Introduction

Small molecular inhibitors targeting the MAPK signaling pathway have achieved unprecedented therapeutic efficacies in the treatment of mutant BRAF melanoma [1,2,3]. However, almost all melanoma patients develop drug resistance through stable genetic alterations or non-genetic phenotype switching [4,5,6,7,8,9]. Phenotypic switching represents a mechanism of adaptive resistance in which tumor cells undergo transcriptional reprograming and switch from a highly proliferative, less invasive state to a highly invasive, poorly proliferative one [8, 10, 11]. In melanoma, long-term treatment of MAPK inhibitors (MAPKi) drives phenotype switching through altering the expressions of certain transcription factors (MITF, SOX10, BRN2/POU3F2) and receptor tyrosine kinases (AXL, ROR1/2) [12,13,14,15]. In general, the MITFlow/AXLhigh or SOX10low melanoma cells are associated with de-differentiated state, reduced proliferation, high invasiveness and MAPKi tolerance. While it is evident that MAPKi-resistant melanoma often acquire more aggressive and invasive phenotype, the molecular control of this enhanced invasiveness associated with drug resistance remains poorly understood.

Laminin β3 (LAMB3) is a β subunit of the extracellular matrix (ECM) protein Laminin-332 (consisting of α3, β3, and γ2 subunits) which is a central component of the basement membrane and participates in the regulation of various cellular processes including cell adhesion, survival, proliferation, differentiation and migration [16,17,18,19,20,21]. Studies have reported that LAMB3 promotes tumor growth, metastasis and invasion in several malignant tumors including pancreatic, colorectal, papillary thyroid cancer and head and neck squamous cell carcinoma, indicating that LAMB3 is a potential therapeutic target of cancer treatment [17,18,19,20,21]. However, the biological function of LAMB3 and its potential roles in melanoma invasion and drug resistance haven’t been investigated.

The transcription factor sex determining region Y (SRY) related HMG box-containing factor 10 (SOX10) is a member of the SOX family transcription factors that plays pivotal regulatory roles in embryonic development and melanoma progression [22]. SOX10 also regulates the migration and invasion of melanoma cells [23]. SOX10 has heterogenous expressions in melanoma and long-term MAPKi treatment causes decreased expression of SOX10 in a subpopulation of melanoma cells [13]. SOX10low melanoma cells acquire adaptive resistance to MAPKi through upregulation of the TGF-β-EGFR/PDGFRβ signaling and is associated with enhanced invasiveness [13, 24]. Additionally, our previous research has determined that SOX10 modulates the adaptive resistance of mutant BRAF melanoma cells to RAF inhibitors by activating its target genes FOXD3 and SAMMSON [25, 26].

In this study, we show that MAPKi-resistant melanoma cells exhibit enhanced invasive phenotype and discover a SOX10low/TGF-β/LAMB3/FAK/MMPs signaling pathway that modulates the migration and invasion potential of MAPKi-resistant melanoma cells. These findings explain the improved invasiveness associated with MAPKi-resistance and provide new targets for the treatment of drug-tolerant melanoma cells.

Results

MAPKi-resistant melanoma cells exhibit increased motility and invasiveness

MAPKi-resistant melanoma cells tend to acquire more invasive phenotypes in concomitance with drug resistance, allowing tumor cells to rapidly metastasize [15, 27,28,29,30,31]. Indeed, in vitro evolved MAPKi-resistance melanoma cells (A375R and 1205LuR) were more motile and invasive than their parental counterparts (Supplementary Fig. S1). To gain mechanistic insight of this pro-invasiveness effect associated with resistance, we performed KEGG enrichment analyses on two GEO datasets (GSE74729 and GSE186108) which compared gene expressions between parental and MAPKi-resistance melanoma cells/tumors [32, 33]. As expected, differential genes were significantly enriched in the MAPK signaling and Ras signaling pathways, which are known MAPKi-resistance associated pathways (Supplementary Fig. S2a). Of note, differential genes were also enriched in pathways that are implicated in cell adhesion and migration, such as focal adhesion and Rap1 signaling pathways, supporting the invasive phenotype of the drug resistance cells [34,35,36,37].

