Heterogeneous ribonucleoprotein (hnRNP) A1 is a member of the A/B subfamily of ubiquitously expressed hnRNPs, which have a wide variety of functions in gene expression and signal transduction. To investigate the biological function and clinical significance of hnRNP A1 in hepatocellular carcinoma (HCC), we measured hnRNP A1 expression in four HCC cell lines and two independent cohorts of HCC patients. We found that hnRNP A1 was overexpressed in the highly metastatic HCC cell lines and in tumor tissues of patients with recurrent HCC. Knockdown of hnRNP A1 in highly metastatic HCC cells caused a significant decrease in cell invasion, while upregulation of hnRNP A1 in poorly metastatic HCC cells led to a significant increase in their invasive capacity. We found that this effect may occur through the regulation of CD44v6 expression by hnRNP A1 in HCC cells. Both quantitative reverse transcription-polymerase chain reaction (qRT-RCR) and immunohistochemistry revealed that hnRNP A1 was upregulated in HCC tissues and coincided with overexpression of CD44v6. HCC patients with high hnRNP A1 tended to have higher levels of CD44v6, shorter overall survival (OS) and higher rates of tumor recurrence. Multivariate analyses revealed that hnRNP A1 alone or in combination with CD44v6 were independent prognostic indicators for OS and time to recurrence and have potential as therapeutic targets. In conclusion, overexpression of hnRNP A1 promotes HCC invasion by regulating the level of CD44v6 and indicates a poor prognosis for HCC patients after curative resection.
Hepatocellular carcinoma (HCC) is one of the most common and aggressive human malignancies worldwide.1 The prognosis of HCC remains dismal due to the high rate of tumor recurrence and metastasis after curative resection.1, 2 The molecular pathogenesis and complicated signal transduction pathways involved in HCC are not fully understood. Although several molecular markers that may define the risk of recurrence and the metastatic potential of HCC have been proposed, none have been approved for routine clinical use.3, 4
Heterogeneous ribonucleoproteins (hnRNPs) are a set of nuclear proteins that bind to nascent transcripts produced by RNA polymerase II and have a wide range of roles in DNA repair, telomere biogenesis, cell signaling and regulating gene expression at both transcriptional and translational levels.5 Emerging evidence suggests that some hnRNPs (including hnRNP A2B1,6–8 hnRNP M49 and hnRNP K10) are essential factors in tumor development and progression.11
HnRNP A1 is a member of the A/B subfamily of ubiquitously expressed hnRNPs, which have a wide variety of functions. It is involved in telomere biogenesis; the telomeres of an hnRNP A1-deficient cell line are shorter than those of a related normal hnRNP A1-expressing cell line.12 This hnRNP is also involved in protein–protein interactions, and through its N-terminal RNA-binding domain, interacts with inhibitory subunit of NF-κB alpha (IκBα), resulting in the activation of nuclear factor κ B (NF-κB).13 Recently, hnRNP A1 was found to be overexpressed in lung,14 breast15 and colon cancer,16 and its overexpression is a potential biomarker for colorectal cancer.17 Other studies have reported that hnRNP A1, as a repressor of splicing factors of the serine/arginine-rich family, participates in regulating the alternative splicing of CD44.18, 19 CD44 is a transmembrane glycoprotein, and the involvement of some variant isoforms (abbreviated CD44v) in tumor development and progression has been reported.20–22 However, the role of hnRNP A1 in HCC, and its relationship with CD44 and/or CD44 variant isoforms during HCC progression, remains elusive.
In this study, we investigated the expression of hnRNP A1 in HCC cell lines and in tumor tissues using quantitative reverse transcription-polymerase chain reaction (qRT-PCR) and western blot analyses. Then, we assessed the role of hnRNP A1 in HCC invasiveness in vitro. Finally, using tissue microarrays (TMAs) and immunohistochemistry, we evaluated the prognostic significance of hnRNP A1 in HCC patients.
