, but not by GST alone (Fig. 1C). Moreover, in vitro translated FLAG-RACK1 was also precipitated specifically by GST-MKK7, but not by GST alone (Fig. 1D). The possible physiological interaction of RACK1 with MKK7 in HCC cells was then analyzed by immunoprecipitating endogenous proteins. MKK7 was present in immunoprecipitates obtained from lysates of HepG2 human HCC cells with an antibody (Ab) against RACK1, whereas no MKK7 coprecipitated when we used a control Ab (healthy rabbit immunoglobulin G; IgG) (Fig. 1E). Moreover, endogenous RACK1 in HepG2 cells coprecipitated with endogenous MKK7 (Fig. 1F). Collectively, our data suggest that RACK1 could engage in a direct interaction with MKK7 in human HCC cells.

Figure 1.

RACK1 interacts with MKK7 in vitro and in vivo. (A and B) 293T cells were transfected with various mammalian expression vectors as indicated. Cell lysates were subjected to immunoprecipitation (IP) with the indicated Abs. Precipitates were then subjected to SDS-PAGE and IB with the indicated Abs. WCL, whole-cell lysates. (C and D) GST-MKK7 or GST bound to glutathione-Sepharose (GSH) beads were incubated with lysates of 293T cells expressing FLAG-RACK1 or in vitro translated FLAG-RACK1. Precipitates were subjected to SDS-PAGE and IB with an anti-RACK1 Ab. (E and F) IB analysis of the interaction between endogenous MKK7 and endogenous RACK1 in lysates of HepG2 cells after IP with an anti-RACK1 Ab (E, control Ab: rabbit IgG) or an anti-MKK7 Ab (F, control Ab: goat IgG). SDS-PAGE, sodium dodecyl sulfate polyacrylamide gel electrophoresis.

Correlation Between Levels of RACK1 Protein and Activity of the JNK Pathway in Human HCC.

RACK1 protein levels in clinical HCC tissues were analyzed by immunohistochemical staining on tissue microarray slides. Among the 130 HCC samples used (Supporting Table 2), matched peritumoral liver tissues from the same patient were available for 12 HCC samples. Consistent with previous reports,17, 18 clinical HCC tissues exhibited more elevated RACK1 expression than matched peritumoral liver tissues (Fig. 2A; Supporting Table 3). Levels of RACK1 protein in the 130 HCC tissues were well associated with the clinical stage (Supporting Table 4). In addition, even not statistically significant, RACK1 expression in HCC tissues showed a tendency to be correlated with tumor size (Supporting Table 4).

Figure 2.

Levels of RACK1 protein are correlated with the activity of the JNK pathway in human HCC. (A) One hudred and thirty formaldehyde-fixed and paraffin-embedded clinical HCC samples were examined for P-JNK and RACK1 staining on tissue microarray slides. Representative paired samples are shown. Scale bar: 50 μm. (B) Levels of P-JNK were correlated with those of RACK1 in clinical HCC tissues. Kendall r = 0.4861; P < 0.0001. (C) Proteins extracted from 12 different human HCC cell lines and one immortalized healthy human hepatocyte line (HL-7702) were subjected to IB with the indicated Abs.

Now that RACK1 could engage in a direct interaction with MKK7 in human HCC cells, it is possible that accumulated RACK1 protein contributes to elevated JNK activity in HCC. In this scenario, levels of P-JNK were also analyzed. As expected, clinical HCC tissues exhibited more elevated P-JNK expression than matched peritumoral liver tissues (Fig. 2A; Supporting Table 5). Levels of P-JNK in the 130 HCC tissues were associated with tumor size and pathological grade (Supporting Table 6). More important, levels of RACK1 protein in clinical HCC tissues were positively correlated with those of P-JNK (Fig. 2A,B).

