PMP22 Regulates Self-Renewal and Chemoresistance of Gastric Cancer Cells

Gastric cancer is a major health burden worldwide. Its incidence ranks second among men and third among women, and the associated mortality still remains second for both males and females despite a downward trend in 2015 (1). The delivery of chemotherapeutic agents, including cisplatin (DDP) and fluoropyrimidine, following surgical resection is currently the standard treatment for advanced gastric cancer patients (2). Although multiagent chemotherapy can improve gastric cancer outcomes, almost all patients develop chemotherapy resistance, and the 5-year survival rate continues to be poor (3).

The cancer stem cell (CSC) hypothesis proposes that CSCs are a group of heterogeneous cells that exhibit self-renewal, multiple differentiation potential, high tumorigenicity, and drug resistance (4, 5). CSCs may have a central role in tumor initiation/maintenance, progression/metastasis, and recurrence (6–10). Conventional anticancer chemotherapies typically kill only differentiated tumor cells, whereas CSCs could be selected for recurrence. Therefore, the development of anticancer drugs that can target all CSC subsets within a tumor to control tumor growth and prevent recurrence is promising. Gastric cancer stem cells (GCSC) have received great attention, and some GCSC candidate surface markers were reported as potential novel therapeutic targets in the treatment of gastric cancer (11). Takaishi and colleagues found that CD44 can be used as a GCSC marker (12). Furthermore, ALDH1⁺, CD90⁺, CD71⁻, CD133⁺, CD44⁺/CD24⁺, EpCAM⁺/CD44⁺, and CD44⁺/CD54⁺ were also identified as potential GCSC markers (13–18). Despite many laboratory efforts, there is no highly specific marker for both GCSCs and chemoresistant gastric cancer cells, and the mechanisms regulating the self-renewal and chemoresistance of gastric cancer cells are not fully understood.

PMP22 is an integral membrane glycoprotein of peripheral nervous system myelin (19). Mutations in the PMP22 gene are the most common causes of Charcot–Marie–Tooth disease (CMT), which is an inherited peripheral nerve disorder (20). The precise expression and function of PMP22 in tumors remain to be clarified. Several studies indicate that PMP22 is a potential tumor suppressor (21, 22), whereas some observations suggest a potential oncogenic function of PMP22 in tumors (23–28). Studies on the role of PMP22 in regulating gastric cancer have not been reported, and we describe here that PMP22 is required for both self-renewal and chemoresistance in gastric cancer.

Bortezomib, a proteasome inhibitor, was approved for the treatment of multiple myeloma by the FDA in 2003 (29, 30). A proposed mechanism for its action is that it prevents the proteasomal degradation of proapoptotic proteins, leading to enhanced apoptosis (31). Jang and colleagues found that bortezomib exhibited marked reduction of endogenous PMP22 mRNA and protein (32). We showed that bortezomib reduced PMP22 expression levels in gastric cancer and significantly suppressed tumor growth when combined with traditional chemotherapy.

Our study provides strong evidence suggesting that PMP22 may act as a therapeutic target mediating the self-renewal and chemoresistance functions in gastric cancer. We also provide a new anticancer regimen that inhibits PMP22 expression to enhance the anticancer effect.

The MGC-803, SGC-7901, and SGC-7901/DDP human gastric cancer cell lines were purchased from the Institute of Cell Biology (Shanghai, China; http://www.cellbank.org.cn). The 293T human embryonic kidney cell line was obtained from the ATCC. Short tandem repeat sequencing has been done for cell authentication. All cell lines used in this study have been validated within the last 6 months. MGC-803, SGC-7901, and SGC-7901/DDP cells were maintained in RPMI1640 medium, whereas 293T cells were maintained in DMEM. All media contained 10% FBS (Gibco) and 1% penicillin/streptomycin (Invitrogen), and cells were maintained at 37°C in a 5% CO₂ humidified atmosphere. PCR tests for mycoplasma were negative. Cisplatin (Qilu Pharmaceutical Co., Ltd.) and bortezomib (Cell Signaling Technology, catalog no. 2204) were used for gastric cancer cell chemotherapy. PCR tests for mycoplasma were negative.

