IGFBP-rP1 suppresses epithelial–mesenchymal transition and metastasis in colorectal cancer

Epithelial–mesenchymal transition (EMT) was initially recognized during organogenesis and has recently been reported to be involved in promoting cancer invasion and metastasis. Cooperation of transforming growth factor-β (TGF-β) and other signaling pathways, such as Ras and Wnt, is essential to inducing EMT, but the molecular mechanisms remain to be fully determined. Here, we reported that insulin-like growth factor binding protein-related protein 1 (IGFBP-rP1), a potential tumor suppressor, controls EMT in colorectal cancer progression. We revealed the inhibitory role of IGFBP-rP1 through analyses of clinical colorectal cancer samples and various EMT and metastasis models in vitro and in vivo. Moreover, we demonstrated that IGFBP-rP1 suppresses EMT and tumor metastasis by repressing TGF-β-mediated EMT through the Smad signaling cascade. These data establish that IGFBP-rP1 functions as a suppressor of EMT and metastasis in colorectal cancer.

Colorectal cancer (CRC) is one of the most common human malignancies, accounting for nearly 1 in 10 cancer deaths. 1 More than 90% of these deaths are attributed to local invasion and distant metastasis. 2 Therefore, it is of great importance to understand the underlying biological processes. Compelling evidence suggests that epithelial-mesenchymal transition (EMT) is a major mechanism of cancer cell metastasis by endowing cells with an invasive phenotype. 3,4 EMT is a highly conserved cellular process that induces epithelial cells to dissociate from the primary tumor, invade surrounding tissues, and migrate to distant organs. 5,6 During EMT, cells lose their epithelial features of polarity and adhesion, gain mesenchymal properties, and exhibit increased motility. 7 Biochemically, epithelial markers such as the adherens junction molecules E-cadherin and zonula occludens-1 (ZO-1) are repressed and mesenchymal markers such as vimentin and N-cadherin are upregulated, together with increased matrix metalloproteinase-9 secretion to degrade the extracellular matrix. 8,9 In addition, transcription factors such as members of the Snail (SNAIL and SLUG), ZEB (ZEB1 and SIP1), and bHLH (E47 and TWIST) families also contribute to the occurrence of EMT by binding to the E-boxes at the E-cadherin promoter to silence its expression. 10 A variety of extracellular signals can trigger EMT, and transforming growth factor β (TGF-β) is considered to be a potent inducer of EMT in an autocrine or paracrine manner. 11 It is known that TGF-β has dual roles during tumor progression. It inhibits tumor growth in the early phase, but paradoxically, in the late phase it acts as a tumor promoter by enhancing EMT and promoting tumor cell invasion in primary carcinomas. 12,13 TGF-β binds to its receptors and activates Smad2 and Smad3, then phosphorylated SMAD2/3 partner with Smad4 to translocate into the nucleus where SMAD complexes control target gene transcription, including EMT-and motility-related genes. 14 Insulin-like growth factor-binding protein-related protein 1 (IGFBP-rP1, also known as IGFBP7) is a secreted factor belonging to the IGFBP family. Mounting evidence indicates that IGFBP-rP1 possesses tumor-suppressor activity in various tumor types. [15][16][17][18][19] Indeed, our previous studies clearly revealed that IGFBP-rP1 inhibits cell proliferation, decreases soft agar colony formation, and induces apoptosis in CRC cell lines. Moreover, the abundance of IGFBP-rP1 is inversely correlated with a poor prognosis. 20 In addition, when we restored the expression of IGFBP-rP1 with 5-aza-2′-deoxycytidine, the migration and invasion abilities of CRC cells decreased. 21 In aggregate, although IGFBP-rP1 has been confirmed as a master regulator of cancer-cell fate, its functional role in EMT and metastasis remains largely unknown.
Here, we used patient samples and several CRC cell models to reveal a unique role for IGFBP-rP1 in inhibiting EMT and metastasis via a TGF-β/smad-dependent pathway.