Furthermore, immunoblotting (IB) analysis revealed that 10 (Huh7, SK-Hep-1, Hep3B, BEL-7404, PLC/PRF/5, HepG2, Li-7, SMMC7721, BEL-7402, and MHCC-97H) of the 12 human HCC cell lines examined exhibited elevated RACK1 expression, albeit to varied extent, as compared with immortalized healthy human hepatocyte line HL-7702 (Fig. 2C). All cell lines with elevated RACK1 expression, except MHCC-97H, showed up-regulated levels of P-JNK, as compared with HL-7702 (Fig. 2C). These data further indicate a correlation between levels of RACK1 protein and JNK activity in human HCC cells. Interestingly, levels of MKK7 phosphorylation (P-MKK7) at Ser271 and Thr275, which is required for MKK7 activity,2-5 were also up-regulated in Huh7, SK-Hep-1, BEL-7404, PLC/PRF/5, HepG2, SMMC-7721, and BEL-7402 cells (Fig. 2C), suggesting that MKK7 is implicated in the regulation of JNK activity by RACK1.

JNK activity has been suggested to contribute not only to the proliferation of liver cancer cells, but also that of healthy hepatocytes.19 It is possible that increased RACK1 levels can simply be the consequence of an increased proliferative activity of hepatocytes. In this scenario, two-thirds partial hepatectomy (PH) was performed to explore the levels of RACK1 protein during liver regeneration. IB analysis revealed a marginal increase in RACK1 protein levels under the conditions that PH triggered robust proliferation of healthy hepatocytes (Supporting Fig. 1). Thus, elevated RACK1 expression does not simply result from an increased proliferative activity of hepatocytes. Collectively, our data suggest a critical role of RACK1 in HCC development.

Enhancement of MKK7/JNK Activity by RACK1 in Human HCC Cells.

To address whether RACK1 indeed affects the activity of the JNK pathway in human HCC cells, we analyzed the effects of RACK1 loss-of-function or gain-of-function in HepG2 cells because this cell model showed moderate RACK1 and P-JNK up-regulation (Fig. 2C). IB analysis revealed that transient or stable silencing of endogenous RACK1 expression by RACK1 small interfering RNA (siRNA) or short hairpin RNAs (shRNAs) in HepG2 cells significantly suppressed basal levels of P-JNK. Reduced P-JNK levels under the condition of RACK1 knockdown were associated with decreased P-MKK7 levels (Fig. 3A-C). Similar phenomena were also observed in Huh7 and SK-Hep-1 cells (Fig. 3B). By contrast, transient ectopic expression of RACK1 in HepG2 cells led to substantially enhanced basal levels of both P-JNK and P-MKK7 (Fig. 3D). Moreover, single-clone HepG2 stable transfectants (named FLAG-RACK1Low and FLAG-RACK1high, respectively, according to levels of FLAG-RACK1 protein) also exhibited augmented levels of P-JNK and P-MKK7, which were well correlated with FLAG-RACK1 expression (Fig. 3E). These data collectively indicate that RACK1 contributes to enhanced levels of P-MKK7/P-JNK in human HCC cells.

Figure 3.

RACK1 contributes to enhanced levels of P-MKK7/P-JNK in human HCC cells. (A-C) Forty-eight hours after transfection of HepG2 cells with RACK1 siRNA or control (Ctrl) nontargeting siRNA (A), or 96 hours after infection of HepG2 cells, Huh7, or SK-Hep-1 cells with control lentivirus or lentivirus carrying RACK1 shRNA 1# (B), or HepG2 single clones stably expressing control shRNA or RACK1 shRNAs 2# and 3# were generated (C), cell lysates were then harvested and subjected to IB with the indicated Abs. (D and E) Forty-eight hours after transfection of HepG2 cells with mammalian expression vectors encoding GFP or GFP-RACK1 (D), or HepG2 single clones stably expressing FLAG-RACK1 and the mock control were generated (E), cell lysates were then harvested and subjected to IB with the indicated Abs.

Mapping RACK1/MKK7 Interacting Regions.