LAMB3 mediates enhanced cell motility and invasiveness of MAPKi-resistant melanoma cells

After exploring differential genes that are common to both datasets, we focused on the LAMB3 gene from the focal adhesion pathway as a potential mediator of the invasive phenotype of the MAPKi-resistant cells (Supplementary Fig. S2b). LAMB3 encodes the β3 subunit of laminin-332 and is implicated in the regulation of cell adhesion, migration and metastasis [16,17,18,19,20,21]. Consistent with the results of bioinformatic analyses, we found LAMB3 expressions were increased in A375R and 1205LuR cells compared to the parental cells (Fig. 1a, b). To investigate whether LAMB3 upregulation is required for the improved invasion and migration abilities of the MAPKi-resistant melanoma cells, LAMB3 was depleted using the CRISPR/Cas9 technology or siRNAs in A375 and 1205Lu cells (Fig. 1c). LAMB3 ablation inhibited the invasion/migration (Fig. 1d, e and Supplementary Fig. S3a–c) and colony-forming abilities of resistant melanoma cells (Fig. 1f and Supplementary Fig. S3d), but increased their sensitivities to Vemurafenib (Fig. 1g and Supplementary Fig. S4a). To further confirm the contribution of LAMB3 to the invasive phenotype, we overexpressed LAMB3 in A375 and 1205Lu cells and found that LAMB3 overexpression enhanced the motility and invasiveness of melanoma cells (Fig. 1h, i and Supplementary Fig. S4b). Since LAMB3 is a component of the Laminin-332 heterotrimeric complex (composed of α3, β3, and γ2) which triggers cell growth, differentiation and migration through binding to integrin receptors α3β1, α6β1 and α6β4 [38], we wondered whether the laminin β3/Integrin signaling is involved in LAMB3-mediated enhancement of migration and invasion of MAPKi-resistant melanoma cells. Western blotting and flow cytometry analyses revealed higher expression levels of total and cell surface α3 and α6 integrins in MAPKi-resistant melanoma cells than in their parental cells (Fig. 2a, b). Moreover, functional blocking of integrin α3 or α6 using specific antibodies strongly inhibited migration and invasion of MAPKi-resistant melanoma cells (Fig. 2c–f). In conclusion, our results indicated that LAMB3 expression was increased in MAPKi-resistant melanoma cells and the LAMB3/integrin signaling was required for the invasive phenotype of the resistant cells.

Fig. 1: LAMB3 mediates enhanced cell motility and invasiveness of MAPKi-resistant melanoma cells.
figure 1

(a) MAPKi-resistant melanoma cells A375R/1205LuR and their parental cells A375/1205Lu were harvested to isolate total RNA for qRT-PCR analysis on LAMB3 using GAPDH as the internal control. (b) same as (a) except that cells were lysed for western blot analysis. (c) A375R cells transduced with Crispr V2, Crispr/LAMB3 sgRNA#1 and sgRNA#4 and 1205LuR cells transfected with non-targeting control or LAMB3 siRNAs for 72 h were harvested for western blot analysis. (d) A375R cells transduced with Crispr V2, Crispr/LAMB3 sgRNA#1 and LAMB3 sgRNA#4 were harvested for cell invasion (6*104) and cell migration (1*104) assay. After culturing for 48/24 h, cells were stained with crystal violet and counted by Image J software. Quantitation of cell invasion and migration numbers are shown. (e) same as (d) except that 1205LuR cells (1*104) transfected with non-targeting control or LAMB3 siRNA#1 and siRNA#2 were used. (f) same as (d), except that cells (1*103) with 2 μM/10 μM vemurafenib were harvested for colony formation assay. Quantitation of cell colony formation numbers are shown. (g) same as (d, e), except that cells treated with 2 μM or 10 μM vemurafenib for 72 h. Cells were then harvested and stained with APC Annexin-V/PI for flow cytometry analysis. Average percent of annexin-V positive cells are shown. (h) A375 and 1205Lu cells (1*104) transduced with control or LAMB3 over-expression lentiviruses were harvested for cell migration and invasion assay. Quantitation of cell migration and invasion numbers are shown. (i) same as (h), except that cells were lysed for western blot analysis. Values in a and d–h represent the mean ± standard deviation of three independent experiments. Significance was determined by ANOVA one-way test, ***p < 0.001. The quantitation of relevant western blot images in b, c and i was provided.

Full size image
Fig. 2: The integrin α3 and α6 are required for LAMB3-mediated enhancement of invasive phenotype of MAPKi-resistant melanoma cells.
figure 2

(a) MAPKi-resistant melanoma cells A375R and their parental cells A375 were lysed for western blot analysis. (b) same as (a) except that cells were harvested for flow cytometry analysis on indicated extracellular proteins. Mouse IgG and Rat IgG were used as negative controls. Representative images of flow traces are shown. (c) MAPKi-resistant melanoma cells (1*104) treated by 10 ng/μL Ctrl IgG, α3 or α6 antibodies were harvested for cell migration assay. After culturing for 24 h, cells were stained with crystal violet and counted by Image J software. Representative images of cell migration are shown. (d) Quantitation of cell migration numbers are shown. (e, f) same as (c–d), except that cells were harvested for cell invasion assay. After culturing for 48 h, cells were stained with crystal violet and counted by Image J software. Representative images of cell invasion are shown. Values in d and f represent the mean ± standard deviation of three independent experiments. Significance was determined by ANOVA one-way test, ***p < 0.001. The quantitation of relevant western blot images in a was provided.