The pLKO.1 lentiviral vector was used to express short hairpin RNA directed against the PMP22 or LacZ control sequence (GTCTCCGAACGTGTCACGTT). Human PMP22 target sequences of PMP22 shRNA1 and PMP22 shRNA2 were CCAAACTCAAACCAAACCAAA and CGGTGTCATCTATGTGATCTT, respectively. The generation of lentiviral vectors was performed by cotransfecting pLKO.1 carrying the expression cassette with helper plasmids pVSV-G and pHR into 293T cells using Lipofectamine 3000 transfection reagent (Invitrogen Life Technologies). The viral supernatant was collected 48 hours after transfection, and cells at 60% to 70% confluence were infected with viral supernatants containing 10 mg/mL polybrene for 24 hours. Then, fresh medium with puromycin was used to select a pool of stably expressed cells.

Total RNA from cells or tissues was extracted using TRIzol reagent (Invitrogen) and reverse-transcribed with the SuperScript First-Strand Synthesis System for RT-PCR (Invitrogen) to produce cDNA according to the manufacturer's protocol. Platinum SYBR Green qPCR SuperMix (Invitrogen) was used for the qPCR reaction, and the expression levels were quantified using the −ΔΔC_t method. GAPDH was used as the internal control. The following primer sequences for quantitative PCR were used:

PMP22: 5′-CTGGTCTGTGCGTGATGAGTG-3′ (forward),

5′-ATGTAGGCGAAACCGTAGGAG-3′ (reverse);

MYC: 5′-CTTCTCTCCGTCCTCGGATTCT-3′ (forward),

5′- GAAGGTGATCCAGACTCTGACCTT -3′ (reverse);

SOX2: 5′-GCCGAGTGGAAACTTTTGTCG-3′ (forward),

5′-GGCAGCGTGTACTTATCCTTCT-3′ (reverse);

DDR1: 5′-GCGTCTGTCTGCGGGTAGAG-3′ (forward),

5′-ACGGCCTCAGATAAATACATTGTCT-3′ (reverse);

GAPDH: 5′-TCTCCTCTGACTTCAACAGCGA-3′ (forward),

5′- GTCCACCACCCTGTTGCTGT -3′ (reverse).

Cells were lysed with NP-40 lysis buffer. Anti-PMP22 (Sigma-Aldrich, catalog no. P0081), anti-claudin-7 (Santa Cruz Biotechnology, catalog no. sc-33532), anti-HLA-DR (Santa Cruz Biotechnology, catalog no. sc-51617), anti-CD16 (Santa Cruz Biotechnology, catalog no. sc-20052), anti-PARP (Sigma-Aldrich, catalog no. P248), anti-CD24 (R&D Systems, catalog no. AF5247), anti-CD44 (Cell Signaling Technology, catalog no. 5640), anti-ALDH1 (R&D Systems, catalog no. AF5869), and anti-β-actin (Sigma-Aldrich, catalog no. A1978) were used for immunoblotting. β-Actin protein was used as the internal control.

The cells (500 cells per well) were cultured in ultra-low-attachment 12-well plates (Corning Life Sciences) in 1 mL of serum-free DMEM/F-12 (Invitrogen) supplemented with B-27 (1:50; Invitrogen), 20 ng/mL EGF (BD Biosciences), 20 ng/mL basic fibroblast growth factor (bFGF; BD Biosciences), and 4 mg/mL insulin (Sigma-Aldrich). Cells were fed every 3 days. The sphere number and size were measured on day 14.

For spheroid colony formation, virus-infected cells were inoculated in each well (three cells per well) of ultra-low-attachment 96-well plates (Corning Life Sciences) in sphere culture medium. After 2 weeks, each well was examined using a light microscope, and the total number of wells with spheres was counted. Images of the spheres were recorded using a Model 1×71 S8F-2 inverted microscope (Olympus America Inc.) and a MicroPublisher 5.0 RTV camera (QImaging). Relative sphere volumes were calculated at the same time points using the following equation: volume = length × (width)² × π/6. Relative sphere length and width were assessed using Adobe Photoshop CS5 software under identical settings.

For the colony formation assay, stable transfected cells were plated in 6-well dishes at 400 cells per well. After 12 days, the cultures were washed twice with PBS, incubated with methanol for 20 minutes, stained with crystal violet for 30 minutes, and washed with tap water. The colonies were counted under a low-magnification microscope, and a group of more than 10 cells was defined as a colony.