MKK7 is composed of an N-terminal JNK-binding domain and a kinase domain, whereas RACK1 contains seven Trp-Asp (WD) repeats.14, 15, 20 RACK1/MKK7-interacting regions were analyzed through generating several deletion mutants (Fig. 4A), followed by Co-IP analysis in 293T cells. FLAG-RACK1 coprecipitated with coexpressed kinase domain of MKK7 (MKK7 ΔN), but not with coexpressed JNK-binding domain of MKK7 (MKK7 ΔC) (Fig. 4B). On the other hand, GFP-MKK7 coprecipitated with coexpressed RACK1 deletion mutant that included WD domains five to seven (RACK1 WD5-7), but not with coexpressed RACK1 WD1-4 (Fig. 4C). Furthermore, a WD6- or WD7-truncated RACK1 mutant (FLAG-WDΔ6 or FLAG-WDΔ7), but not FLAG-WDΔ5, showed significant reduced association with coexpressed GFP-MKK7 (Fig. 4D).

Figure 4.

Mapping RACK1/MKK7-interacting regions. (A) Deletion mutants of MKK7 and RACK1 used in domain-mapping experiments. Numbers in parentheses indicate amino acids included in construct. (B-F) 293T cells were transfected with various mammalian expression vectors, as indicated. Cell lysates were immunoprecipitated with indicated Abs. Precipitates were then subjected to sodium dodecyl sulfate polyacrylamide gel electrophoresis and IB with the indicated Abs.

WD6 and WD7 of RACK1 are the docking domains for various proteins, including MEKK4.14, 15 To analyze whether the direct interaction between RACK1 and MKK7 enhances the activity of the JNK pathway, it is of importance to identify the specific binding sites in RACK1. In this scenario, molecular simulations of MKK7 and RACK1 were performed according to the reported three-dimentional crystal structures of the proteins (Supporting Fig. 2A), followed by molecular docking. Among the candidates for the complex structure, the one with WD6 and WD7 in the interfaces was chosen. The predicted model suggested that three previously unidentified sites (amino acids 225-231 in WD6 and amino acids 269-272 and 275-280 in WD7) of RACK1 were essential for anchoring the kinase domain of MKK7 (Supporting Fig. 2A). Mutation of amino acids 269-272 [(269-272)Mut] or 275-280 [(275-280)Mut] in WD7 of RACK1 into alanines was predicted to lead to reduced binding activity without changes in the global structure of the protein, whereas mutation of amino acids 225-231 in WD6 of RACK1 into alanines was predicted to lead to changes in the global structure of the protein (Supporting Fig. 2B-D). Indeed, Co-IP analysis revealed that compared to wild-type (WT) FLAG-RACK1, FLAG-RACK1(269-272)Mut and FLAG-RACK1(275-280)Mut showed significantly reduced association with coexpressed GFP-MKK7 (Fig. 4E), but exhibited similar association with coexpressed HA-MEKK4 (Fig. 4F) and GFP-JNK1 (Supporting Fig. 3) in 293T cells. Thus, amino acids 269-272 and 275-280 are the specific binding sites in RACK1 to anchor MKK7.

Interaction of RACK1 With MKK7 Is Required for Enhancement of MKK7/JNK Activity by RACK1 in Human HCC Cells.

To test the possibility that RACK1 enhances MKK7/JNK activity in human HCC cells by directly binding to MKK7, HepG2 cells were transfected with mammalian expression vectors encoding tagged-RACK1 WT or mutants. IB analysis revealed that the mutation of either MKK7-binding site abrogated the enhancement of P-MKK7/P-JNK levels by RACK1 (Fig. 5A,B). Interestingly, ectopic expression of either RACK1 WT or mutants showed marginal effects on the phosphorylation of MKK4 (Fig. 5B), suggesting that RACK1 does not regulate MKK4 and MKK7 in the same manner. Such treatment also exhibited little effects on the phosphorylation of other substrates of MAP3Ks, including IKKα/β and MKK3/6 (Fig. 5B), suggesting the changes in P-MKK7 levels are not associated with overt alteration of the overall MAP3K activity. Notably, tagged-RACK1 WT and RACK1 mutants were expressed at comparable levels (Fig. 5A,B). To determine whether the mutants did lose the ability to interact with MKK7 in HepG2 cells, Co-IP analysis was performed. As expected, tagged-RACK1 WT, but not tagged-RACK1(269-272)Mut or tagged-RACK1(275-280)Mut, interacted with endogenous MKK7 in HepG2 cells (Fig. 5A,B). Thus, the interaction of RACK1 with MKK7 is indeed essential for the enhancement of MKK7/JNK activity by RACK1 in human HCC cells.