TTR, time to recurrence; OS, overall survival; AFP, α-fetoprotein; γ-GT, γ-glutamyl transferase; TNM, tumor-nodes-metastasis; HR, hazard ratio; CI, confidential interval; NA, not adopted.
Material and Methods
Four human HCC cell lines were used in this study: two highly metastatic human HCC cell lines (MHCC97H and HCCLM3) that were established at our institute23 and two poorly metastatic human HCC cell lines were Hep3B and PLC/PRF/5 (American Type Culture Collection). All cell lines were routinely maintained in high-glucose Dulbecco's modified eagle medium (DMEM) supplemented with 10% fetal bovine serum (FBS) at 37°C under 5% CO2.
Patients and follow-ups
Two independent cohorts of HCC patients were enrolled in this study. To investigate the expression of hnRNP A1 in HCC tissues and its relationship with CD44v6, we collected 94 tumorous and peritumoral samples for qRT-PCR analysis (Cohort 1). Samples were consecutively collected from HCC patients undergoing curative resection at the Liver Cancer Institute, Zhongshan Hospital, between February 2007 and June 2007. These HCC patients were monitored after surgery, until November 11, 2010. All specimens were collected immediately after resection, transported in liquid nitrogen and stored at −80°C. Two peritumoral samples were excluded from the analysis due to poor RNA quantity. To evaluate the prognostic role of hnRNP A1 and/or CD44v6 in HCC, we obtained tumor specimens used in TMA analyses from 323 consecutive HCC patients who underwent curative resection at the Liver Cancer Institute, Zhongshan Hospital, between January 2003 and March 2004 (Cohort 2). These HCC patients were monitored after surgery, until March 15, 2009, with a median follow-up of 60 months (range: 2.0–74.0 months). Detailed clinicopathological features are listed in Supporting Information Table 1.
Liver function was assessed by Child-Pugh classification. Tumor stage was determined according to the 2010 International Union Against Cancer tumor-nodes-metastasis (TNM) classification system. Tumor differentiation was graded by the Edmondson grading system. Curative resection was defined as: (i) the complete resection of all tumor nodules and the cut surface being free from cancer on histological examination; (ii) no cancerous thrombus found in the portal vein (main trunk or two major branches), hepatic veins or bile duct; (iii) no extrahepatic metastasis and (iv) negative serology and imaging studies at 2 months after operation. Overall survival (OS) was defined as the interval between surgery and death or between surgery and the last observation point. For surviving patients, the data were censored at the last follow-up. Time to recurrence (TTR) was defined as the interval between the date of surgery and the date of diagnosis of any type of relapse (intrahepatic recurrence and extrahepatic metastasis).24 Ethical approval for the use of human subjects was obtained from the Research Ethics Committee of Zhongshan Hospital, and informed consent was obtained from each patient.
Cell transfection and clone selection
U6/Neo/GFP/RNAi-hnRNPA1 and U6/Neo/GFP/RNAi-NC plasmids were purchased from Shanghai Genechem Company (Shanghai, China). HnRNP A1 cDNA was cloned into the pcDNA3.1 vector to generate an expression vector. The U6/Neo/GFP/RNAi-hnRNPA1 plasmid was transfected into HCCLM3 cells, and the pcDNA3.1-hnRNP A1 plasmid was transfected into Hep3B cells using Lipofectamine 2000 (Invitrogen, Carlsbad, CA). U6/Neo/GFP/RNAi-NC and pcDNA3.1 plasmids were used as controls. The transfected cells were selected under 800 μg/mL G418 (Sigma) for 3–5 weeks. Stably transfected clones were validated by qRT-PCR and immunoblotting for hnRNP A1.
The siRNA sequences against CD44v6 were as previously described25: 5′-UGA GGG AUA UCG CCA AAC ATT-3′ (sense); 5′-UGU UUG GCG AUA UCC CUC ATT-3′ (antisense). CD44v6 cDNA was cloned into the pcDNA3.1 vector, and the expression vector was transfected into HCC cells using Lipofectamine 2000 reagent (Invitrogen).