Full size image

SOX10 negatively regulates LAMB3

Previous studies have revealed that MAPKi-resistant melanoma cells often had decreased SOX10 expression and SOX10-deficiency led to reduced proliferation and invasive properties [13, 24]. RNA sequencing studies showed that SOX10 mRNA levels were reduced in BRAFi-resistant melanoma tumors versus the control tumors and that LAMB3 expression was increased in SOX10-depleted melanoma cells [13, 32]. These results suggested that SOX10 may negatively regulate LAMB3 expression. We found SOX10 and LAMB3 expressions showed negative correlation in parental and MAPKi-resistant melanoma cells (Fig. 3a). SOX10 knockdown induced LAMB3 expression at both mRNA and protein levels in different melanoma cells (Fig. 3b, c). Of note, ectopic expression of HA-SOX10 could nullify LAMB3 induction by SOX10 depletion in A375 cells (Fig. 3d) and also reduce LAMB3 expression in MAPKi-resistant cells (A375R and 1205LuR) (Fig. 3e). Together, our loss-of-function and rescue experiments consistently demonstrated that SOX10 negatively regulated LAMB3 expression in melanoma cells and SOX10-deficiency led to upregulation of LAMB3 in MAPKi-resistant cells.

Fig. 3: SOX10 negatively regulates LAMB3.
figure 3

(a) MAPKi-resistant melanoma cells A375R/1205LuR and their parental cells A375/1205Lu were harvested for western blot analysis. (b) Melanoma cells transduced with shRNA control, SOX10 shRNA#1 and SOX10 shRNA#2 were harvested for western blot analysis. (c) same as (b) except that cells were harvested to isolate total RNA for qRT-PCR analysis on LAMB3 using GAPDH as the internal control. (d) A375TR HA-SOX10 WT cells were transfected with control or SOX10 siRNAs for 72 h in the absence or presence of 100 ng/mL Doxycycline. Cells were then lysed for western blot analysis. (e) A375, A375R and A375R HA-SOX10 WT (1205Lu, 1205LuR and 1205LuR HA-SOX10 WT) cells were lysed for western blot analysis. Values in c represent the mean ± standard deviation of three independent experiments. Significance was determined by ANOVA one-way test, **p < 0.01, ***p < 0.001. The quantitation of relevant western blot images in a, b, d and e was provided.

Full size image

SOX10-deficiency induces LAMB3 upregulation via activating TGF-β signaling

Sun et al. have demonstrated that SOX10 depletion caused upregulation of the TGF-β type II receptor, TGFBR2 and activation of the TGF-β signaling, which drove adaptive MAPKi resistance in melanoma [13]. Another study showed that TGF-β treatment induced production of laminin-332 in cutaneous squamous cell carcinoma [39]. Inspired by these studies, we asked whether SOX10-deficiency could trigger LAMB3 upregulation by activating the TGF-β signaling. In line with this, we found SOX10 knockdown not only induced LAMB3 expression, but also increased phospho-Smad2 level, indicating the activation of TGF-β signaling (Fig. 4a). Importantly, TGF-β receptor type I/II inhibitor LY2109761 effectively negated the induction of LAMB3 by SOX10-depletion in a dose- and time-dependent manner (Fig. 4a and Supplementary Fig. 5). In addition, TGF-β1 treatment induced protein and mRNA expressions of LAMB3 and two known TGF-β signaling targets, EGFR and PDGFRβ in different melanoma cells (Fig. 4b, c). The upregulation of LAMB3 in MAPKi-resistant melanoma cells was associated with increased expression of the TGF-β type II receptor, TGFBR2 and knockdown of TGFBR2 reduced LAMB3 expression (Fig. 4d). These results support our hypothesis that SOX10-deficiency induces LAMB3 expression through activating the TGF-β signaling in MAPKi-resistant melanoma cells.

Fig. 4: SOX10-deficiency induces LAMB3 upregulation via activating TGF-β signaling.
figure 4

(a) Melanoma cells transduced with Ctrl shRNA or SOX10 shRNA#2 were treated with 0, 5, 10 μM LY2109761 for 24 h or 48 h, and cells were then lysed for western blot analysis. (b) Melanoma cells were treated with 5 ng/mL TGF-β1 for 0, 12, 24 and 48 h, and then lysed for western blot analysis. (c) same as (b) except that cells were harvested to isolate total RNA for qRT-PCR analysis on indicated genes using GAPDH as the internal control. (d) MAPKi-resistant melanoma cells A375R/1205LuR and their parental cells A375/1205Lu were transduced with control shRNA or TGFBR2 shRNAs, and then cells were lysed for western blot analysis on indicated proteins. Values in c represent the mean ± standard deviation of three independent experiments. Significance was determined by ANOVA one-way test, *p < 0.05, **p < 0.01, ***p < 0.001, ns, no significance. The quantitation of relevant western blot images in a, b and d was provided.