For soft agar colony formation, agar (Nacalai Tesque) was dissolved in culture medium to 0.5% and plated in 6-well plates (bottom layer). Then, cells were seeded at 400 cells per well in 0.3% agar (top layer) over the bottom layer. Cells were covered with liquid growth media and fed every 3 days. The colony number was measured on day 14.

Confluent cells were washed once with PBS and then harvested with 0.05% trypsin/0.025% EDTA. Detached cells were washed with PBS containing 2% FBS (FACS buffer) and resuspended in the FACS buffer. The cell suspensions were incubated with isotype controls (BD Biosciences) or an mAb against human PMP22 (Sigma-Aldrich), followed by the secondary antibodies anti-mouse IgG/IgM conjugated with FITC (BD Biosciences). For cell analysis of the PMP22⁺ population, 10⁶ cells were analyzed using a Beckman EPICS XL flow cytometer that assessed 10,000 events. For cell sorting of PMP22⁺ and PMP22⁻ populations, cells were sorted with a Beckman MoFlo XDP cell sorter. The purity of the sorted cells was estimated to be greater than 95%. Cell apoptosis was evaluated by FACS using the Annexin V-FITC/PI Apoptosis Detection Kit (Sigma-Aldrich) following the manufacturer's instructions.

Cell growth rates and cytotoxicity were determined using MTS (Promega) following the manufacturer's instructions. The absorbance at 490 nm was measured with a Multiskan GO microplate spectrophotometer (Thermo Scientific).

The antibody microarray (ASB600) was obtained from Full Moon BioSystems. Each glass slide contains 656 highly specific and well-characterized antibodies in duplicate. Proteins (whole-cell lysates) were extracted, biotinylated, hybridized to the microarray, and detected with fluorescent-labeled strepatavidin using the Antibody Microarray Detection Kit (Spring Bioscience) according to the manufacturer's protocol. A change of approximately 2-fold was used as the cut-off value to evaluate the differential expression of proteins between consecutive DDP-treated and control xenograft cells.

Microarray experiments to investigate the transcriptional profiles of the MGC-803 Ctrl shRNA and PMP22 shRNA2 cells were performed by Shanghai KangChen Biotech Company (http://www.kangchen.com.cn). Sample labeling and microarray (Agilent Human 4×44K Gene Expression Microarrays, >41,000 probes) hybridization were performed according to the Agilent One-Color Microarray-Based Gene Expression Analysis protocol (Agilent Technology). Corresponding GEO accession numbers were GSE94714.

Experiments were conducted in male BALB/c athymic nude mice aged 5 weeks old. Animal care and handling procedures were performed in accordance with the Guide for the Care and Use of Laboratory Animals, and the animal study protocol was approved by the Institutional Animal Care and Use Committee of Xiamen University (reference no. XMULAC20150166). MGC-803 gastric cancer cells (3 × 10⁶ cells for each mouse) were injected subcutaneously into the back fat pad of nude mice. After tumors reached approximately 100 mm³ in size, mice were randomized to treatment with PBS or different doses of cisplatin for 4 days of a 20-day period. For GCSC enrichment, consecutively passaged equal numbers (3 × 10⁶) of MGC-803 xenograft cells from the 3 mg/kg DDP group and 0.9% NaCl group were treated in vivo with the same dose of DDP or NaCl for 4 days of a 32-day period as one passage.

For in vivo chemotherapy experiments, nude mice were subcutaneously inoculated with SGC-7901/DDP cells (2 × 10⁶ cells per mouse). After 8 days, tumor-bearing mice were treated with 5 mg/kg DDP, 0.4 mg/kg bortezomib, or 2.5 mg/kg DDP combined with 0.25 mg/kg bortezomib for 3 days over a 15-day period (n = 6 mice per group). The tumor volumes were monitored and recorded every three days. The investigator who monitored the tumor volumes was blinded to group allocation. The tumor volumes were estimated using the following formula: 0.5 × length × width².

The cohort of 38 eligible patients with resectable locally advanced adenocarcinoma of the stomach (AJCC, stage III) diagnosed from 2010 to 2013 was assigned to receive 6 cycles of perioperative epirubicin, cisplatin and 5-fluorouracil (ECF) chemotherapy. PMP22 expression levels were analyzed in the tumor samples obtained by surgery after preoperative chemotherapy. According to the follow-up survey, the patients were divided into two groups based on whether they experienced recurrence within 2 years. The tumor material was collected after informed consent was obtained, in agreement with the Institutional Review Board of Zhongshan Hospital of Xiamen University. The investigator who performed the examination of PMP22 expression was blinded to the sample information.

