Objective Despite the recent advances in treatment of colon cancer, the prognosis is unfavourable for patients with distant metastases. The aim of this study was to identify targets for prevention and/or therapy of colon cancer metastasis.
Design CMT93 cells, a murine rectal cancer cell line with poor metastasising activity, were transduced with lentiviral shRNA library and transplanted into the rectum of syngeneic C57BL/6 mice. Genomic DNA was collected from metastatic lesions, and the integrated shRNA were retrieved by PCR for sequencing, followed by identification of the candidate genes targeted by the shRNA.
Results The genome-wide shRNA library screen identified Hnrnpll (heterogeneous nuclear ribonucleoprotein L-like) encoding a pre-mRNA splicing factor as a candidate metastasis suppressor gene. Knockdown of Hnrnpll enhanced matrigel invasion activity of colon cancer cells in vitro, as well as their metastatic ability in vivo. An RNA-immunoprecipitation analysis showed Hnrnpll-binding to Cd44 pre-mRNAs, and the level of Cd44 variable exon 6 (Cd44v6), a poor prognosis marker of colorectal cancer, was increased by knocking down Hnrnpll. A neutralising Cd44v6 antibody suppressed the matrigel invasion ability induced by Hnrnpll knockdown. HNRNPLL expression was downregulated when colon cancer cells were induced to undergo epithelial-mesenchymal transition (EMT). Immunohistochemistry of clinical samples indicated that colorectal cancer cells with low E-cadherin expression at the invasion front exhibited decreased HNRNPLL expression.
Conclusions HNRNPLL is a novel metastasis suppressor of colorectal cancer, and modulates alternative splicing of CD44 during EMT.
- COLORECTAL CANCER
- MOLECULAR MECHANISMS
- COLORECTAL METASTASES
Statistics from Altmetric.com
Significance of this study
What is already known on this subject?
Patients with colorectal cancer with distant metastasis suffer poor prognosis.
Whereas many metastasis promoters have been identified and characterised, the list of metastasis suppressors is still very short.
Although splicing variants of CD44 have been implicated in colorectal cancer progression, the molecular mechanisms and regulators of the CD44 alternative splicing remain unclear.
What are the new findings?
Through an in vivo genome-wide shRNA library screen, HNRNPLL (heterogeneous nuclear ribonucleoprotein L-like) encoding a pre-mRNA splicing factor was identified as a colorectal cancer metastasis suppressor gene.
Knockdown of HNRNPLL endowed colorectal cancer cells with increased invasion ability in vitro and increased metastatic ability in vivo.
Increased expression of CD44 variable exon 6 was partly responsible for the enhanced invasion activity induced by HNRNPLL knockdown in colorectal cancer cells.
HNRNPLL expression was downregulated during epithelial-mesenchymal transition (EMT) of colorectal cancer cells. The link between HNRNPLL and EMT was further suggested by immunohistochemical analysis of clinical samples.
How might it impact on clinical practice in the foreseeable future?
Elucidation of the mechanism underlying the downregulation of HNRNPLL and further identification of its splicing targets might unveil novel prevention and therapeutic targets for colorectal cancer metastasis.
Colorectal cancer is one of the leading causes of death worldwide.1 Although recent advances in early detection, increased awareness and developments in treatment have contributed to better prognosis of patients with colorectal cancer, those with distant metastasis still suffer poor prognosis, necessitating novel strategies for prevention and/or therapy of metastasis. Metastasis is a complex process composed of multiple steps called the metastatic cascade: local invasion, intravasation, survival in the blood flow, extravasation and colonisation.2 These steps are regulated by a number of biomolecules, which can be conceptually divided into metastasis promoters and suppressors. Whereas many metastasis promoters have been identified and characterised, the list of metastasis suppressors is still very short.3
In recent years, the shRNA library has gathered researchers' attention as a useful tool for functional genomics. Several metastasis suppressor genes have been identified by in vitro or in vivo screens using shRNA libraries. Gumireddy et al4 identified KLF17 as a breast cancer metastasis suppressor gene from an in vivo screen, in which 168FARN murine breast cancer cells carrying a genome-wide shRNA library were orthotopically injected into mouse mammary fat pads, followed by collection of lung metastases 7 weeks later. The only reported study that used an in vivo shRNA screen for identifying colon cancer metastasis suppressor genes employed tail-vein injection of CC14 primary human colon cancer cells carrying a genome-wide shRNA library.5 Using this model, Duquet et al5 identified shRNAs targeting TMED3 and SOX12 from metastatic lesions collected 8 weeks after the injection.