Results
Negative correlation of IGFBP-rP1 detected by immunohistochemical staining with EMT features in clinical specimens. Considering the proposed correlation between IGFBP-rP1 and clinical outcome, we extended our analysis to the relationship between IGFBP-rP1 expression and clinicopathological parameters. The staining intensity of IGFBP-rP1 was inversely correlated with histological grade (r = − 0.161, P = 0.017), lymph-node metastasis (r = − 0.174, P = 0.011), and TNM stage (r = − 0.149, P = 0.028) (Table 1). Then, we investigated whether IGFBP-rP1 was also associated with EMT marker expression in our clinical cohort. Immunostaining of IGFBP-rP1 and the epithelial markers E-cadherin (r = 0.24, P = 0.014) and β-catenin (r = 0.28, P = 0.002) revealed a positive correlation in membrane expression. Conversely, we found a negative correlation between IGFBP-rP1 and fibronectin expression (r = − 0.196, P = 0.030) ( Table 2). Overall, our clinical results imply that IGFBP-rP1, as a suppressor of CRC progression, may function to oppose the EMT program.
Overexpression of IGFBP-rP1 hinders EMT and cell migration in CRC cells. To directly investigate the potential role of IGFBP-rP1, we stably overexpressed it in CRC cell lines and evaluated its effect on EMT. We found that IGFBP-rP1 overexpression increased the expression levels of the epithelial cell marker E-cadherin and blocked that of the mesenchymal markers N-cadherin, vimentin, and MMP9 in SW620 and CW2 cells (Figures 1a and b). As IGFBP-rP1 is a secreted protein, we sought to determine whether exogenous IGFBP-rP1 similarly changed EMT marker protein levels. In agreement with the changes in IGFBP-rP1-overexpressing cells, addition of recombinant human IGFBP-rP1 protein (rhIGFBP-rP1) resulted in high E-cadherin and low mesenchymal markers levels in SW620 cells (Figure 1c). Meanwhile, upregulated E-cadherin accumulated in the cell-to-cell junctions, and nuclear β-catenin re-localized to the membrane and cytoplasm of SW620 cells, as expected ( Figure 1d). Furthermore, SW620-IGFBP-rP1 cells exhibited organized F-actin fibers as well as reduced stress fibers and lamellipodia at the periphery compared with the control cells and the parental cells ( Figure 1d). Functionally, overexpression of IGFBP-rP1 decreased the motility ability (Figures 1e and f) of SW620 cells. Collectively, these data indicate that IGFBP-rP1 is a negative regulation of EMT in CRC.
Knockdown of IGFBP-rP1 induces EMT and promotes invasion in CRC cells. We next determined whether knockdown of IGFBP-rP1 reversed its inhibitory role in EMT. We used short-hairpin RNA (shRNA) to stably knock down IGFBP-rP1 in SW480 cells, which have detectable endogenous IGFBP-rP1 expression (Figure 2a). To rule out clone-to-clone variations, two clones named SW480-sh2 and SW480-sh3 were selected. The knockdown efficiency of IGFBP-rP1 was further confirmed at the mRNA and protein levels ( Figure 2a). IGFBP-rP1 downregulation induced EMT, leading to augmented N-cadherin, vimentin, and MMP9 protein levels as well as a significant decrease in ZO-1 expression (Figure 2b). Nuclear expression of β-catenin detected by western blot exhibited increased nuclear β-catenin levels in two SW480/shIGFBP-rP1 cell lines. Similarly, immunofluorescence showed that knockdown of IGFBP-rP1 results in relocalization of β-catenin from adherens junctions of the membrane to cytoplasm and nucleus ( Figure 2c). Functionally, knockdown of IGFBP-rP1 dramatically increased the migration ability of SW480 cells (Figure 2d). In a complementary experiment, we determined whether the re-expression of IGFBP-rP1 rescued the EMT process induced by silencing it, and found that addition of rhIGFBP-rP1 protein restored ZO-1 expression as well as rescued the increased Vimentin, N-cadherin, and MMP9 expression triggered by IGFBP-rP1 knockdown in SW480 cells ( Figure 2e). Consistently, the elevated transwell migratory cells were also reduced due to the restoration of IGFBP-rP1 ( Figure 2f). Taken together, the functional role of IGFBP-rP1 in inhibiting EMT in CRC is further strengthened.
