Periostin regulates autophagy through integrin α5β1 or α6β4 and an AKT‐dependent pathway in colorectal cancer cell migration

Abstract Colorectal cancer (CRC) is one of the most fatal cancers with highly invasive properties. The progression of CRC is determined by the driving force of periostin (PN) from cancer‐associated fibroblasts (CAFs) in the tumour microenvironment. This present work aims to investigate autophagy‐mediated CRC invasion via the receptor integrin (ITG) by PN. The level of PN in 410 clinical CRC tissues was found increased and was an independent poor prognosis marker (HR = 2.578, 95% CI = 1.218‐5.457, P‐value = .013) with a significant correlation with overall survival time (P‐value < .001). PN activated proliferation, migration and invasion of CRC cells, but with reduced autophagy. Interestingly, the reduction of LC3 autophagic protein corresponded to the increased ability of CRC cell migration. The siITGα5‐treated HT‐29 and siITGβ4‐treated HCT‐116 CRC cells attenuated epithelial‐to‐mesenchymal transitions (EMT)‐related genes and pAKT compared with those in siITG‐untreated cells. The reduction of pAKT by a PI3K inhibitor significantly restored autophagy in CRC cells. These evidences confirmed the effect of PN through either ITGα5β1 or ITGα6β4 and the AKT‐dependent pathway to control autophagy‐regulated cell migration. In conclusion, these results exhibited the impact of PN activation of ITGα5β1 or ITGα6β4 through pAKT in autophagy‐mediated EMT and migration in CRC cells.

periodontal ligaments and gastrointestinal tracts. 4,5 PN is commonly expressed in cancer-associated fibroblasts (CAFs) of many cancers including oesophageal, gastric, colorectal, prostate, breast and ovarian. [6][7][8][9][10] In CRC, PN promotes cancer cell survival, angiogenesis and resistance to chemotherapeutic drugs; is also involved in epithelial-mesenchymal transitions (EMT) leading to cancer invasiveness and metastasis; and is related to poor prognosis in several cancers including colon, pancreatic, ovarian, breast, head and neck, thyroid and gastric. 4,11,12 The receptor integrin (ITG) has been proven to be activated by PN through the AKT/PKB signalling pathway to promote these tumorigenic properties. 13 The importance of autophagy is becoming widely recognized as it demonstrates both pro-and anti-tumorigenic functions. 14 In CRC, cell autonomy and non-autonomous roles for autophagy are essential in growth and progression; however, the mechanisms downstream of autophagy to reduce or enhance tumour growth are not well known. 15 It was recently reported that autophagy protein expression levels in mouse osteoblasts were increased after PN silencing. 16 Up to present knowledge, the alterations of autophagy on CRC tumour promotion after exposure to PN are still unclear.
In this study, 410 CRC patient samples were investigated for PN expression status using immunohistochemistry and the correlation of PN levels to clinicopathological features and disease-free survivals of the patients were analysed to validate the impact of PN.
Additionally, the role of PN in reduction of autophagy in CRC cells through an ITG-dependent pathway was shown for the regulation of PN-stimulated CRC cell migration. It is likely to conclude that CAFsderived PN affects cancer cells through the ITGα5β1 or the α6β4, via an AKT-dependent signalling pathway to attenuate autophagy leading to the induction of cancer cell migration. The molecules in the PN-ITG-autophagy axis are proposed as targets to attenuate disease progression in CRC patients.

| Immunohistochemistry of PN in human CRC tissues
The 4-µm paraffin-embedded CRC tissues were processed for PN staining by the protocol as previously reported. 17 Inc, Buffalo Grove, IL, USA). The scoring values were evaluated by the percentage of positive staining cells (P) and the intensity of the staining signal (I). For P, 0%-25%, 26%-50%, 51%-75% and 76%-100% were classified as grades 0, 1, 2 and 3. For I, remained unstained, slightly stained, intermediately stained and strongly stained were classified as 0, 1, 2 and 3. The expression scoring was calculated by P x I that covered the total core of 0-9. Each stained slide was observed by 2 investigators double-blinded to the clinical data. Patients with a total score of more than 4 (the median value of the overall scores) were grouped as the high PN expression group whereas those with less than or equal to 4 were in low PN group.

| Small interference RNA (siRNA) knockdown of ITGα5 and ITGβ4
Human Heidelberg, Germany). 19 The fluorescein conjugate-A (sc-36869, Santa Cruz Biotechnology, Inc) and the control siRNA-A (sc-37007, Santa Cruz Biotechnology, Inc) were utilized as the positive and negative controls.