Full size image

LAMB3 regulates the expression of multiple MMPs

To understand how LAMB3 mediates the invasive phenotype of MAPKi-resistant cells, we performed RNA-sequencing on control and LAMB3-KO A375R cells. Comparative transcriptome analyzes revealed that multiple matrix metalloproteinases (MMPs), including MMP1, MMP2, MMP3 and MMP13 were significantly down-regulated in LAMB3 KO cells (Supplementary Fig. S6). The regulation of MMP1, 2, 3, 13 by LAMB3 was further validated by qRT-PCR and western blotting experiments except that MMP1 expression was not apparently affected by LAMB3-KO in 1205LuR cells (Fig. 5a, b). Consistently, MMP2, 3, 13 were more highly expressed in A375R and 1205LuR cells than in the parental cells (Fig. 5c, d), which correlated with LAMB3 expression (Fig. 1b). MMP1 had higher expression in A375R cells but not in 1205LuR cells. Moreover, overexpression of LAMB3 in parental melanoma cells resulted in upregulation of MMP1, 2, 3, 13 (Fig. 5e, f). Studies have shown that degradation of the basement membrane and extracellular matrix by MMPs is an essential step in melanoma cell migration, invasion, and metastasis [40,41,42]. Therefore, LAMB3 may exert its pro-invasiveness function by upregulating MMPs.

Fig. 5: LAMB3 regulates the expression of multiple MMPs.
figure 5

(a) MAPKi-resistant melanoma cells A375R and 1205LuR transduced with Crispr V2, Crispr/LAMB3 sgRNA#1 and LAMB3 sgRNA#4 were harvested to isolate total RNA for qRT-PCR analysis on indicated genes using GAPDH as the internal control. (b) same as (a) except that cells were harvested for western blot analysis. (c) MAPKi-resistant melanoma cells A375R/1205LuR and their parental cells A375/1205Lu were harvested to isolate total RNA for qRT-PCR analysis on indicated genes using GAPDH as the internal control. (d) same as (c) except that cells were lysed for western blot analysis. (e) A375 and 1205Lu cells transduced with control or LAMB3 over-expression lentiviruses were harvested for qRT-PCR analysis. (f) same as (e) except that cells were lysed for western blot analysis. Values in a, c and e represent the mean ± standard deviation of three independent experiments. Significance was determined by ANOVA one-way test, *p < 0.05, **p < 0.01, ***p < 0.001, ns, no significance. The quantitation of relevant western blot images in b, d and f was provided.

Full size image

LAMB3 regulates MMPs through FAK

KEGG pathway analyzes revealed that differential genes enriched in the focal adhesion pathway were reduced in two independent LAMB3-KO cell lines than Ctrl KO cells (Supplementary Fig. S7). FAK, the major kinase of focal adhesion signaling, has been implicated in cell growth, migration and invasion in melanoma, through promoting ERK1/2 signaling and MMP2 activity [34, 35]. Inspired by these studies, we hypothesize that LAMB3 may regulate MMPs expression through activating FAK signaling. In support of this, pY397-FAK levels were higher in MAPKi-resistant cells than in parental cells (Fig. 6a), which correlated with LAMB3 upregulation in resistant cells (Fig. 1b). LAMB3 ablation in resistant cells reduced pY397-FAK levels, while overexpression of LAMB3 in parental cells enhanced FAK phosphorylation at Y397 (Fig. 6a, b), indicating that LAMB3 can activate FAK.

Fig. 6: LAMB3 regulates MMPs through FAK.
figure 6

(a) MAPKi-resistant melanoma cells A375R/1205LuR and their parental cells A375/1205Lu transduced with Crispr V2, Crispr/LAMB3 sgRNA#1 or LAMB3 sgRNA#4 were harvested for western blot analysis. (b) Melanoma cells A375 and 1205Lu transduced with control or LAMB3 over-expression lentiviruses were lysed for western blot analysis on indicated proteins. (c) MAPKi-resistant melanoma cells A375R and 1205LuR were treated with 0, 1, 2, 5, 10 μM Defactinib for 24 h, and then cells were lysed for western blot analysis. (d) A375R control KO/LAMB3 KO cells and 1205LuR control KD/LAMB3 siRNA KD cells were transduced with or without HA-FAK WT lentiviruses, and then lysed for western blot analysis. (e) same as (d) except that cells were harvested for cell migration and invasion assay. After culturing for 24/48 h, cells were stained with crystal violet and counted by Image J software. Quantitation of cell migration and invasion numbers are shown. Values in e represent the mean ± standard deviation of three independent experiments. Significance was determined by ANOVA one-way test, *p < 0.05, ***p < 0.001. The quantitation of relevant western blot images in a-d was provided.