The data were analyzed using GraphPad Prism software. The data are presented as the mean ± SD, n ≥ 3. All data were analyzed using two-sided Student t tests, and P < 0.05 was considered statistically significant.

Resistance to chemotherapy distinguishes CSCs from other cancer cells (10). Therefore, GCSCs could be enriched by in vivo consecutive passage of gastric cancer cells in NOD/SCID mice treated with chemotherapy. First, we treated mice with xenograft MGC-803 cell tumors by tail vein injection of different doses of cisplatin (DDP). As shown in Fig. 1A, xenograft growth was effectively inhibited only when the concentration of DDP reached 3 mg/kg. Then, we chose this dosage (3 mg/kg) for the treatment of two subsequent passages of xenograft tumors. The relative tumor weight ratio in the 0.9% NaCl control group versus the DDP-treated group was 30% for the second-passage xenografts, whereas this relative tumor weight ratio reached 70% in the third-passage xenografts, which suggests the gain of a chemoresistance phenotype in the process of consecutive chemotherapy (Fig. 1B). Moreover, the mRNA levels of the self-renewal–related and drug resistance–related genes MYC, SOX2, and DDR1 were dramatically increased in the third-passage xenografts treated with consecutive chemotherapy (Fig. 1C).

To examine whether chemotherapy may enrich for GCSCs, freshly isolated cells from the third-passage DDP-treated xenografts were cultured in suspension to generate tumor spheres. The number of spheres reflects the quantity of cells capable of in vitro self-renewal, whereas the number of cells per sphere indicates the self-renewal capacity of each sphere-generating cell (33, 34). As shown in Fig. 1D, freshly isolated cells from the third passage of DDP-treated xenografts generated an increased number of spheres that were larger than the controls. Finally, we analyzed the protein expression profiles of the third-passage DDP-treated xenografts versus the 0.9% NaCl-treated controls using the explorer antibody microarray. Notably, several cell surface proteins, including Claudin7, PMP22, HLA-DR, and CD16, were significantly altered (Supplementary Table S1), and Western blotting confirmed these results (Fig. 1E). Together, these results indicated that chemotherapy selectively enriches for self-renewing gastric cancer cells.

Given that PMP22 is the most upregulated cell surface protein in DDP-selected xenografts, we determined whether PMP22 is also upregulated in tumorspheric cells. Higher levels of PMP22⁺ cells were detected in tumorspheric cells than in adherent cultures of MGC-803 cells according to FACS analysis (Fig. 2A). Consistently, PMP22 mRNA and protein levels were also increased in tumorspheric cells (Fig. 2B). Next, PMP22⁺ and PMP22⁻ cells sorted from tumorspheric MGC-803 cells cultured in suspension were plated on collagen under differentiating conditions in serum. Only approximately 5.97% remained PMP22⁺ by day 5 in the PMP22⁺ fraction, whereas the PMP22⁻ fraction remained >97% by day 5 (Fig. 2C). Therefore, PMP22⁺ cells not only have self-renewing capability, but, at the same time, PMP22⁺ cell levels were negatively correlated with GCSC differentiation process in vitro.

Next, we detected other known GCSC markers as well as the mRNA levels of the self-renewal and drug resistance genes MYC, SOX2, and DDR1 in PMP22⁺ and PMP22⁻ MGC-803 cells. As shown in Fig. 2D, proteins levels of GCSC marker ALDH1 in PMP22⁺ MGC-803 cells was more than 3-fold in PMP22⁻ cells, but the CD24 and CD44 proteins levels had no obvious change. The qPCR results showed that the expression of MYC, SOX2, and DDR1 mRNA was significantly increased in PMP22⁺ cells compared with PMP22⁻ cells (Fig. 2E). Moreover, PMP22⁺ cells formed more colonies and spheres than PMP22^– cells in the soft agar colony formation (Fig. 2F) and sphere formation (Fig. 2G) assays. Using the MTS assay, we also observed an increased cell growth rate in PMP22⁺ cells compared with PMP22⁻ cells (Fig. 2H); this phenotype may be due to a combination of factors such as PMP22 induced stem cell expansion and increased expression of some oncogenes in gastric cancer cells. Collectively, these results demonstrate that PMP22 is upregulated in self-renewing gastric cancer cells and that PMP22⁺ cells possess the features of self-renewal.