Here we present identification of HNRNPLL as a candidate colorectal cancer metastasis suppressor gene, and demonstrate the mechanisms by which HNRNPLL downregulation can contribute to colorectal cancer invasion and metastasis, focusing on its involvement in regulating the alternative splicing of CD44, as well as on its relevance to epithelial-mesenchymal transition (EMT).
Materials and methods
shRNA library screen
C57BL/6N mice were purchased from CLEA Japan (Tokyo, Japan). MISSION pooled mouse Lentiplex shRNA library (Sigma-Aldrich, St. Louis, Missouri, USA) was introduced into CMT93 cells expressing Venus fluorescent protein at a multiplicity of infection of 5, which had been determined from preliminary experiments using pLKO.1-based TurboGFP lentivirus (Sigma-Aldrich). After expansion and selection with puromycin at 5 μg/mL for a week, the cells were suspended in phosphate buffered saline (PBS)(−) and mixed with BD Matrigel Basement Membrane Matrix Growth Factor Reduced (BD Biosciences, San Jose, California, USA) at 1:1 ratio, and then injected into the rectal submucosa at 2×106 cells/mouse. The mice were maintained at a specific pathogen-free facility for around 3 months. After euthanasia, metastatic lesions were visualised and collected under an M205 FA fluorescent stereomicroscope (Leica, Wetzlar, Germany). Genomic DNA was extracted from the collected organ specimens with ISOGEN reagent (Nippon Gene, Tokyo, Japan) and was subjected to PCR using the following primers: CGATACAAGGCTGTTAGAGAGATAATTGGA (forward) and CAAAGTGGATCTCTGCTGTCCCTGTAATAA (reverse). The PCR products were subcloned using a Zero Blunt TOPO PCR Cloning Kit (Thermo Fisher Scientific, Waltham, Massachusetts, USA), and sequenced using a primer with the following sequence: GAAAGTAATAATTTCTTGGGTAGTTTGCAG. Target genes of the shRNAs were identified by matching their sequences with the database provided by Sigma-Aldrich.
CMT93, SW480, T84, HT29 and SW1116 cells were obtained from the American Type Culture Collection (ATCC; Manassas, Virginia, USA), and CaR-1 cells were obtained from the Health Science Research Resources Bank (Osaka, Japan). The human cell lines were authenticated by short tandem repeat profiling. All of the cell lines used were maintained in Dulbecco's modified eagle medium (DMEM) supplemented with 10% fetal bovine serum (FBS, Thermo Fisher Scientific). For inducing EMT, SW480 cells were treated with epidermal growth factor epidermal growth factor (EGF) (Sigma-Aldrich) at 20 ng/mL and basic fibroblast growth factor (FGF) (Sigma-Aldrich) at 10 ng/mL in the absence of FBS, as previously described,6 while CaR-1 cells were treated with transforming growth factor β (TGF-β) (PeproTech, Rocky Hill, New Jersey, USA) at 10 ng/mL. Mesenchymal-epithelial transition (MET) was induced by culturing EMT-induced cells in DMEM supplemented with 10% FBS for at least 7 days. Cell counting was performed using TC20 cell counter (Bio-Rad, Hercules, California, USA). Hepatocyte growth factor (HGF, PeproTech) was added at a final concentration of 100 ng/mL for 12 hours.
Plasmids, transfection and lentiviral transduction
pDisplay-Venus provided by Dr Atsushi Miyawaki (RIKEN Brain Science Institute, Saitama, Japan) was transfected into cells using Lipofectamine LTX (Thermo Fisher Scientific). Lentiviral shRNA constructs were purchased from Sigma-Aldrich (see online supplementary table S1). Lentiviral cDNA expression vectors were prepared by subcloning cDNA into pLEX-MCS (GE Healthcare, Buckinghamshire, England), and were introduced into HEK293T cells using Lipofectamine LTX. Supernatants were collected and used for infecting cells with 8 μg/mL of polybrene (Sigma-Aldrich).