IGFBP-rP1 is involved in tumorigenicity of CRC cells in nude mice. To explore the biological role of IGFBP-rP1 in tumor formation in vivo, we inoculated CRC cell lines with and without IGFBP-rP1 expression as xenografts in nude mice. In accord with the finding that IGFBP-rP1 inhibited cell proliferation in vitro, the xenografted tumors in the   Well-established EMT markers were analyzed by western blot after IGFBP-rP1 transfection in SW620 and CW2 cells. E-cadherin expression was markedly induced and the mesenchymal markers N-cadherin, vimentin, and MMP9 were decreased in IGFBP-rP1-transfected cells (-rP1) compared with control vector-transfected (-ctrl) and parental cells (-blk). Equal loading was confirmed by GAPDH. (c) SW620 cells were treated with rhIGFBP-rP1 protein (4 μg/ ml) for 48 h and the expression of EMT markers was detected by western blot. (d) Immunofluorescence images of SW620 cells stained for E-cadherin, β-catenin, and F-actin. The images were taken at × 630 (for E-cadherin and β-catenin ) and × 1000 (for F-actin). Arrow: Lamellipodia and microspike formation. DAPI and β-catenin staining was showed seperately and then the merged images were showed. (e and f) Wound-healing and transwell motility assays for SW620-ctrl and SW620-rP1 cells. The motility was drastically decreased in IGFBP-rP1 transfected SW620 cells (×100). Cell motility detemined by wound-healing assay was quantified as an inverse ratio of gap distance (GD) at 48 h relative to that at 0 h. *Po 0.05. The data were expressed as mean+S.D. of three independent experiments IGFBP-rP1 inhibits colorectal cancer metastasis. Considering the role of IGFBP-rP1 in controlling EMT and migration in vitro, we extended our study to an IGFBP-rP1 overexpressing SW620 cells led to a significant reduction of lung metastasis compared with scramble control cells, as shown in the gross images (Figure 4a, left) and hematoxylin and eosin (H&E) staining (Figure 4a, right). Decreased numbers of lung metastatic nodules were quantified as shown in Figure 4b. Interestingly, we also observed less probability of metastatic lesion formation and decreased nodule numbers in the pleura of IGFBP-rP1 overexpressing group when compared with control group (Figures 4c and d). H&E staining indicated tumor cells invaded the pleural (Figure 4e, left top) and even the costal cartilage (Figure 4e, left bottom) in the group without IGFBP-rP1 expression. Therefore, these analyses again show that IGFBP-rP1 strongly opposes colorectal cancer distant metastasis.
Effect of IGFBP-rP1 on TGF-β receptor and its downstream mediators. TGF-β is considered as a very potent inducer of EMT and metastasis. Canonical Smad signaling, which is activated by TGF-β, has critical roles during the induction of EMT. 22,23 Previous microarray results revealed a relationship between IGFBP-rP1 and TGF-β signaling, especially Smad3. Therefore, we explored whether the effect of IGFBP-rP1 on EMT is mediated by TGF-β signaling. As expected, IGFBP-rP1 overexpression inactivated both TGF-β R I and R II in SW620 cells. Accordingly, phosphorylation of smad2 and smad3 was also decreased in the IGFBP-rP1 overexpressing cells as compared with the control cells (Figure 5a). In support of this, IGFBP-rP1 knockdown in SW480 cells increased the levels of TGF-β receptor and its main downstream effectors, phospho-Smad2/3 (Figure 5b). To further validate whether IGFBP-rP1 modulated the TGF-β/ Smad signaling pathway, we used SB431542, which is a selective inhibitor of TGF-β/ALK5/Smad2 signaling. Western blots confirmed the inhibition of TGF-β R II and smad2/3 phosphorylation by SB431542 in two clones of IGFBP-rP1 knockdown SW480 cells. SB431542 rescued the EMT status induced by IGFBP-rP1 knockdown (Figure 5c). Overall, our data support that IGFBP-rP1 upregulates TGF-β receptor expression, with subsequent activation of TGF-β receptor pathways.