Cell invasion assay
Cell invasion analysis was performed using a transwell insert (3422, Corning, NY) as previously described.26 The HCC cells, suspended in 100-μL DMEM with 1% FBS, were seeded into 8-μm pore membrane Boyden chambers (Costar, Washington, DC) coated with matrigel (BD Bioscience, Bedford, MA). DMEM containing 10% FBS was used in the lower chamber as a chemo-attractant. After incubation for 48 hr, migrating cells, on the bottom surface, were fixed with formaldehyde, stained with 0.1% crystal violet for 15 min and finally counted under a light microscope (magnification, ×200). All experiments were performed in triplicate.
qRT-PCR, western blot and immunohistochemistry
qRT-PCR and western blot were performed as previously described.27 The primers and primary antibodies used are listed under the Materials and Methods section in the Supporting Information. All experiments were performed in triplicate. Construction of TMA and immunohistochemistry are included in the Materials and Methods section in the Supporting Information.
The statistical analyses were performed using SPSS statistical software (version 16.0; SPSS, Chicago, IL). The cumulative recurrence and survival probability were evaluated using the Kaplan–Meier method, and differences were assessed using the log-rank test. Cox multivariate regression analysis was used to determine independent prognostic factors. Quantitative data between groups were compared using the Student t-test. Categorical data were analyzed by the chi-square test or Fisher exact test. Values were expressed as the mean ± standard deviation. Differences were deemed statistically significant when p < 0.05.
HnRNP A1 promotes a metastatic phenotype in HCC cell lines
Western blots and qRT-PCR showed that hnRNP A1 expression was significantly increased in the highly metastatic cell lines (MHCC97H and HCCLM3) relative to the poorly metastatic HCC cell lines (Hep3B and PLC/PRF/5; Fig. 1a; p < 0.01). Stable overexpression of hnRNP A1 in Hep3B cells and successful shRNA-mediated knockdown of hnRNP A1 expression in HCCLM3 cells were confirmed by qRT-PCR and western blot analyses (p < 0.01, Fig. 1b). In vitro invasion assays showed that the number of invasive hnRNP A1 shRNA-treated HCCLM3 cells was significantly decreased when compared with the control (26.7 ± 7.0 vs. 42.9 ± 12.0, p < 0.01; Figs. 1c and 1d), while the number of invasive hnRNP A1 cDNA-transfected Hep3B cells was significantly higher than that of the control cells (18.4 ± 7.1 vs. 8.3 ± 4.6, p < 0.01; Figs. 1c and 1d). These data indicate that hnRNP A1 expression is positively correlated with the metastatic potential of HCC cells, and hnRNP A1 is necessary to confer this metastatic phenotype in HCC cells.
HnRNP A1 regulates CD44v6 expression in HCC cells
To explore the mechanism underlying overexpression of hnRNP A1 in the promotion of HCC cell invasiveness, we investigated whether modulations of hnRNP A1 expression altered CD44 and/or CD44 isoform (CD44v4-5, CD44v6, CD44v7-8 and CD44v10) expression, which is associated with the progression of HCC.22 As shown in Figure 2a, upregulation of hnRNP A1 in Hep3B cells led to significantly increased CD44v6 expression (p < 0.01). Conversely, knockdown of hnRNP A1 in HCCLM3 cells resulted in a significant downregulation of CD44v6 expression (p < 0.01; Fig. 2a). However, the levels of CD44, CD44v7-8 and CD44v10 were not altered. Although CD44v4-5 expression changed slightly, this change is not statistically significant (p > 0.05, data not shown). In addition, we found that CD44v6 expression in HCC cells positively correlated with hnRNP A1 expression as measured by qRT-PCR and western blot (Fig. 2b).