Figure 5.

Interaction of RACK1 with MKK7 is required for enhancement of MKK7/JNK activity by RACK1 in human HCC cells. HepG2 cells were transfected with various mammalian expression vectors, as indicated. Lysates of these cells were subjected to immunoprecipitation with indicated Abs. Precipitates as well as whole cell lysates were then subjected to sodium dodecyl sulfate polyacrylamide gel electrophoresis and IB with indicated Abs. P-MKK4, phospho-MKK4; P-MKK3/6, phospho-MKK3/6; P-IKKα/β, phospho-IKKα/β.

Interaction of RACK1 With MKK7 Facilitates the Binding of MKK7 to MAP3Ks.

The molecular mechanism(s) by which RACK1/MKK7 interaction leads to enhanced MKK7 activity are of interest. Because RACK1 interacts with the kinase domain of MKK7 (Fig. 4B) and the MAP3K docking site stretches within the C-terminal part of the MKK7 catalytic domain,2 it is possible that RACK1/MKK7 interaction might affect the docking interaction of MKK7 with upstream MAP3Ks. Indeed, Co-IP analysis revealed that the physiological interaction between MKK7 and several MKK7-specific MAP3Ks (MEKK2, MEKK3, and TAK1) in HepG2 cells (Fig. 6A) decreased under the condition of RACK1 knockdown (Fig. 6B). Consistently, nonradioactive in vitro kinase assays revealed that the phosphorylation of GST-MKK7 by suboptimal amount of HA-MEKK1, but not the autophosphorylation of GST-MKK7, was enhanced in the presence of FLAG-RACK1 immunoprecipitated from lysates of HepG2 cells (Fig. 6C). Thus, RACK1 promotes the binding of MKK7 to upstream MAP3Ks. RACK1/MKK7 interaction is essential during this process because the mutation of either MKK7-binding site abrogated the enhancing effects of RACK1 (Fig. 6D).

Figure 6.

Interaction of RACK1 with MKK7 facilitates the association between MKK7 and MAP3Ks. (A and B) IB analysis of the interaction between endogenous MKK7 and endogenous MEKK2/MEKK3/TAK1 in lysates of HepG2 cells with or without RACK1 knockdown after immunoprecipitation with an anti-MKK7 Ab (control Ab: goat IgG). (C and D) GST-MKK7 was incubated with or without FLAG-RACK1 WT or mutants immunoprecipitated from lysates of HepG2 cells for 3 hours at 4°C in kinase buffer. At the end of the incubation, nonradioactive adenosine triphosphate and a suboptimal amount of HA-MEKK1 were added to indicated samples. Samples were kept at 30°C for 60 minutes with gentle shaking and were then subjected to sodium dodecyl sulfate polyacrylamide gel electrophoresis and IB with indicated Abs. P-GST-MKK7, phospho-GST-MKK7.

Enhanced MKK7 Activity Plays a Key Role in the Protumorigenic Effects of RACK1 in HCC.