IGFBP-rP1 reverses TGF-β1-mediated EMT by attenuating Smad pathways. The relevance of the pathway was next examined in HT29 colorectal cancer cells. Here, addition of TGF-β1 induced a time-dependent increase of phosphorylated Smad2/3 protein levels, with 24 h TGF-β having the maximal activation effect in HT29 cells (Figure 6a). In response to TGF-β1 stimulation for 24 h, although the level of E-cadherin showed no apparent change, HT29 cells underwent EMT as determined by downregulation of the epithelial marker ZO-1 with concomitant upregulation of MMP9 and Snail (Figure 6b) as well as increased cell migration (Figure 6c). These data suggested that we had not only established the classical model of TGF-β1-induced EMT, but that it also activated the TGF-β/Smad cascade in HT29 cells. To further evaluate whether IGFBP-rP1 reversed TGF-β1-induced EMT-related changes, we performed a rescue experiment by adding exogenous rhIGFBP-rP1 protein. Consistent with our findings in the other cell lines, treatment with rhIGFBP-rP1 alone decreased TGF-β receptor and phospho-Smad2/3 expression levels as well as inhibited EMT program (Figure 6d, lane 2). This was accompanied by an obvious decrease of HT29 cells invasive ability (Figure 6e, lane 2). Treatment with TGF-β alone activated EMT program, TGF-β receptors, and its downstream molecules (Figure 6d, lane 3). Of note, exogenous rhIGFBP-rP1 protein decreased the activation of TGF-β signaling and restored the upregulation of TGF-β1-induced EMT-related markers (Figure 6d, lane 4 compare with lane 3). Accordingly, rhIGFBP-rP1 decreased the promotion of migration triggered by TGF-β1 (Figure 6e). These data collectively provide evidence that IGFBP-rP1 suppresses EMT by attenuation of TGF-β/Smad signaling in colorectal cancer.

Discussion
IGFBP-rP1 is commonly regarded as a tumor suppressor in CRC, 20 prostate cancer, 15 breast cancer, 16 lung cancer, 17 melanoma, 18 and thyroid cancer. 19 Previous work from our laboratory suggested that IGFBP-rP1 expression in CRC tissue is correlated with a favorable prognosis. 20 Although the prognosis relies heavily on tumor recurrence and distant metastasis after surgical resection, little was known about the function of IGFBP-rP1 in tumor cell motility.
Although not well investigated, IGFBP-rP1 seems to have a role in cell migration. Downregulation of its expression promotes invasion by hepatocellular cancer cells. 24 IGFBP-rP1-overexpressing cells in breast cancer show reduced migration, 25 and IGFBP-rP1 is also correlated with early metastasis of CRC. 26,27 We previously demonstrated that the restoration of IGFBP-rP1 by 5-aza-dC inhibits migration and invasion in CRC cell lines. Consistent with these results, the present study demonstrated that depletion of IGFBP-rP1 promoted the migration of SW480 cells, whereas its overexpression inhibited CRC cell migration in vitro and distant metastasis in vivo. As expected, the TGF-β-induced migration was reversed by rhIGFBP-rP1. Moreover, downregulation of IGFBP-rP1 occurred in patients with positive lymph-node metastasis. Overall, IGFBP-rP1 showed an inhibitory effect on cell motility and tumor metastasis in CRC.