Full size image

To test whether LAMB3 regulates MMPs through FAK, we used Defactinib, a specific FAK inhibitor, to block FAK activity in A375R and 1205LuR cells. Defactinib reduced pY397-FAK levels and attenuated the expression of MMP3 and MMP13 in a dose-dependent manner (Fig. 6c), indicating FAK activity is required for the expression of MMP3 and MMP13 in MAPKi-resistant cells. Conversely, while LAMB3 KO reduced the expression levels of pY397-FAK, MMP3 and MMP13 in resistant cells, ectopic expression of HA-FAK could at least partially rescue these effects (Fig. 6d). Furthermore, overexpression of HA-FAK could also partially rescue the migration and invasion abilities of LAMB3-depleted resistant cells (Fig. 6e and Supplementary Fig. S8). Together, our results consistently demonstrate that LAMB3 regulates MMPs to confer cell invasiveness by modulating FAK activity.

LAMB3 depletion reduces metastatic potential of MAPKi-resistant melanoma tumors in vivo

The pro-invasiveness effect of LAMB3 was further investigated in vivo using a mouse lung-metastasis model. Luciferase-expressing A375 and A375R cells with or without LAMB3 KO were separately injected into mice through the tail vein and lung metastases were monitored over time. After 3 weeks, mice inoculated with A375R cells bore much more lung metastases than mice with A375 cells as determined by bioluminescence signals (Fig. 7a) and direct counts of metastatic nodules (Fig. 7b). In addition to lung metastases, A375R cells also formed metastases in oral cavity and limbs, which was barely detected in mice with A375 cells. Importantly, LAMB3 KO brought down the metastatic capacity of A375R cells to the level of the parental cells (Fig. 7a–c). These results demonstrate that MAPKi-resistant melanoma cells have higher metastatic potential than the parental cells in vivo and that LAMB3 is a major mediator of the improved invasiveness of the MAPKi-resistant melanoma cells.

Fig. 7: LAMB3 depletion reduces metastatic potential of MAPKi-resistant melanoma tumors in vivo.
figure 7

(a) A375 control KO, A375R control KO, A375R LAMB3 KO#1 and #4 cells transduced with HA-tagged Luciferase were injected into the tail vein of nude-mice (5×106 cells per mice). After 3 weeks, mice injected with 150 mg/kg luciferin were monitored by an IVIS bioluminescence imaging system. Representative bioluminescence images of mice with metastatic lesions are shown (left). Bioluminescent quantifications of metastatic lesions are shown (right). (b) Representative lungs with metastatic lesions (black arrow) are shown (left). Quantifications of metastatic lesions in lungs are shown (right). (c) H&E staining of representative lungs are shown. Metastatic lesions are indicated by black arrow. Values in a and b represent the mean ± standard deviation of three independent experiments. Significance was determined by ANOVA one-way test, ***p < 0.001.

Full size image

Altered SOX10/LAMB3/MMPs signaling in treatment-resistant human melanoma samples

To explore the clinical relevance of our findings, we compared expressions of SOX10, LAMB3 and MMPs in matched pre-treatment and relapsed human melanoma samples. Out of 4 sets of melanoma samples tested, two sets showed decreased SOX10 expression in relapsed versus pre-treatment samples, with a concomitant increase of LAMB3 and MMP3 expression (Supplementary Fig. S9). We also analyzed a published RNA-seq data (GEO database, GSE65185) comparing transcriptomes between pre-treatment and resistant human melanoma samples. Out of total 18 sets of matched melanoma samples, 11 sets showed decreased SOX10 expression in resistant versus pre-treatment samples. Among these, 4 sets showed enhanced LAMB3, MMP3 and EGFR (a target gene of the TGFb signaling pathway) expressions in resistant samples (Supplementary Fig. S10). Therefore, at least a subset of MAPKi-resistant melanoma is associated with decreased expression of SOX10 and increased expressions of LAMB3 and MMP3.

Discussion

A subset of melanoma develops adaptive resistance to MAPK inhibitors through phenotypic switching from a proliferative to an invasive state. While the molecular switch that controls the phenotype transition is well understood, how drug-tolerant melanoma cells gain improved invasiveness remains unclear. LAMB3 is an ECM protein implicated in the migration and invasion of many cancer cells but its role in melanoma hasn’t been elucidated [17,18,19,20,21]. We show that LAMB3 is upregulated in MAPKi-resistant melanoma cells and confers improved migration/invasion abilities as well as increased drug tolerance on melanoma cells in vitro. Furthermore, LAMB3 depletion efficiently inhibited the metastatic potential of MAPKi-resistant melanoma cells in a mouse lung metastasis model. These results indicate that LAMB3 is a key determinant of the enhanced invasiveness associated with MAPKi-resistant melanoma cells. ECM remodeling has been implicated in the regulation of tumor cell phenotype switching, migration/invasion and therapeutic responses [43]. In PTEN-null BRAF mutant melanoma cells, BRAFi-induced fibronectin production abrogates the drug response of tumor cells [44]. Furthermore, ECM stiffness modulated by collagen abundance contributes to melanoma cell phenotypic switching and therapeutic resistance through mechano-signaling [45, 46]. Together, these findings suggest ECM components may be potential targets for the treatment of drug-tolerant melanoma.