Furthermore, we used RNA-affinity precipitation (RNAP) to identify binding between hnRNP A1 and CD44v6 exon sequences in HCCLM3 cells. We set up RNAP using biotinylated RNA oligonucleotides corresponding to three subdomains of the CD44v6 exon (L, M and R) (Supporting Information Fig. S1A).As a result, we observed that each of the three subdomains of the CD44v6 exon were bound to hnRNP A1 (Supporting Information Fig. S1C, Lanes 1–3). Western blot analysis of the RNA-affinity precipitates in Hep3B-pcDNA3.1hnRNP A1 cells showed that there was a larger amount of hnRNPA1 protein detectable, compared with parent cells (Supporting Information Fig. S1D), which further demonstrates that hnRNP A1 binds to CD44v6 exon sequences.
HnRNP A1 promotes HCC invasion through regulating CD44v6 expression
Next, we investigated whether CD44v6 plays a crucial role in the promotion of invasiveness of HCC cells by hnRNP A1. Downregulation of CD44v6 by siRNA in HCCLM3 cells led to a significant decrease in the number of invasive cells (Figs. 2c–2e). Meanwhile, overexpression of CD44v6 through transfection with the pcDNA3.1-CD44v6 plasmid in Hep3B cells caused a significant increase in HCC cell invasion (Figs. 2c–2e). These results demonstrate that CD44v6 is involved in HCC cell invasion and metastasis. Then, we found that downregulation of CD44v6 in Hep3B- pcDNA3.1-hnRNP A1 cell was sufficient to abolish the increase in cell invasion produced by hnRNP A1 overexpression (Figs. 2c–2e). Conversely, upregulation of CD44v6 in HCCLM3-shRNA-hnRNPA1 cells completely reversed the inhibition of cell invasion produced by hnRNP A1 knockdown (Figs. 2c–2e). Therefore, we suggest that hnRNP A1 expression promotes HCC cell invasion through its regulation of CD44v6 expression.
HnRNP A1 is upregulated in HCC tissues and coincides with upregulation of CD44v6 expression
Expression of HnRNP A1 at the mRNA level was measured by qRT-PCR analysis of 94 paired HCC samples in Cohort 1. We found that hnRNP A1 was significantly overexpressed in tumors when compared with corresponding peritumoral tissues (p < 0.01; Fig. 3a). Samples from patients with tumor recurrences (57/94) had higher levels of hnRNP A1 than those from patients without recurrences (37/94; p < 0.01; left, Fig. 3c). These results were confirmed by western blot analyses of 12 HCC samples selected from the 94 HCC cases (right, Fig. 3c).
Similar to hnRNP A1, CD44v6 mRNA expression in HCC tissues was significantly higher than that observed in peritumoral tissues (p < 0.05; Fig. 3b). We further investigated the relationship between hnRNP A1 and CD44v6 in HCC samples. Scatter plot analyses revealed a significant positive correlation between hnRNP A1 and CD44v6 mRNA levels in cancerous tissues (Pearson's correlation, n = 94, r = 0.508, p < 0.001; left, Fig. 3d). Similar results were also observed at the protein level in the other cohort of patients (n = 323, r = 0.523, p < 0.001; right, Fig. 3d) by immunohistochemistry.
We investigated the expression of hnRNP A1 and CD44v6 by immunohistochemical staining in a TMA composed of primary tumors from 323 HCC patients in Cohort 2. The immunohistochemistry results showed that hnRNP A1 was primarily located in the nucleus and the cytoplasm, whereas CD44v6 protein was observed in the cell membranes of tumor cells (Figs. 4c and 4d). The hnRNP A1high and CD44v6high groups accounted for 49.8% (161/323) and 50.2% (162/323) of the total patients, respectively. The Pearson's chi-square test indicated that hnRNP A1 was significantly related to tumor encapsulation (p = 0.030; Supporting Information Table 2). However, CD44v6 was significantly related to microvascular invasion (p = 0.029), preoperative γ-glutamyl transferase levels (p = 0.015) and TNM stage (p = 0.006; Supporting Information Table 2).