EMT is a critical process for cancer metastasis. Until recently, IGFBP-rP1 had not been associated with EMT regulation. In this report, we demonstrated the role of IGFBP-rP1 in regulating EMT. Our study choosed SW620 and CW2 cell lines, the two colon cancer cell lines without endogenous IGFBP-rP1 expression, to observe the transfection of IGFBP-rP1 cDNA on the EMT status of the cells. SW480 cells that were used shRNA to stably knock down express endogenous IGFBP-rP1. HT29 is a widely used colorectal cell line for a classical model system: TGF1-induced EMT. We used a combination of in vitro cell-line models and in vivo xenograft nude mice to show that gain of IGFBP-rP1 facilitated the acquisition of numerous epithelial features and the absence of mesenchymal characteristics. Conversely, loss of IGFBP-rP1 induced apparent EMT. Although the decrease of IGFBP-rP1-mediated E-cadherin expression is not obvious in HT29 cells, we found that E-cadherin was induced in overexpressing SW620 and CW2 cells compared with the control vector transfected cells and the parental cells. Moreover, another epithelial marker ZO-1 in SW480 and HT29 cells was investigated. E-cadherin that is a transmembrane protein represents a well-characterized molecular marker of the epithelial phenotype. Numerous studies have demonstrated that abolishment of E-cadherin expression or function could promote cell invasiveness, which is a critical event in EMT. However, E-cadherin losing is not a unique pivotal event during EMT. The forcing expression of E-cadherin was not sufficient to restore epithelial characteristic in murine spindle carcinoma cells 28 as well as failed to reverse the fibroblastic morphology of MDCK-Snail cells. 29 Furthermore, reduced E-cadherin expression triggered by overexpression of Slug and Twist failed to induce full EMT and enhance metastasis in CRC cells. 30 More importantly, clinical data analyses also indicated that IGFBP-rP1 is inversely associated with the classical EMT markers. These data indicate that IGFBP-rP1 may be a negative regulator to monitor the EMT program.
IGFBP-rP1 is known to bind to IGF-1 and IGF-2 with 100fold lower affinity than IGFBP1 to IGFBP6, 31 so it functions in both an IGF-dependent and an IGF-independent manner. 32 IGFBP-rP1 may be involved in the TGF-β signaling pathway through binding to activin, 33 a member of the TGF-β superfamily. IGFBP-rP1 in prostate epithelial cells is also modulated by TGF-β1. 34 Previous microarray analysis provided evidence that IGFBP-rP1 has a close relationship with TGF-β, and further influences Smad3 expression. 35 Therefore, whether IGFBP-rP1 participates in TGF-β signaling by a Smaddependent or Smad-independent manner remains to be confirmed. Interestingly, we found that TGF-β signaling inhibition was due to overexpression of IGFBP-rP1 levels in the SW620 cells. Reversely, knockdown of IGFBP-rP1 in SW480 cells activated the TGF-β signaling pathway. To strengthen our data and provide further insight, exogenous rhIGFBP-rP1 protein decreased the absence of TGF-β receptor and its downstream protein levels activated by TGF-β1 in HT29 cells. EMT as a potential target for IGFBP-rP1 via regulation of TGF-β /Smad signaling is a unique finding based on our studies. In the study, we delineate a pathway in which IGFBP-rP1 attenuates the TGF-β signaling pathway to suppress EMT, thereby decreasing migration.
TGF-β exerts a potent effect on EMT initiation and tumor metastasis. It has been reported that TGF-β induces the expression of transcription factors such as Snail, Slug, Zeb, and Twist. 10 Of these, Snail is greatly upregulated through the Smad pathway. 36 During TGF-β-mediated EMT, the Snail-Smad3/4 transcriptional complex functions as a transcriptional repressor. 37 HMGA2 and Smads bind to the Snail promoter and acquire the EMT phenotype. 38 These compelling data suggest that Snail is regulated by TGF-β/Smad signaling from the tumor microenvironment. In line with these observations, Snail was clearly downregulated by either IGFBP-rP1 (Supplementary Figure 1B) or the addition of rhIGFBP-rP1 to TGF-β1-stimulated cells (Figure 6d) and was upregulated by IGFBP-rP1 knockdown (Supplementary Figure 1C). Therefore, the potential involvement of the IGFBP-rP1-Smad-Snail axis in CRC warrants further investigation.