Cells were lysed in radio-immunoprecipitation assay (RIPA) buffer (Thermo Fisher Scientific) containing blends of protease and phosphatase inhibitors (Roche Life Science, Mannheim, Germany), and subjected to sodium dodecyl sulfate (SDS)- polyacrylamide gel electrophoresis (PAGE) followed by transfer onto polyvinylidene fluoride (PVDF) membranes (Bio-Rad). After blocking with Blocking-One (Nacalai Tesque, Kyoto, Japan), the membranes were incubated with primary antibodies and then with horseradish peroxidase (HRP)-conjugated secondary antibody (Southern Biotech, Birmingham, Alabama, USA) at appropriate dilutions. The signals were detected with Immobilon Western Chemiluminescent HRP Substrate (Merck Millipore, Billerica, Massachusetts, USA). Antibodies used are listed in online supplementary table S2.
First strand cDNAs were prepared from total RNA using a High-Capacity cDNA Reverse Transcription Kit with oligo(dT) primers (Thermo Fisher Scientific). Conventional reverse transcription (RT)-PCR was performed with KOD Plus Neo (Toyobo, Osaka, Japan) and primers listed in online supplementary table S3 using a GeneAmp PCR System 9700 (Thermo Fisher Scientific). For quantitative RT-PCR, cDNA samples were mixed with carboxyfluorescein (FAM)-labelled TaqMan Gene Expression Assays and TaqMan Gene Expression Master Mix (Thermo Fisher Scientific), followed by amplification using 7500 Fast Real-Time PCR System (Thermo Fisher Scientific) according to the manufacturer’s protocol. Assay identifications (IDs) for CD44, Cd44 and HNRNPLL are Hs01075861_m1, Mm01277161_m1 and Hs00293181_m1, respectively. The results were calculated by the comparative cycle threshold (CT) method, with relative transcript levels determined as 2−ΔΔCT.
Matrigel invasion activity was assayed using FluoroBlok Cancer Cell Invasion Assay System (BD Biosciences) according to the manufacturer’s instruction. Briefly, Venus-labelled cells (2×105) were seeded in serum-free culture medium onto the Matrigel-coated filters with or without preincubation with 40 μg/mL of anti-CD44v6 neutralising antibody (clone 2F10; R&D, Minneapolis, Minnesota, USA), anti-Cd44v6 neutralising antibody (clone 9A4; GeneTex, Irvine, California, USA) or control IgG1 for 30 min. Culture medium supplemented with 10% FBS was added to the lower part of the chambers. After incubation in 5% CO2 at 37°C for 20 hours, fluorescence intensity of the invaded cells was determined with a Genios microplate reader (Tecan, Männedorf, Switzerland).
RNA-immunoprecipitation was performed using a Magna RNA-immunoprecipitation (RIP) RNA-Binding Protein Immunoprecipitation Kit (Merck Millipore) according to the manufacturer's protocol. Immunoprecipitated RNA was subjected to reverse transcription using a High Capacity cDNA Reverse Transcription Kit (Thermo Fisher Scientific). Random primers were used for the reverse transcription of the immunoprecipitated RNA.
Murine tissue sections were fixed overnight in 4% paraformaldehyde. Paraffin-embedded blocks were sliced at 6 μm thickness and subjected to H&E staining. Human colon cancer tissue sections were obtained from surgical samples in the Aichi Cancer Center Hospital under informed consent. Paraffin sections were deparaffinised, heated in citrate buffer for antigen retrieval, blocked with PBS(−) containing 5% normal goat serum and 0.3% Triton X-100 for 60 min, incubated with primary antibodies overnight at 4°C, and then with Alexa488 or Alexa555-conjugated secondary antibodies (Thermo Fisher Scientific) for 2 hours at room temperature. Hoechst 33342 (Dojindo, Kumamoto, Japan) was used for nuclear staining. Stained sections were observed with a confocal microscope, LSM510 (Carl Zeiss, Jena, Germany). All the investigations were conducted according to the Declaration of Helsinki principles. Staining intensity of HNRNPLL was quantified using the ImageJ software National Center for Biotechnology Information (NCBI).
The statistical significance was assessed by unpaired Student's t-test using the Microsoft Excel software (Microsoft, Redmond, Washington, USA).