| Statistical analysis
The correlation of PN expression in CRC tissues with clinicopathological data was verified by Fisher's exact univariate analysis. The Kaplan-Meier log rank test was used for survival analysis. The Cox proportional hazards model was used to examine the association of each clinicopathological parameter with overall survival. The significant parameters were further determined in stepwise modelled multivariate analysis. The P-value of less than .05 was considered statistically significant. All statistical analyses were performed with SPSS version 23 (SPSS Inc, Chicago, IL, USA).

| Demographics and clinicopathological parameters correlation
Among 410 total cases, 232 (57%) cases were male. The patients were aged ranging from 29 to 95 years with a mean age of 64 years (  Figure 1A, i-viii) and was found very low in normal colorectal tissues ( Figure 1A, ix). High PN levels were correlated with statistical significance to lymphovascular invasion, perineural invasion, tumour staging, lymph node metastasis and distant metastasis (P-value < .001).
Multivariate analysis exhibited the significance of high PN levels in the tissues as an independent risk factor with an HR of 2.578 (P-value = .013). In contrast, serum CEA showed no significant risk factors ( Table 2). The lymphovascular (P-value = .002), perineural (P-value = .029) invasions and late stages of tumours (stages III-IV) (P-value < .001) were independently bad prognosis markers with an HR of more than 1. In addition, 3-years and 5-years overall survival using the Kaplan-Meier log rank test showed high levels of PN expression significantly correlated with patient short survival time with statistical significance ( Figure 1B). Interestingly, the level of tissue PN and serum CEA exhibited the significant linear correlation (Table S1).

| PN stimulates CRC spheroid formation and migration
The size of CRC spheroids of all CRC cell lines was increased in a time-dependent manner when exposed to rPN more than that of negative controls treated with 1% FBS media (Figure 2A-D), but only HT-29 showed a significant increase (P-value < .05) ( Figure 2B). Interestingly, both HCT-116 and HT-29 CRC cells revealed the increased cell invasion that protruded out of the spheroid masses representing the capability of cancer cell invasion ( Figure 2E,F) with statistical significance detected in HCT-116 CRC cells ( Figure 2E). In the wound-healing cell migration assay, rPN induced both CRC cells to move in a 2D way ( Figure 3A-D) but the significant effect was only observed in HCT-116 CRC cells ( Figure 3A,C).

| PN reduces migration capability of CRC cells through ITGα5β1 and α6β4
The expression of ITGs on CRC cells was explored and the re-

| PN attenuated autophagy in CRC underwent EMT through ITGα5β1 or ITGα6β4 and AKT-or ERKdependent pathways
The

| D ISCUSS I ON
PN is well accepted for its up-regulated expression in many cancers including CRC. 5,[8][9][10] Herein, the increased PN was confirmed in Thai CRC tissues and strongly correlated with patient short survival time as reported previously. 10,13,22 Moreover, these results supported the findings that CAFs were the major sources of PN production in CRC tissues and CAFs-derived PN was a poor prognostic marker in CRC patients. 9,10,23 The mechanism of PN to activate predominant ITGα5β1 or ITGα6β4 receptors on CRC cells to activate cancer cell aggressiveness was revealed.
The association of ITGα1β1 and the progression of CRC has been reported. 24 The ITGα8 mRNA was reported as a potential diagnostic biomarker whereas ITGα5 may serve as an independent prognosis indicator for CRC. 25 13 The results were confirmed by the increased levels of pAKT and pERK after PN treatment, but not in cells with impaired levels of these two receptors.
Notably, the increased level of pERK by PN activation was not attenuated which may have implied no involvement of the ERK pathway.
PN-activated PI3K/AKT/survivin was related to chemoresistance in CRC. 26 The PN-ITG-AKT-ERK-dependent pathway is related in cell F I G U R E 5 PN and autophagy in CRC cell lines. (A-C) CRC cells, HCT-116 and HT-29 were treated with 100 ng/ml of PN in 1% FBS medium, 30 µM of CLQ was added at 8 h before harvest, and then subjected for IF staining of PN (orange colour) and LC3 (green colour). Bar graph represents mean ± SD; *P-value < .05; **P-value < .01. D, E, Co-localization of PN and LC3 in tumour clinical sections of CRC patients. Scale bar = 20 µM; magnification = 63×. F Schematic diagram illustrates the potential of PN secreted from cancer-associated fibroblasts (CAFs) to reduce autophagy in CRC-mediated migration through ITGα5β1 or ITGα6β4 and AKT-dependent pathways proliferation and induction of invasion. 19,27,28 The evidence in this study highlights the roles of PN-ITGα5β1or PN-ITGα6β4 through an AKT-dependent pathway in CRC cell migration.
Importantly, the reduction of autophagy through ITGα5β1 and

ACK N OWLED G M ENTS
The authors would like to thank Prof. James A. Will, University of Wisconsin-Madison, USA for his English editing on this article.

CO N FLI C T O F I NTE R E S T
The authors confirm that there are no conflicts of interest.