In brief summary, our analyses have concluded that IGFBP-rP1, through the inhibition of TGF-β/Smad signaling, endows cells with properties linked to metastasis, including a reverse EMT process and decreased motility.
Cell culture. The human CRC cell lines SW620 and SW480 were purchased from the American Type Culture Collection (Manassas, VA, USA). The human CRC cell lines HT29 and CW2 were obtained from the Shanghai Institute of Cell Biology, Chinese Academy of Science (Shanghai, China). All cell lines were maintained in RPMI-1640 supplemented with 10% fetal bovine serum (FBS) at 37°C, in a humidified atmosphere of 5% CO 2 . The medium was changed every 2 days. The stably-transfected cell lines SW620-ctrl/rP1 and CW2-ctrl/rP1 had been established previously. 20,39 Transfection. The IGFBP-rP1-shRNA expression vector kit was from Shanghai Genepharma Co., Ltd. (Shanghai, China), and the target sequences were TAGAGGAGATACCAGCACCCAGCCA and TGCTGATGCTGAAGCCTGTCCTTGG. All transfection experiments were performed using Lipofectamine 2000 (Invitrogen) according to the manufacturer's protocol. Stably-transfected cell clones were selected using G418 (500 μg/ml).
RNA isolation and real-time RT-PCR. Total RNA was extracted from cells using TRIzol reagent (Invitrogen) according to the manufacturer's protocol. One microgram of total RNA was reverse transcribed using a PrimeScript RT reagent kit (Perfect Real Time) and real-time PCR was performed using SYBR Premix Ex Taq (Tli RNaseH Plus), both from TaKaRa Biotechnology Co., Ltd (Dalian, China). The 10-μl reactions were performed in 384-well plates on the ABI Prism 7900 sequence detection system (Applied Biosystems, Foster City, CA, USA). The settings were initial denaturation at 95°C for 10 s, followed by 40 cycles of 95°C for 5 s and 60°C for 30 s, and elongation at 72°C for 40 s. For the normalized expression ratios of genes, the △△Ct method was used. The primer sequences used were as follows: GAPDH, forward: 5′-ACCACAGTCCATGCCATCAC-3′ and reverse: 5′-TCCACCA CCCTGTTGCTGTA-3′; IGFBP-rP1, forward: 5′-CACTGGTGCCCAGGTGTACT-3′ and reverse: 5′-TTGGATGCATGGCACTCATA-3′.
Western blot analysis. Concentrated cell culture medium and total cell proteins were prepared as described previously. Western blot analyses were performed using standard procedures. Proteins were separated by SDS-PAGE and transferred onto nitrocellulose membrane. After blocking in 5% non-fat milk in TBS-Tween 20 for 30 min at room temperature, membranes were incubated with primary antibodies overnight at 4°C, and then incubated with secondary antibodies for 1 h at room temperature. The blots were detected using the Odyssey system (Li-COR Recombinant human IGFBP-rP1 protein treatment. Recombinant human IGFBP-rP1 (rhIGFBP-rP1) protein stimulation assays were done as described previously. Cells were seeded into 6-well plates (3-5 × 10 5 cells/plate). After attachment for 24 h, the cells were exposed to fresh culture medium containing different concentrations of rhIGFBP-rP1 protein (SW620 cells, 4 μg/ml; HT29 cells, 1 μg/ml). After 48 h, the cells were collected and protein extraction performed.
Recombinant human TGF-β1 treatment. HT29 cells were grown to 70% confluence and exposed to serum-free medium. After starvation for 24 h, the cells were incubated with recombinant human TGF-β1 at 10ng/ml for the indicated periods. EMT markers were analyzed by western blot.
Phalloidin staining. Cells were grown on glass slides, washed three times with PBS, and fixed in 100% methanol for 10 min at − 20°C. They were washed again with PBS, permeabilized with 0.1% Triton X-100 in PBS for 15 min, and blocked in 10% bovine serum for 30 min at room temperature. After further washing with PBS, the cells were incubated with primary antibody Alexa-Fluor 488 phalloidin (1 : 1000; Invitrogen) for 2 h in the dark at room temperature, and washed with PBS. The nuclei were stained with 4',6-diamidino-2-phenylindole (1 : 10000; Invitrogen) for 30 min. Confocal images were captured using an LSM 510 META confocal microscope (Carl Zeiss, Oberkochen, Germany).