Identification of Hnrnpll as a candidate colorectal cancer metastasis suppressor gene
To identify metastasis suppressor genes of colorectal cancer, we developed an in vivo screening system using orthotopic transplantation of mouse colorectal cancer in syngeneic mice (figure 1A). We chose C57BL/icrf-derived CMT93 rectal cancer cells as host cells of the shRNA library, because their high tumorigenicity and poor metastatic ability were suitable for the screening of metastasis suppressors. The cells labelled with Venus fluorescent protein were transduced with whole genome-targeted lentiviral shRNA library, and then injected into the rectal submucosa of C57BL/6 mice. After 12–16 weeks, 14 out of 68 mice that had been transplanted displayed distant metastasis, and the metastatic tumours were collected under a fluorescent microscope. A portion of the collected tumours was used for histological confirmation of adenocarcinoma (figure 1B), and the rest was used for extracting genomic DNA. We identified 47 shRNA sequences from the genomic DNA obtained from 18 metastatic lesions (14 lung metastases, 3 mesenteric lymph node metastases and one liver metastasis) and their targets were determined by matching the shRNA sequences with the database listing their target genes (see online supplementary table S4). Among the identified target genes, we focused on Hnrnpll (heterogeneous nuclear ribonucleoprotein L-like), because shRNA targeting this gene was independently detected in metastatic tumours from two mice. In view of the possibility that the high metastatic ability of the original Hnrnpll shRNA-transduced cells could be due to an off-target effect or clonal variation of the parental cells, CMT93 cells were newly transduced with the same shRNA sequence identified in the screen (sh1) or a distinct shRNA sequence targeting Hnrnpll (sh2), with or without an shRNA-resistant Hnrnpll cDNA for rescuing Hnrnpll expression (figure 1C), and then transplanted into the rectal submucosa of C57BL/6 mice. Both sh2-transduced cells and sh1-transduced cells formed significantly more metastatic nodules than control cells, which could be rescued by the shRNA-resistant Hnrnpll cDNA (figure 1D), demonstrating that the reduced level of Hnrnpll was responsible for enhanced metastasis of CMT93 cells.
Hnrnpll suppresses cell invasion in vitro
We next examined phenotypical alterations that Hnrnpll knockdown induced in cultured colorectal cancer cells. The growth curve analysis showed that Hnrnpll knockdown significantly suppressed proliferation of CMT93 cells, which could be rescued by the shRNA-resistant Hnrnpll (figure 2A). Matrigel invasion assays revealed that Hnrnpll knockdown significantly enhanced invasion activity, which could also be rescued by the shRNA-resistant Hnrnpll (figure 2B). To confirm the results in human colon cancer cells, the human ortholog HNRNPLL was knocked down by two different shRNAs in SW480 and HT29 human colon cancer cells (see figure 2C and online supplementary figure S1A). Both of the HNRNPLL-knocked down SW480 and HT29 cells showed suppressed cell proliferation (see figure 2D and online supplementary figure S1B) and accelerated matrigel invasion (see figure 2E and online supplementary figure S1C), which was rescued by an shRNA-resistant HNRNPLL. These results indicated that HNRNPLL suppressed metastasis by inhibiting invasion.
Knockdown of HNRNPLL increases CD44v6 expression
We next investigated the molecular mechanism by which HNRNPLL knockdown increased the invasion activity. Since HNRNPLL is known as an RNA-binding protein involved in the alternative splicing of precursor messenger RNAs (pre-mRNAs), we addressed the issue by identifying genes whose splice variant expression is affected by knocking down HNRNPLL. A preceding study on HNRNPLL by Oberdoerffer et al7 provided a list of genes displaying significantly altered exon usage upon HNRNPLL knockdown in peripheral CD4+ T cells. Among the listed genes, we were interested in CD44 because of its well established involvement in cancer progression.8 The human CD44 gene consists of 10 standard exons (s1–10) that are constitutively expressed and 9 variable exons (v2–10) that are selectively expressed due to alternative splicing (figure 3A). RT-PCR analysis showed increased expression of exons v3–10 in HNRNPLL-knocked down SW480 cells (see online supplementary figure S2A). To determine whether this effect on alternative splicing was mediated by direct binding of HNRNPLL, we next performed RNA-immunoprecipitation assays using FLAG-HNRNPLL-transduced cells. As shown in online supplementary figure S2B, anti-FLAG antibody immunoprecipitated RNAs containing these exons. Similar analyses using CMT93 cells revealed marked increase in exon v6 expression (see online supplementary figure S2A) and Hnrnpll binding to RNAs containing exon v6 (see online supplementary figure S2B). CD44 exon v6 is clinically known as a marker of poor prognosis in patients with colon cancer.9–11 Quantitative RT-PCR analysis confirmed that a ratio of the expression level of exon v6 to that of exons s8–9 was significantly higher in the HNRNPLL (Hnrnpll)- knocked down cells (see figure 3B and online supplementary figure S2C). In western blot analysis, we found that expression of the exon v6-containing CD44 isoforms was increased in the HNRNPLL-knocked down cells, whereas expression of the standard CD44 isoform (CD44s) was decreased (see figure 3C and online supplementary figure S2D). These results suggest that reduced level of HNRNPLL enhances the CD44v6 expression in colon cancer cells.