The phenotype switching of melanoma is analogous to the epithelial-mesenchymal-transition (EMT) process observed in many malignances. TGF- β signaling, a strong inducer of EMT and invasiveness has been shown to be required for melanoma phenotype switching and development of adaptive resistance to BRAFi [13, 47]. We show that TGF-β signaling is both necessary and sufficient for LAMB3 upregulation in melanoma cells. While the direct cue for TGF-β signaling activation remains obscure, evidence from others’ and this study suggests that SOX10 deficiency is associated with TGF-β signaling activation in melanoma cells [13, 24]. How SOX10low status leads to TGF-β signaling activation will be an interest of further study. A previous study demonstrates that SOX10 promotes cell migration and invasion via activating melanoma inhibitory activity (MIA), which appears to contradict our findings. This difference may result from differential roles of SOX10 played before and after phenotype switching. Melanoma cells of proliferative phenotype express high levels of SOX10 to maintain the differentiated state and promote cell proliferation, survival and migration/invasion through transcriptional control of oncogenic genes such as MITF, SAMMSONand MIA [23, 26, 48, 49]. After phenotype switching, melanoma cells enter a de-differentiated slow-cycling state and SOX10 deficiency become beneficial to sustain the de-differentiated state and improve invasiveness though activating the TGF-β signaling. Nevertheless, these studies indicate the complexity and diversity of gene regulation in melanoma cells during small molecule inhibitor treatment.

MMPs are key players in the invasion and dissemination of cancer cells for their abilities to degrade the ECM, activate latent cytokines/growth factors and modulate integrin signaling [40,41,42]. Sandri. et al. have found that melanoma cells that develops resistance to vemurafenib induces significant changes in the tumor microenvironment mainly by MMP2 upregulation, associated with enhanced metastatic potential [50]. Our data indicate LAMB3 positively regulates the expression of multiple MMPs in melanoma cells, which explains the link between LAMB3 upregulation and enhanced invasiveness observed in drug-tolerant melanoma cells. MMPs induction by LAMB3 appears to be mediated by FAK, a focal adhesion pathway kinase activated by the ECM/integrin signaling. Consistently, we found FAK inhibition or LAMB3/Integrin signaling blockade inhibited the pro-invasiveness effect of LAMB3 and FAK overexpression could rescue the effects of LAMB3 depletion. However, the transcription factor(s) responsible for the induction of MMPs downstream FAK still need further investigation.

In summary, we identify a SOX10low/TGF-β/LAMB3/FAK/MMPs signaling axis that governs the migration and invasion capacities of MAPKi-resistant melanoma cells. Our study also provides rationales for co-targeting components of this pathway (eg. LAMB3, TGFBR, FAK) to curb the metastasis of melanoma cells in targeted therapy.

Materials & Methods

Reagents

Vemurafenib, defactinib and LY2109761 were purchased from Selleck Chemicals LLC (Houston, TX); doxycycline from MCE (South Brunswick Township, NJ); Recombinant Human TGF-β1 from Pepro Tech (Cranbury, NJ); G418 sulfate from INALCO (San Luis Obispo, CA); Puromycin from InvivoGen (San Diego, CA); Crystal violet and luciferin from Beyotime (Shanghai, China).

Cell culture

1205Lu cells were generously gifted by Dr. Meenhard Herlyn at The Wistar Institute (Philadelphia, PA). M238 and M229 cells were provided by Dr. Antoni Ribas at University of California (Los Angeles, CA). A375 and HEK293FT cells were purchased from American Type Culture Collection (ATCC) (Manassas, VA). A375TR cells are sublines constructed with high Tet repressor expression [51]. BRAFi-resistant melanoma cell lines A375R and 1205LuR were obtained by treating their parental cells with 5 μmol L−1 vemurafenib through over 4 weeks. 1205Lu, 1205LuR, M238 and M229 cells were cultured in RPMI 1640 medium with 10% fetal bovine serum and 1% penicillin/streptomycin. A375, A375R, A375TR and HEK293FT cells were cultured in H-DMEM medium with 10% fetal bovine serum and 1% penicillin/streptomycin. All the cell lines were mycoplasma free.