To determine the role of PMP22 in gastric cancer cells, PMP22 was significantly knocked down in the MGC-803 cells by lentivirus-expressed shRNAs (Fig. 3A). First, we identified PMP22 downstream-regulated genes by screening a microarray system. Remarkably, by performing ANOVA of normalized microarray data (Supplementary Table S2), we identified a set of genes associated with drug resistance, stemness, and tumorigenesis whose expression levels in PMP22 knockdown MGC-803 cells significantly differed from those in the control cells (Fig. 3B). These results indicate that PMP22 may play functional roles in regulating the carcinogenesis, self-renewal, and chemoresistance of gastric cancer cells.

We also performed plate colony formation (Fig. 3C) and spheroid colony formation (Fig. 3D) assays, and the results showed that PMP22 knockdown cells generated significantly fewer and smaller colonies in both experiments. The MTS assay results also showed that PMP22 knockdown significantly inhibited the cell growth rate in MGC-803 cells (Fig. 3E). Next, we subcutaneously injected these cells into SCID mice. After 4 weeks, we observed that the PMP22 knockdown cells produced much smaller tumors than those derived from the control cells (Fig. 3F). These results imply that PMP22 is involved in tumor initiation of gastric cancer.

Studies have suggested that CSC-enriched cells in several solid tumors exhibited greater resistance to anticancer chemotherapy drugs than the non-CSC population (35, 36). To investigate whether PMP22 is associated with the chemoresistance of gastric cancer cells, we performed cell survival assays using different doses of DDP. The PMP22⁺ fraction of MGC-803 cells exhibited increased resistance to the anticancer drug DDP compared with the PMP22⁻ fraction (Fig. 4A). In addition, PMP22 knockdown in MGC-803 cells significantly inhibited chemoresistance to DDP (Fig. 4B). SGC-7901/DDP was an existing and approved DDP-resistant gastric cancer cell line, so we chose SGC-7901 as well as corresponding DDP-resistant cell lines SGC-7901/DDP to carry out related experiment. Increased levels of PMP22⁺ were also detected in SGC-7901/DDP cells compared with SGC-7901 cells (Fig. 4C). Consistently, qPCR results showed that PMP22 mRNA expression was significantly increased in SGC-7901/DDP cells compared with SGC-7901 cells. Next, we knocked down PMP22 expression in SGC-7901/DDP cells to assess chemotherapeutic sensitivity (Fig. 4D). As shown in Fig. 4E, SGC-7901/DDP cells naturally exhibited increased resistance to DDP compared with SGC-7901 cells; however, PMP22 knockdown significantly increased chemotherapeutic sensitivity to DDP. Finally, we assessed the effect of PMP22 on the sphere formation ability of DDP-resistant cells. As expected, SGC-7901/DDP cells formed significantly more and larger spheres than SGC-7901 cells, but PMP22 knockdown in SGC-7901/DDP cells significantly inhibited the sphere formation ability (Fig. 4F). Together, our findings show that reduced PMP22 could inhibit chemoresistance to DDP in gastric cancer cells and that this effect may be associated with self-renewal phenotypes.

Standard chemotherapy approaches for gastric cancer are still associated with a high rate of recurrence and mortality, as resistance is rapidly acquired (37). Therefore, we hypothesized that standard chemotherapy combined with PMP22 targeting may yield better therapeutic effects than either used alone. Previous studies have reported that bortezomib was used as a redifferentiation agent for the treatment of multiple myeloma patients and significantly inhibited PMP22 expression levels (30, 32). However, bortezomib has not been used for the treatment of gastric cancer. Therefore, we used bortezomib as a PMP22 inhibitor in combination with DDP for chemoresistant gastric cancer therapy. The MTS results showed that the survival rate of SGC-7901/DDP cells was greater than 60% after treatment with 8 μg/mL DDP or 40 μmol/L bortezomib, separately, for 36 hours; however, treatment with 4 μg/mL DDP combined with 20 μmol/L bortezomib for 36 hours resulted in a cell survival rate of less than 25% (Fig. 5A). Moreover, this combined chemotherapy regimen dramatically induced the activation of PARP cleavage (Fig. 5A). Consistently, when we used flow cytometry to determine the percentage of apoptotic cells after exposure to different chemotherapies for 24 hours, the results indicated that cell apoptosis in the combined chemotherapy group was greatly increased compared with that resulting from either agent alone (Fig. 5B). Interestingly, the qPCR results revealed that PMP22 mRNA levels were significantly increased by DDP treatment and were significantly reduced by bortezomib treatment. However, there were no significant differences in PMP22 mRNA levels between the combined chemotherapy and PBS treatment groups (Fig. 5C). Collectively, the above in vitro results verify the hypothesis that standard cisplatin chemotherapy combined with PMP22 inhibition by bortezomib could increase chemotherapeutic sensitivity by inducing cell apoptosis.