To examine the possible involvement of other splicing factors in the increased CD44 variant exon expression in HNRNPLL-knocked down cells, we next determined the expression levels of HNRNPL, HNRNPM and ESRP1 (epithelial splicing regulatory protein 1). HNRNPL is a paralog of HNRNPLL with 69% amino acid homology, and HNRNPM and ESRP1 were recently reported to regulate CD44 exon v6 expression in breast cancer cells. Western blot analysis showed that the levels of these splicing factors were not significantly altered by knocking down HNRNPLL (see online supplementary figure S2E).
Increased CD44 exon v6 expression induced by HNRNPLL knockdown contributes to enhanced invasion ability
We next investigated whether the increased expression of the exon v6-containing isoforms of Cd44 in Hnrnpll-knocked down cells is involved in their enhanced invasion ability. The 2F10 and 9A4 monoclonal antibodies are known to neutralise the exon v6-containing human CD44 and murine Cd44 in living cells, respectively.12–15 Preincubation with these antibodies suppressed matrigel invasion activity of Luc shRNA-transduced cells and HNRNPLL (Hnrnpll)-knocked down cells, indicating that increased CD44 (Cd44) exon v6 expression contributed to enhanced cell invasion (see figure 3D and online supplementary figure S2F). Invasion activity of the 9A4-treated Hnrnpll-knocked down cells was still significantly higher than that of 9A4-treated control cells, suggesting the presence of Cd44 exon v6-independent mechanisms, which remain to be elucidated.
CD44 exon v6-containing isoforms are known to function as coreceptors of the HGF receptor (c-Met). Orian-Rousseau et al16 demonstrated that CD44v6-deficient cancer cells were unable to activate c-Met and that anti-CD44 exon v6 antibody inhibited phosphorylation of c-Met.8 c-Met activation has been implicated in invasion, metastasis and poor prognosis of colon cancer.17–21 The level of HGF-induced c-Met phosphorylation was much higher in HNRNPLL (Hnrnpll)-knocked down colon cancer cells as compared with control cells (see figure 3E and online supplementary figure S2G). These results suggest that the enhanced invasion activity of HNRNPLL knockdown cells may be caused in part by the increased c-Met activation mediated by CD44v6.
HNRNPLL is transcriptionally downregulated during EMT
To investigate the clinical relevance of the invasion/metastasis suppressive effects of HNRNPLL, we first checked genetic alterations of HNRNPLL in colorectal cancer using the catalogue of somatic mutations in cancer (COSMIC), a public database of somatic mutations in cancer (http://cancer.sanger.ac.uk/cosmic/), and found no frequent genetic alterations that may cause loss of HNRNPLL function. Next we immunohistologically assessed the expression levels of HNRNPLL in primary and metastatic colorectal cancer tissues surgically obtained from the same patients (figure 4A). Semiquantitative analysis of the expression level of HNRNPLL using the ImageJ software showed no significant difference between primary and metastatic cancers (figure 4B), suggesting that HNRNPLL may be downregulated transiently or only in a small subset of cancer cells. Therefore, we next tested a possible link between HNRNPLL downregulation and EMT, which can endow cancer cells with transiently enhanced invasion activity. EMT is defined by various phenotypical alterations including fibroblastic morphology, loss of epithelial cell polarity, downregulation of E-cadherin and upregulation of EMT markers such as SNAI1, ZEB1 and N-cadherin.22 The phenotypical alterations during EMT are reversed through MET in metastatic organs, which is required for the subsequent tumour growth.23 ,24 We induced EMT in SW480 cells using a combination of EGF and basic FGF,6 and in CaR-1 cells using TGF-β. MET was induced by removing these factors and culturing the cells in culture media supplemented with 10% FBS. Successful induction of EMT was confirmed by fibroblastic morphology (see figure 4C and online supplementary figure S3A), transcriptional upregulation of EMT markers, and transcriptional downregulation of E-cadherin (see figure 4D and online supplementary figure S3B), which were all reversed by subsequent MET induction. Expression of HNRNPLL was decreased both at the mRNA and protein levels in EMT-induced cells and was recovered by MET induction (see figure 4E and online supplementary figure S3C), whereas the expression levels of HNRNPL, HNRNPM and ESRP1 showed no significant alteration (see online supplementary figure S3D). Real-time RT-PCR analysis demonstrated that EMT induction increased a ratio of the expression level of exon v6 to that of exon s8–9, which was cancelled by MET induction or by introducing HNRNPLL cDNA (see figure 4F and online supplementary figure S3E). Increased expression of exon v6-containing CD44 isoforms in EMT-induced cells was confirmed by western blot, which was also cancelled by MET induction or overexpression of HNRNPLL (see figure 4G and online supplementary figure S3F). These data indicate that HNRNPLL expression is decreased during EMT, leading to increased CD44 variant isoform expression.