JNK activity contributes to HCC growth by promoting HCC cell proliferation and resistance to TRAIL- or Fas-mediated apoptosis.6, 11-13 Because RACK1 enhances MKK7/JNK activity by directly binding to MKK7 in human HCC cells, it is of importance to investigate how RACK1 may affect tumorigenic growth. HepG2 single clones stably expressing RACK1 shRNAs (Fig. 3C) exhibited dramatically reduced anchorage-independent growth and more apoptosis in response to TRAIL or anti-Fas Ab (CH11) (Fig. 7A,B). Similar effects of RACK1 knockdown (Fig. 3B) were also observed in Huh7 and SK-Hep-1 cells (Supporting Fig. 4). Furthermore, RACK1 knockdown led to impaired in vivo tumor growth (Fig. 7C). By contrast, anchorage-independent growth, resistance to TRAIL- or Fas-mediated apoptosis, and in vivo tumor growth were enhanced in HepG2 single clones stably expressing FLAG-RACK1 (Fig. 7D-F), which were well correlated with the levels of exogenous RACK1 protein (Fig. 3E).

Figure 7.

RACK1 contributes to HCC growth. (A and D) HepG2 single clones stably expressing FLAG-RACK1 (D) or corresponding shRNAs (A) were plated in soft agar (1 × 103/well) and assayed for colony number (no.) after 14 days. Values shown are mean ± SD. *P < 0.05; **P < 0.01. (B and E) HepG2 single clones stably expressing FLAG-RACK1 (E) or corresponding shRNAs (B) were treated with CH11 or TRAIL at indicated doses for 24 hours, followed by apoptosis analysis (Annexin V staining, mean ± SD). (C and F) HepG2 single clones stably expressing FLAG-RACK1 (F) or corresponding shRNAs (C) were injected into nude mice. After various periods of time, as indicated, tumors were measured with Vernier calipers (mean ± SD). Because nude mice inoculated with HepG2 FLAG-RACK1High cells died rapidly after 25 days postinjection, the measurement had to be stopped. SD, standard deviation.

To test whether enhanced MKK7 activity plays a key role in the protumorigenic effects of RACK1 in human HCC cells, we expressed MKK7 in HepG2 cells with RACK1 knockdown because overexpressed MKK7 has considerable basal enzymatic activity, possibly resulting from autophosphorylation.2, 5 We found that anchorage-independent growth, resistance to TRAIL- or Fas-mediated apoptosis, and in vivo tumor growth were dramatically decreased under the condition of RACK1 knockdown, but the cells became insensitive to the loss of RACK1 when MKK7 was ectopically expressed (Fig. 8). By contrast, ectopic expression of MKK4, which also has considerable basal enzymatic activity, possibly resulting from autophosphorylation,2 in HepG2 cells with RACK1 knockdown led to more impaired anchorage-independent growth and more apoptosis in response to TRAIL or anti-Fas Ab (Supporting Fig. 5). Taken together, these results suggest that RACK1 promotes HCC growth by enhancing MKK7 activity.

Figure 8.

Enhanced MKK7 activity plays a key role in the protumorigenic effects of RACK1. (A) HepG2 cells stably expressing RACK1 shRNA and MKK7 were generated, and cell lysates were then harvested and subjected to IB with indicated Abs. (B-D) Overexpression of MKK7 in HepG2 cells stably expressing RACK1 shRNA restored anchorage-independent growth (B), resistance to TRAIL- or Fas-mediated apoptosis (C; Annexin V staining), and in vivo tumor growth (D). Values shown are mean ± standard deviation. Differences between the RACK1 knockdown group and either of the other two groups are shown.

Discussion

RACK1, an adaptor protein implicated in the regulation of multiple signaling pathways, plays a context-dependent role in tumorigeneis.21 Our data show that human HCC tissues and cell lines exhibit augmented levels of RACK1 protein (Fig. 2), which contribute to HCC growth through, at least partially, enhancing the activity of the JNK pathway (Figs. 7 and 8). It should be noted that SMMC7721, BEL-7402, and BEL-7404 cells show significantly elevated RACK1 expression, but the levels of P-JNK in these cells are only weakly up-regulated (Fig. 2C). Thus, the role of RACK1 on the activity of the JNK pathway might be compromised by other genetic mutations. In addition, our data show that MHCC-97L, MHCC-97H, and HCCLM3, which have higher invasive capacity than the other cells, exhibit lower levels of RACK1 and P-MKK7/P-JNK (Fig. 2C). These findings are consistent with a recent report,22 and suggest that the role of JNK in HCC metastasis should be reevaluated. Because the levels of RACK1 protein are correlated with the clinical stage, whereas those of P-JNK are not (Supporting Tables 4 and 6), other signaling pathways regulated by RACK1 might collaborate with the JNK pathway to promote HCC development. The elucidation of the interplay of these signaling pathways and the underlying mechanisms of RACK1 overexpression might shed light on the treatment of HCC in the clinic.