Expression of hnRNP A1 and combined expression of hnRNP A1 and CD44v6 predict poorer prognosis in HCC patients
By the last follow-up, 54.2% (175/323) of the patients had suffered from recurrence and 51.1% (165/323) had died. The 1-, 3- and 5-year OS rates in the whole cohort were 85.4, 62.2 and 50.7%, respectively. The 1-, 3- and 5-year cumulative recurrence rates were 25.4, 50.2 and 59.7%, respectively. Furthermore, we found that the 1-, 3- and 5-year survival rates of the hnRNP A1low patients were significantly higher than the survival rates of those of the hnRNP A1high group (90.1% vs. 80.7%, 72.2% vs. 52.2% and 61.7% vs. 39.7%, respectively; Fig. 5a). Similarly, hnRNP A1high HCC patients had the poorest prognosis at 1-, 3- and 5-years, with higher cumulative recurrence rates than hnRNP A1low patients (33.7% vs. 17.6%, 62.0% vs. 39.7% and 72.4% vs. 48.7%, respectively; Fig. 5b). Expression of CD44v6 significantly correlated with OS (p < 0.001, hazard ratio (HR) = 2.062) and TTR (p < 0.001, HR = 1.731; Figs. 5c and 5d; Table 1), which was consistent with previously reported results.22
Table 1. Univariate and multivariate analyses of prognostic factors with TTR and OS in HCC (n = 323)
When evaluating the combined effect of hnRNP A1 and CD44v6 on the prognoses of HCC, we found that the 1-, 3- and 5-year OS rates of hnRNP A1low/CD44v6low patients were 93.2, 76.1 and 67.5%, respectively, and were significantly higher than the OS rates for hnRNP A1high/CD44v6high patients (77.8, 47.9 and 35.0%, respectively; Fig. 5e). The 1-, 3- and 5-year cumulative recurrence rates in hnRNP A1low/CD44v6low patients were 14.7, 38.5 and 46.0%, respectively, which were significantly lower than those for hnRNP A1high/CD44v6high patients (35.8, 66.4 and 75.6%, respectively; Fig. 5f). Univariate and multivariate analyses revealed that along with tumor size, encapsulation, microvascular invasion and TNM stage, hnRNP A1 and the coindex (hnRNP A1/CD44v6) were independent prognostic factors for both OS (p < 0.001, HR = 2.009 and p < 0.001, HR = 2.594, respectively) and TTR (p < 0.001, HR = 2.015 and p < 0.001, HR = 2.394, respectively; Table 1).
HnRNP A1 is a versatile hnRNP within the A/B subfamily of proteins.5 Several studies have indicated overexpression of hnRNP A1 in many tumors, including lung cancer,14 breast cancer15 and colon cancer.16 It has been reported that hnRNP A1 may promote cancer cell proliferation and tumor angiogenesis by mediating translation of fibroblast growth factor-2.28 Christofk et al.29 also reported that upregulated transcription of hnRNP A1 (along with hnRNP A2 and hnRNP I) led to a high PKM2/PKM1 ratio, leading to the alternative splicing event required for tumor cell proliferation.30 Moreover, this hnRNP may prevent tumor cells from entering senescence by binding to telomeric DNA sequences.31 Until now, the role of hnRNP A1 in HCC progression has not been clearly defined. In this study, we report that hnRNP A1 is preferentially expressed in highly metastatic HCC cell lines. Moreover, our results revealed that hnRNP A1 overexpression confers invasiveness in HCC cells, while hnRNP A1 downregulation suppresses cell invasion in vitro. The above results suggest that overexpression of hnRNP A1 leads to the metastatic potential of HCCs.