Finally, we assessed whether the link between HNRNPLL and EMT could be observed in clinical samples. Immunostaining with antibodies for HNRNPLL and an epithelial marker cytokeratin 18 (CK18) showed markedly reduced level of HNRNPLL expression in a small portion of cancer cells at the invasion front compared with that in cancer cells in the central region of the tumour mass (figure 4H). Furthermore, immunostaining with HNRNPLL and E-cadherin antibodies showed that cancer cells at the invasion front showing low E-cadherin expression exhibit prominently reduced HNRNPLL expression compared with those in the central region (figure 4I). Taken together, these data suggest that HNRNPLL downregulation is associated with EMT in vivo.
More than 90% of human genes are estimated to produce splice variants via alternative splicing,25 ,26 60% of which are translated into distinct protein isoforms.27 Alternative splicing is regulated by interactions between cis-regulatory pre-mRNA sequences and trans-regulatory splicing factors that bind to these cis-regulatory elements.28 ,29 The functional output of the splicing factors is determined by competition between splicing factors promoting exon inclusion and those promoting exon skipping. Whereas alternative splicing is tightly regulated in normal cells, dysregulated expression or activity of the splicing factors, or mutations in the cis-regulatory elements can lead to the production of proteins with aberrant functions in cancer cells.30–32 Previous reports implicated dysregulated expression of splicing factors in colon cancer,33–38 and functional involvement of the dysregulation in cancer progression has been partially elucidated. For example, Ghigna et al39 demonstrated an intriguing mechanism by which the splicing factor SF2/alternative splicing factor (ASF) regulates cell motility. They showed that SF2/ASF promotes the skipping of exon 11 of RON, resulting in the accumulation of a constitutively active form of RON, which is known to enhance cell dissociation, motility and matrix invasion. However, few reports have demonstrated the involvement of splicing factors in cancer metastasis in vivo. In the present study, we have provided in vivo evidence that HNRNPLL suppresses colon cancer metastasis.