The molecular mechanism by which RACK1 regulates JNK activity seems to be cell context dependent.14, 15 Our present study has revealed a novel molecular mechanism by which RACK1 regulates the JNK pathway—RACK1 augments MKK7/JNK activity by directly binding to MKK7 and enhancing MKK7 activity in human HCC cells. Because FLAG-RACK1 immunoprecipitated from lysates of HepG2 cells does not enhance the phosphorylation of GST-MKK7 without any manually added MAP3K in nonradioactive in vitro kinase assays (Fig. 6C), RACK1/MKK7 interaction does not enhance MKK7/JNK activity by enhancing MKK7 autophosphorylation. These data also indicate that no significant amount of endogenous MKK7-specific MAP3Ks coprecipitated with FLAG-RACK1. Consistently, we failed to detect endogenous MKK7-specific MAP3Ks in immunoprecipitates obtained from lysates of HepG2 human HCC cells with an Ab against RACK1 (data not shown). In addition, ectopic expression of RACK1 leads to no overt alteration of the overall MAP3K activity (Fig. 5B). Thus, it seems unlikely that RACK1 binds to any MKK7-specific MAP3K(s) directly in human HCC cells. Despite that, our data show that RACK1/MKK7 interaction facilitates the association of MKK7 with upstream MAP3Ks and, consequently, enhances P-MKK7 levels in human HCC cells (Fig. 6).

It is interesting that RACK1 shows no significant effects on P-MKK4 levels in human HCC cells (Fig. 5B and Supporting Fig. 5A). Consistently, we failed to detect the interaction of endogenous RACK1 with endogenous MKK4 in HepG2 cells (data no shown), most likely resulting from the poor homology of MKK4 and MKK7.3, 5 Moreover, our data suggest that MKK4 overexpression worsens, but not compensates, the loss of P-MKK7 in human HCC cells (Supporting Fig. 5). The same phenomenon has been reported in MKK7-null mouse embryonic fibroblasts, even though MKK4 deficiency leads to decreased proliferation of mouse embryonic fibroblasts, similar to MKK7 deficiency.23 The roles of endogenous MKK4 in HCC development remain to be explored. Increased MKK4 abundance also inhibits the proliferation of human fetal lung diploid fibroblasts.24 The regulation of p38 and/or yet unknown substrate(s) have been proposed to contribute to the inhibitory effects of MKK4.24, 25

In addition, RACK1 regulates the phosphorylation of both p54JNK and p46JNK (Figs. 3 and 5), whereas JNK1 (p46JNK1 as the predominant splicing form and p54JNK1 as the minor splicing form3-5, 20) is the major JNK isoform, which shows up-regulated activity in human HCC.6-8 We tried to resolve this puzzle with Abs that specifically immunoprecipitated JNK1 or JNK2 (Supporting Fig. 6A). Immune complex kinase assays suggest that RACK1 enhances JNK1 activity, but not JNK2 activity (Supporting Fig. 6B). It is possible that the specific activation of MKK7 and the enhanced binding to certain MAP3Ks make the discrepancy.

Although the above-mentioned uncertainties still await future studies, our data have confirmed the previously proposed notion that there may be factors other than chronic inflammation responsible for the augmented JNK activity in HCC6 and have identified that RACK1 overexpression is one such factor.

Acknowledgements

The authors thank Mrs. Xiaoling Lang and Chunmei Hou for their technical assistance.

Ancillary