Moreover, it is well known that hnRNP A1 plays a critical role in pre-mRNA splicing and may regulate splicing of CD44 variants.18 CD44 and CD44 variant isoforms are involved in a variety of physiological and pathological processes, including tumor development and progression. They have attracted our attention due to their association with poor clinical outcome for HCC patients.22 In addition, some variant CD44 isoforms have been demonstrated to contribute to metastatic potential in a nonmetastatic rat tumor cell line,32 which was confirmed by a selective knockdown of CD44v4-v7 in the highly metastatic BSp73ASML cell line.33 In the present study, we found that overexpression of hnRNP A1 promoted HCC cell invasion through upregulating CD44v6 expression. First, our results showed that hnRNP A1 level was elevated in MHCC97H and HCCLM3 cells (highly metastatic cell lines), in which expression of CD44v6 was also high. Poorly metastatic cell lines (Hep3B and PLC/PRF/5) expressed low levels of hnRNP A1 and CD44v6. Second, we found that hnRNP A1 binds to CD44v6 exon sequences in vitro. Upregulation of hnRNP A1 expression in HCC cells dramatically increased CD44v6 expression and cell invasion, whereas inhibition of CD44v6 reversed the promotion of HCC cell invasion produced by hnRNP A1 overexpression. CD44v6 can initiate c-Met activation through hepatocyte growth factor (HGF) binding,34 thereby enhancing the motility and invasiveness of the cells. In addition, CD44v6 is associated with poor clinical outcome for HCC patients and is significantly related to vascular invasion,22 suggesting that this isoform may be particularly associated with liver cancer. These data lend further credence to the notion that hnRNP A1 plays a crucial role in HCC invasion and metastasis.
We then investigated the expression of hnRNP A1 and CD44v6 in clinical HCC samples. In 94 tumors (Cohort 1), we found that hnRNP A1 was highly expressed at the mRNA level in HCC tissues when compared with peritumoral tissues. Samples from patients with recurrent HCC exhibited higher levels of hnRNP A1 than those without recurrence. Moreover, our data in the present study revealed a novel relationship between hnRNP A1 and CD44v6. HCC patients with high hnRNP A1 expression tended to have higher levels of CD44v6. In the other independent cohort of patients (Cohort 2), similar results were obtained at the protein level. Because our in vitro invasion assays confirmed that hnRNP A1 promoted HCC invasion by regulating CD44v6 expression, we speculated that CD44v6 overexpression in HCC may be due to the dysregulated expression of hnRNP A1.
Another important finding from this study is the correlation between hnRNP A1 expression and poor prognosis for HCC patients. Our results show that overexpression of hnRNP A1 in HCC patients predicts lower OS rates and higher recurrence rates. Multivariate analyses validated hnRNP A1 as a significant independent predictor for OS and TTR. These observations were similar to previous reports on other malignancies, such as colorectal cancer,17 and strongly indicated that hnRNP A1 may act as both a marker for tumor aggressiveness and a predictor for prognosis in HCC. Our TMA results also confirmed that high CD44v6 expression was a reliable indicator of poor prognosis for HCC patients after resection, when compared with only 107 samples analyzed in a previous study.22 This finding is compatible with that of Endo et al.,22 suggesting this CD44 isoform is significantly related to vascular invasion and ultimately to metastasis. Clinical analyses further supported our in vitro invasion assay finding that CD44v6 may play the crucial role in the promotion of HCC cell invasion by hnRNP A1. We then determine whether coexpression of hnRNP A1 and CD44v6 could act as a predictor of prognosis of HCC. Our results indicate that the subgroup of patients presenting as hnRNP A1high/CD44v6high were more prone to recurrence and suffered worse survival rates after curative resection. Conversely, the HCC patients who expressed low levels of both hnRNP A1 and CD44v6 had the best prognosis.
In conclusion, hnRNP A1 is positively related to the metastatic potential of HCC cells and it promotes HCC cell invasion by regulating CD44v6 expression. The expression levels of hnRNP A1 alone or in combination with CD44v6 in HCC patients are important because they provide not only a predictor for HCC prognosis but also a therapeutic target for future studies.