We demonstrated that HNRNPLL binds to CD44 pre-mRNA and that expression of the variable exons was increased by knocking down HNRNPLL in colon cancer cells (see figure 3B and online supplementary figure S2C). This finding is in line with the previous reports that showed increased expression of CD44 variable exons in HNRNPLL-knocked down CD4+ T cells7 and plasma cells.40 Hnrnpll bound predominantly to Cd44 pre-mRNAs containing the exon v6 in CMT93 cells (see online supplementary figure S2B), and HNRNPLL knockdown induced expression of CD44v6 (figure 3B,C), a known marker of the poor prognosis of patients with colorectal cancer.9–11 Suppressive effects of the neutralising anti-CD44v6 antibody clearly indicated the involvement of CD44v6 in enhanced matrigel invasion activity caused by HNRNPLL knockdown (figure 3D). However, involvement of other variable exons remains to be elucidated, especially because the transcript levels of other CD44 variable exons were also increased by knocking down HNRNPLL in SW480 cells (see online supplementary figure S2A). The roles of CD44 variant isoforms other than CD44v6 in cancer invasion and metastasis have been documented for several types of cancer.8 ,41 Notably, Saya et al reported that a subpopulation of 4T1 breast cancer cells expressing CD44v8–10 showed high lung-metastatic ability in an orthotopic transplantation model, in a manner dependent on interaction between the cystine transporter cystine/glutamate transporter (xCT) and CD44v8–10.42 ,43
Whereas CD44v6 has been implicated in the poor prognosis of patients with colorectal cancer the molecular mechanism underlying the CD44v6 expression in cancer cells has been unclear. A role of epithelial splicing regulatory proteins (ESRPs) in the inclusion of CD44 variable exons has been well documented in studies using immortalised normal epithelial cells. Namely, Warzecha et al44 showed that knockdown of ESRPs caused decreased expression of CD44v6–10 and CD44v8–10 expression in immortalised human prostatic epithelial cells. More recently, HNRNPM, a heterogeneous nuclear ribonucleoprotein (hnRNP) family protein, was shown to compete with ESRP1 for binding to the cis-regulatory pre-mRNA element of CD44 and to promote skipping of the variable exons in immortalised human mammary epithelial cells.45 Our data showed comparable levels of ESRP1 expression in HNRNPLL-knockdown cells and in EMT-induced cells compared with the control cells (see online supplementary figures S2E and S3D), suggesting a possibility that downregulation of HNRNPLL might result in the predominant occupancy of the cis-regulatory element of the CD44 pre-mRNAs by ESRP1, although involvement of other splicing factors in the variable exon inclusion remains possible.
In this study, we have demonstrated that HNRNPLL negatively regulates invasion and metastasis of colorectal cancers by controlling the alternative splicing of CD44. It is possible that downregulation of HNRNPLL contributes to cancer progression through other unidentified mechanisms as well. A positive link between HNRNPLL and cell proliferation has been suggested by HNRNPLL knockdown experiment in vitro (see figure 2A,D, and online supplementary figure S1B). Intriguingly, EMT induction in colorectal cancer cells reduced the level of HNRNPLL, which could be restored by inducing MET (see figure 4E and online supplementary figure S3C). This behaviour of HNRNPLL was paralleled with cell proliferation, which was suppressed during EMT and reactivated during MET. These findings may suggest a possible role of HNRNPLL as a regulator of cell proliferation during EMT and MET, and provide a clue for future studies.
It has been well documented that the EMT process involves deregulated pre-mRNA alternative splicing programme of various genes.46 ,47 Recently, Todaro et al12 demonstrated that HGF, osteopontin and stromal-derived factor 1α can induce expression of both CD44v6 and EMT-related genes in colorectal cancer stem cells. This report is in line with our data showing that CD44v6 is increased upon EMT induction, though the responsible splicing factors may be different. In clear contrast, EMT is known to induce CD44s expression in breast epithelial cells,48 which has been explained by competition of ESRP1 by HNRNPM as mentioned above.45 The link between CD44v6 and EMT may thus differ among cancers of different origins or between normal and cancer cells.
In summary, we have demonstrated that HNRNPLL is a colon cancer metastasis suppressor that inhibits invasion partly through decreased CD44v6 expression (figure 4J). Downregulation of HNRNPLL was associated with EMT both in colon cancer cell lines and clinical samples. Elucidation of the mechanism underlying the downregulation of HNRNPLL and further identification of its splicing targets may unveil novel prevention and therapeutic targets for colorectal cancer metastasis.
The authors thank Dr Atsushi Miyawaki (RIKEN Brain Science Institute, Saitama, Japan) for providing a pDisplay-Venus construct.
Correction notice This article has been corrected since it published Online First. Figure 1 has been updated to include labels C and D.
Contributors MA and KS designed the study, analysed and interpreted the data, and wrote the manuscript. KS carried out the experiments. ES, KKi, KKo, YS and YY provided clinical samples. All authors have seen and approved the final version of the manuscript.
Funding This work was supported in part by Grant-in-Aid for Challenging Exploratory Research (24650627) and Grant-in-Aid for Scientific Research B (26290045) from the Japan Society for the Promotion of Science, grants from the Naito Foundation, Suzuken Memorial Foundation, Foundation for Promotion of Cancer Research, and Nagono Medical Foundation, Japan.
Competing interests None declared.
Ethics approval All the animal experiments were conducted according to the protocol approved by the Animal Care and Use Committee of Aichi Cancer Center Research Institute. This study was carried out in accordance with the institutional ethics committee of Aichi Cancer Center.
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.