Western Blotting

Melanoma cell lysates were separated on SDS-PAGE gels and then transferred to PVDF membranes. After blocking with 1% BSA for 1 h, the membranes were incubated with primary antibodies at 4 °C with shaking overnight. In the next day, the membranes were incubated with horseradish peroxidase-conjugated secondary antibodies for 1 h at room temperature. The protein blots were developed using an enhanced chemiluminescence western blotting detection kit (BioRad, Hercules, CA,). Antibodies against β-actin (A2066) and FAK (06–543) were purchased from Sigma-Aldrich (St. Louis, MO); Anti-SOX10 (A-2, #SC-365692), anti-LAMB3 (A-6, #SC-133178), anti-TGFBR2 (C-4, #SC-17791), anti-MMP-1 (3B6, #SC-21731), anti-MMP-2 (2C1, #SC-13594), anti-MMP-13 (C-3, #SC-515284), anti-EGFR (A-10, #SC-373746), anti-Integrin α3 (P1B5, #SC-13545) and anti-Integrin α6 (GOH3, #SC-19622) from Santa Cruz Biotechnology (Santa Cruz, CA); Anti-HA-tag (clone 6E2, #2367), anti-phospho-Smad2 (Ser465/467, clone 138D4, #138D4), anti-PDGFRβ (clone 28E1, #3169), anti-phospho-FAK (Tyr397, #3283 s) and anti-Smad2 (clone 86F7, #3122) from Cell Signaling Technology (Beverley, MA); Anti-MMP3 (#A11418) from ABclonal (Wuhan, China).

Quantitative RT-PCR

The TRIzol Reagent (TAKARA, Osaka, Japan) was used to extracted the total RNA in melanoma cells. The RNA was then reverse transcribed into cDNA using iScript™ cDNA Synthesis Kit (TAKARA). PCR reactions were performed using TB Green Premix Ex Taq II (TAKARA) and analyzed by a CFX Connect real-time PCR detection system (BioRad). Relative mRNA levels were calculated using the comparative Ct (ΔCt) method. Each Quantitation of mRNA levels represents data from three independent experiments. The indicated primer sequences are provided as below: GAPDH (forward, 5’-GGATTTGGTCGTATTGGG-3’; reverse, 5’- GGAAGATGGTGATGGGATT-3’), LAMB3 (forward, 5’-GCAGCCTCACAACTACTACAG-3’; reverse, 5’-CCAGGTCTTACCGAAGTCTGA-3’), EGFR (forward, 5’-TCCTCTGGAGGCTGAGAAAA-3’; reverse, 5’-GGGCTCTGGAGGAAAAGAAA-3’), PDGFRβ (forward, 5’-CAGGAGAGACAGCAACAGCA-3’; reverse, 5’-TGTCCAGAGCCTGGAACTGT-3’), MMP1 (forward, 5’-ATGAAGCAGCCCAGATGTGGAG −3’; reverse, 5’-TGGTCCACATCTGCTCTTGGCA −3’), MMP2 (forward, 5’-AGCGAGTGGATGCCGCCTTTAA-3’; reverse, 5’-CATTCCAGGCATCTGCGATGAG-3’), MMP3 (forward, 5’-CACTCACAGACCTGACTCGGTT-3’; reverse, 5’-AAGCAGGATCACAGTTGGCTGG-3’), MMP13 (forward, 5’-CCTTGATGCCATTACCAGTCTCC-3’; reverse, 5’-AAACAGCTCCGCATCAACCTGC-3’).

Production of Lentiviral Vectors and Cell Lines

The CRISPR/Cas9 system was used to generate the LAMB3 knock out BRAFi-resistant melanoma cell lines [52]. LAMB3 sgRNAs were provided as below: #1 5′-GGCCTGCTATCCACCTGTTG-3′, #4 5′-TGTGACTGCCACCAGCGCAT-3′.

Wild-type LAMB3, SOX10, FAK and Luciferase cDNA were cloned into pENTR™/D-TOPO vector (Thermo Fisher Scientific, Rockford, IL). SOX10 siRNA-resistant mutant was constructed using the Quickchange site-directed mutagenesis kit (Agilent Technologies Inc., Santa Clara, CA), pENTR™/HA-SOX10 WT was using as the template. Then the recombination assay was performed between the entry plasmids and pLentipuro/TO/V5-DEST or pLentiNEO/TO/V5-DEST using the LR Clonase II kit (Invitrogen, Carlsbad, CA) to generate lentiviral plasmids. Lentivirus were produced in HEK293FT cells, then melanoma cells were infected with concentrated lentivirus for 72 h before selection with 2 μg mL−1 puromycin or 500 μg mL−1 G418.

For SOX10 and TGFBR2 shRNA constructs, DNA oligonucleotides were annealed and cloned into pENTR/H1/TO plasmid (Thermo Fisher Scientific). The shRNA sequences were provided below: SOX10 (#1 CCGTATGCAGCACAAGAAA, #2 GTATGCAGCACAAGAAAGA), TGFBR2 (#1 GCTTCTCCAAAGTGCATTA, #2 GACCTCAAGAGCTCCAATA). The entry shRNA cassettes were recombined into pLentipuro/H1TO/block it/DEST destination vector. Lentiviruses were produced and melanoma cells were transduced as described above.