Wound-healing assay. Cells were grown to~80% confluence in 6-well plates. A straight scratch was made gently across the center of the well using a yellow Eppendorf tip. Detached cells were rinsed away with PBS and serum-free medium was added. Cell migration to the scratch was monitored at the indicated Inhibition of EMT and metastasis by IGFBP-rP1 S Zhu et al time periods in the same area. The images were captured at × 100 magnification (Nikon TS100, Tokyo, Japan). The experiments were repeated three times.
Transwell migration assay. Boyden chambers (8 μm pores; Costar, Corning, NK, USA) were used for transwell assays to measure cell migration. Cells (1 × 10 5 ) in 0.2 ml serum-free medium were seeded into the upper chamber, while 0.6 ml medium containing 10% FBS was added to the lower chamber. The cells were then incubated for 48 h at 37°C in a humidified atmosphere of 5% CO 2 .
After the assay was completed, the cells on the upper side of the membrane were removed with a wet cotton swab and those on the lower side were fixed in 95% ethanol and stained with crystal violet at room temperature for 10 min. The membrane was gently removed and mounted on a glass slide. The images were captured at × 100 magnification. The number of cells was counted from four random fields using ImageJ (NIH, Bethesda, MD, USA). Each experiment was performed at least three times.
Tumor formation assay. BALB/c-nu/nu mice (5 weeks old, male) were purchased from Shanghai Laboratory Animal Center. All experimental procedures were approved by the Animal Care and Use Committee of Zhejiang Chinese Medical University. Cells were harvested and suspended in PBS. Control vector overexpressing SW620 cells and IGFBP-rP1 overexpressing SW620 cells (3 × 10 6 cells; n = 4 mice each); control shRNA SW480 cells, IGFBP-rP1 knockdown SW480 cells (8 × 10 6 cells; n = 4 mice each) in 0.1 ml PBS were injected subcutaneously into the nape of mice. The length and width of the tumor were measured using digital calipers twice a week until the end point. Tumor volume was calculated using the following formula: π/6 × (length) × (width) 2 . After 6-7 weeks, the mice were euthanized, and the tumors were excised, weighed, and then used for RNA and protein extraction.
Tumor metastasis assays. NOD/SCID mice (5 weeks old, male) were purchased from Shanghai Laboratory Animal Center. To determine the distant metastatic potential of cancer cells in vivo, 5 × 10 6 cells (two groups including SW620-vector cells and SW620-IGFBP-rP1 cells) resuspended in 100 μl PBS were injected into the tail vein of severe combined immunodeficient NOD/SCID mice (n = 4/group). After 6 weeks, the number of metastatic lesions on the lung and pleura was counted. Individual organs were excised and analyzed by hematoxylin and eosin (H&E) staining.
Clinical specimens. All 217 patients were enrolled with informed consent in accordance with the Review Boards of Zhejiang University. Data of the clinicopathological parameters were obtained from the Xiaoshan tumor registry system database. The 217 specimens were from 117 males and 100 females. The age at diagnosis ranged from 26 to 85 years (median, 61) and follow-up ranged from 1 to 207 months (median, 95).
Statistical analysis. The statistical package SPSS (version 20.0; IBM, New York, NY, USA) was applied. For results from cell lines, all data are reported as mean ± S.D. Unpaired Student's t tests were used for normally distributed data and non-parametric Mann-Whitney U-tests were used for non-normally distributed data to compare central tendencies. For results in CRC tissues, comparisons of clinicopathological parameters and EMT markers in the IGFBP-rP1-low and IGFBP-rP1-high groups were made by the χ 2 test or Fisher's exact test. Correlations were analyzed by the Spearman coefficient test. Significance was set at Po0.05.

Conflict of Interest
The authors declare no conflict of interest.