The treatment effects of PMP22 inhibition and DDP chemotherapy were next examined in vivo. SGC-7901/DDP cells were subcutaneously injected into the back of mice to form xenografts. After tumors reached 50 to 100 mm³ in size, mice were treated with PBS or different chemotherapies 3 days for 15 days. Control tumors treated with PBS grew to over 800 mm³ by 15 days following treatment, whereas the tumors treated separately with 5 mg/kg DDP or 0.4 mg/kg bortezomib grew to over 500 mm³. However, the combination of 2.5 mg/kg DDP and 0.25 mg/kg bortezomib dramatically reduced tumor growth to 200 mm³ (Fig. 6A and B). Similar results were achieved when we measured tumor weight (Fig. 6C). Notably, the relative body weight of mice pre- and postchemotherapy (d8/d23) was significantly increased with the combined chemotherapy compared with 5 mg/kg DDP alone (Fig. 6D). FACS and qPCR were performed to determine the PMP22⁺ cell levels and PMP22 mRNA levels in gastric cancer xenograft cells after treatment with different chemotherapies. PMP22⁺ cell levels and PMP22 mRNA levels were significantly increased after 5 mg/kg DDP chemotherapy for 15 days, whereas 0.4 mg/kg bortezomib treatment significantly inhibited PMP22 expression. As expected, PMP22⁺ cell levels and PMP22 mRNA levels after the combined chemotherapy were similar to those in the PBS treatment group (Fig. 6E and F).

To investigate the association of PMP22 with clinical chemoresistance in gastric cancer patients, we selected 38 gastric cancer specimens from patients who received 6 cycles of perioperative chemotherapy to assess PMP22 mRNA levels by qPCR. The patients were divided into two groups according to clinical outcome as follows: group 1, patients with recurrence; group 2, patients without recurrence 2 years after surgery. The results showed that the PMP22 levels were significantly increased in gastric cancer tissues from patients with recurrence compared with the group without recurrence (Fig. 6G). Together, these results strongly indicate that PMP22 activation is involved in chemoresistance in gastric cancer patients and that standard cisplatin chemotherapy combined with PMP22 inhibition using bortezomib may yield better therapeutic effects.

Chemotherapy-induced drug resistance remains a major obstacle to the clinical management of gastric cancer, resulting in relapse and metastasis. Thus, novel strategies to overcome drug resistance are currently under intense investigation. GCSCs are a major factor in gastric cancer resistance to radiation and chemotherapy (11). CSCs isolated from gastric cancer cell lines using the SP method exhibit drug tolerance for chemotherapy (12, 13). However, this is not a direct method for the simultaneous enrichment of GCSCs and chemoresistant gastric cancer cells. As resistance to chemotherapy distinguishes CSCs from other cancer cells (10), the in vivo consecutive passage of cancer cells in NOD/SCID mice treated with chemotherapy is an effective method for the enrichment of CSCs, as demonstrated in previous studies and our study (38). A long period of chemotherapy also simulates the induction of drug resistance in tumors in patients undergoing clinical treatment. Therefore, this method is suitable for the identification of diagnostic or therapeutic targets for both CSCs and chemoresistance.