RNA Interference

Melanoma cells in 6-well plate were transfected with 12.5 nmol small-interfering RNAs (siRNA) and 2 μL Lipofectamine RNAiMAX transfect reagents (Thermo Fisher Scientific, Rockford, IL) for 72 h. siRNAs for SOX10 (#1 CCGUAUGCAGCACAAGAAA; #2 GUAUGCAGCACAAGAAAGA), LAMB3 (#1 CCAGCGAGGCUACUGUAAU; #2 GUGUGUGCAAGGAGCAUGU) and non-targeting siRNA control (5’-UUCUCCGAACGUGUCACGU-3’) were purchased from Shanghai GenePharma Co., Ltd. (Shanghai, China).

Annexin V/PI Apoptosis Assay

Melanoma cells were collected and washed by cold PBS twice. Cells were then stained using the APC Annexin-V-PI Apoptosis Detection Kit (Biolegend, San Diego, CA, USA) according to the manufacturer’s protocol. After staining, cells were detected by flow cytometry on a CytoFLEX system (Beckman Coulter, Indianapolis, IN). The data were analyzed with Flowjo software (Three Star, Inc., Ashland, OR).

Extracellular Staining

Melanoma cells were washed twice by PBS and blocked with 0.5% BSA for 15 min on ice. Then the control IgG or primary antibody (α3 or α6 antibodies, 10 ng μL−1) were added into cells for 1 h on ice. After washing by PBS, cells were resuspended in secondary antibody conjugated with fluorophore for 30 min on ice. Cells were fixed in methanol free formaldehyde for 30 min and then detected by flow cytometry on a CytoFLEX system. The data were analyzed with Flowjo software.

Cell Migration and Invasion Assay

Polycarbonate Membrane Inserts with 8.0 µm Pore (Corning, Lowell, MA, USA) were used in the migration assay. Melanoma cells were resuspended in H-DMEM medium with 1% fetal bovine serum, and then add into the upper compartment of transwell inserts. Normal H-DMEM medium with 10% fetal bovine serum was added into the lower compartment. After culturing about 24 h, cells were washed and stained by crystal violet (Beyotime). Migrated cells were then counted by Image J software.

For cell invasion assay, Matrigel matrix (Corning) was pre-coated on the transwell inserts. Cells were cultured for over 48 h, and then stained by crystal violet (Beyotime).

Wound Healing Assay

Melanoma cells were plated in the two wells of silicone culture-inserts (Ibidi, Fitchburg, Wisconsin). In the next day, after removing the inserts, microphotographs of cell-free gaps were taken by an Inverted Microscope System (Nikon ECLIPSE Ti, Tokyo, Japan) as before healing. After culturing cells for about 16 h, microphotographs of cell-free gaps were taken again as after healing. The areas of cell-free gaps at 0 h and 16 h were calculated by Image J software. The wound closure rates were calculated using the following formula: wound closure rate = (S1 − S2) S1–1, where S1 represents the area of cell-free gap at 0 h; S2 represents the area of cell-free gap at 16 h.

Tail Vein Assay of Melanoma Metastasis

Five-week-old female BALB/c nude mice (Shanghai SLAC Laboratory Animal CO. LTD, Shanghai, China) were randomly divided into 4 treatment groups (10 mice per group). A375 HA-luciferase cells and A375R HA-luciferase cells carrying Crispr Ctrl KO, LAMB3 KO#1, #4 lentiviruses were injected into tail vein of mice (5× 106 per mouse). Mice injected with A375R Ctrl KO, LAMB3 KO#1, #4 cells were treated intraperitoneally with vemurafenib (30 mg kg−1) every other day. After 3 weeks post-injection, mice were imaged by IVIS Spectrum In Vivo Imaging System (PerkinElmer, Waltham, MA). Then mice were euthanized and removed lungs. The numbers of metastatic tumors were counted using a dissecting microscope. The lungs were then paraffin embedded for pathological examination of H&E slides.

All animal protocols were approved by the Institutional Animal Care and Use Committee of Xi’an Jiaotong University (no. 2019-0032). The investigator was not blinded to the group allocation during the experiment.

Immunohistochemical staining (IHC)

Four pairs of matched human melanoma samples, each of which comprises a pre-treatment sample and a relapsed (post-treatment) sample were tested by immunohistochemical (IHC) in Y&K biotech Inc (Xi’an, China). All tumor samples have obtained the informed consent from patients. The protocols were approved by the Ethics Committee of Xi’an Jiaotong University.

Statistical Analysis: Statistical analysis was performed using Student’s two-tailed t-test for two groups’ comparison, and One-Way ANOVA test for multiple groups. The software Graphpad Prism7 was used for statistical analyses. A p value of <0.05 was considered statistically significant.