In a search for cell surface protein(s) that are markers of self-renewing and chemoresistant gastric cancer cells using the Explorer Antibody Microarray, we identified PMP22, a target gene of the Wnt/β-catenin pathway (39, 40), as the most upregulated cell surface protein in the xenografts of the consecutive chemotherapy group. The biological functions of PMP22 proteins in mammalian cells are complex. PMP22 comprises an estimated 2% to 5% of the total myelin proteins in the peripheral nervous system (41). Its functional importance is highlighted by the fact that mutations in the PMP22 gene are the most common causes of inherited peripheral nerve disorders (20). In addition to its expression in neural cells, PMP22 is also observed in non-neural cells and tissues during development, such as epithelial cells of the lung, intestines, and stomach (42–44). The precise expression and function of PMP22 in tumors remain to be clarified. Several studies indicate a potential tumor suppressor function of PMP22. For example, PMP22 depletion results in increased proliferation in breast cancer cells (21). High expression of PMP22 results in growth arrest and apoptosis in fibroblasts (22). In contrast, some observations suggest a potential oncogenic function of PMP22 in tumors. PMP22 participates in the oncogenesis, progression, and metastasis of osteosarcoma (23–26). Furthermore, high-level PMP22 amplification was observed in high-grade glioma (27, 28). In the current study, we revealed for the first time the role of PMP22 in gastric cancer in potentially mediating self-renewal and chemoresistance in gastric cancer cells. This role is strongly supported by the following findings: (i) PMP22 expression levels and PMP22⁺ cell levels were upregulated in both tumor spheres and a DDP-resistant gastric cancer cell line; (ii) PMP22⁺ gastric cancer cells exhibited enhanced self-renewing ability and drug tolerance for chemotherapy compared with PMP22⁻ cells; (iii) knockdown of PMP22-inhibited self-renewal and chemoresistance in gastric cancer cells was found in vitro and in vivo; and (iv) the expression pattern and functional relevance of PMP22 in tumor model and cancer samples were identified. Because CSCs possess chemoresistant features, we hypothesized that chemoresistant cancer cells acquire more stem cells. Indeed, we observed an expanded stem cell population in SGC-7901/DDP cells compared with SGC-7901 cells and revealed a reciprocal link between stem cell properties and chemoresistance in cancer cells. Moreover, the inhibition of PMP22 expression further reduced the GCSC population in the resistant gastric cancer cells. Above all, these results indicated that PMP22 is more than just a marker of the self-renewing capacity and chemoresistance of gastric cancer cells. PMP22 functionally contributes to the self-renewal, tumorigenicity, and chemoresistance of gastric cancer cells. We also predict that when combined with other GCSC surface markers, such as CD44, PMP22 can precisely recognize the GCSC population.

Conventional chemotherapy only kills differentiated tumor cells, resulting in tumor size reduction. However, the tumors relapse after some time, possibly due to the presence of residual and selected CSCs (11). Recent advanced strategies have been suggested for targeting GCSCs, including a combination of chemotherapy-associated apoptosis, tumor stem cell differentiation induction, therapies targeting GCSC surface molecules, and inhibition of GCSC-related pathways (11–13, 45, 46). However, these strategies are difficult to implement directly in clinical treatment. In our study, which aimed to investigate an existing clinical drug to target PMP22 in gastric cancer cells, we identified bortezomib, which has been used for the clinical treatment of multiple myeloma (30, 31). More importantly, bortezomib markedly reduces endogenous PMP22 mRNA and protein (32). Indeed, we observed effective inhibition of PMP22 by bortezomib in chemoresistant gastric cancer cells. When combined with DDP, bortezomib treatment dramatically increased chemotherapeutic sensitivity by inducing cell apoptosis.

In summary, we have shown that cisplatin chemotherapy selectively enriches self-renewing gastric cancer cells and that standard cisplatin chemotherapy combined with bortezomib enhances the effects of chemotherapy. We identified that PMP22 not only acts as a marker for gastric CSCs but may also have an essential role in regulating the self-renewal and chemoresistance of gastric cancer. Our findings suggest that PMP22 has clinical value for the prognosis and treatment of chemoresistant gastric cancer.

No potential conflicts of interest were disclosed.

Conception and design: W. Cai, Q. Luo, B. Li, J. Cai

Development of methodology: W. Cai, G. Chen, Q. Luo, J. Liu, X. Guo

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): G. Chen, T. Zhang, F. Ma, L. Yuan

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): W. Cai, G. Chen, J. Liu, F. Ma

Writing, review, and/or revision of the manuscript: W. Cai, B. Li, J. Cai

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): W. Cai, G. Chen

Study supervision: B. Li

This work was supported by grants from the National Natural Science Foundation of China: grant numbers 81372616 (to J. Cai), 81272384 (to B. Li), and 81602148 (to W. Cai); and the Natural Science Foundation of Fujian Province: grant number 2014D018 (to J. Cai).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.