Abstract
Cutaneous squamous cell carcinoma (cSCC) is the most common metastatic skin cancer. The prognosis of patients with metastatic cSCC is poor emphasizing the need for new therapies. We have previously reported that the activation of Ras/MEK/ERK1/2 and transforming growth factor β (TGF-β)/Smad2 signaling in transformed keratinocytes and cSCC cells leads to increased accumulation of laminin-332 and accelerated invasion. Here, we show that the next-generation B-Raf inhibitor PLX8394 blocks TGF-β signaling in ras-transformed metastatic epidermal keratinocytes (RT3 cells) harboring wild-type B-Raf and hyperactive Ras. PLX8394 decreased phosphorylation of TGF-β receptor II and Smad2, as well as p38 activity, MMP-1 and MMP-13 synthesis, and laminin-332 accumulation. PLX8394 significantly inhibited the growth of human cSCC tumors and in vivo collagen degradation in xenograft model. In conclusion, our data indicate that PLX8394 inhibits several serine-threonine kinases in malignantly transformed human keratinocytes and cSCC cells and inhibits cSCC invasion and tumor growth in vitro and in vivo. We identify PLX8394 as a potential therapeutic compound for advanced human cSCC.
Introduction
Epidermal keratinocyte-derived cutaneous squamous cell carcinoma (cSCC) is the most common metastatic skin cancer with increasing incidence worldwide [1, 2]. The molecular basis of cSCC progression from premalignant lesion, actinic keratosis, to cSCC in situ and finally to invasive cSCC is incompletely understood [2], and the prognosis of patients with advanced disease is poor [3, 4].
Laminin-332 is a major protein component in cutaneous basement membrane, and in many cancers, including cSCC, its high expression level correlates with tumor invasiveness [5,6,7]. We have recently shown that hyperactive H-Ras and stromal fibroblast-related induction in TGF-β signaling co-operate in the synthesis of laminin-332 in cSCC cells, leading to enhanced cancer cell invasion [8]. H-Ras is mutated in ~10% of cSCCs [9,10,11,12]. Mitogen-activated protein kinases (MAPKs) can also be activated by other mechanisms. Recently, a meta-analysis of cSCC driver mutations was performed and 30 genes were appointed to operate in signaling pathways known to be disrupted in cSCC [11]. Mutations that activate the MAPK and/or phosphoinositide 3-kinase (PI3K) pathways occurred in 31% of the tumors [11]. Accumulation of laminin-332 can be prevented by inhibition of either Ras or TGF-β signaling, which consequently decrease cSCC invasion [8]. The TGF-β signaling pathway may also promote the invasion of cSCC via activation of matrix metalloproteinase (MMP) production [13].
These previous observations encouraged us to search for new strategies for blocking the activation of Ras and TGF-β signaling pathways in cSCC. Ras is known to act via activation of ERK1/2 MAPK pathway, composed of MAPK kinase kinases (Rafs), MAPK kinases (MEKs), and MAPKs (ERKs). B-Raf inhibitors, such as dabrafenib and vemurafenib, are used in the treatment of V600EB-Raf positive cancers, including metastatic melanoma and non-small-cell lung cancer [14,15,16] alone or in combination with a MEK1/2 inhibitor (trametinib). Undesirably, dabrafenib and vemurafenib can activate ERK1/2 pathway in cells that exhibit high Ras activity [17,18,19]. Recently, next-generation Raf inhibitors have been developed. PLX7904 and its optimized analog PLX8394 do not elicit paradoxical ERK1/2 activation [20,21,22]. PLX8394 has been shown to suppress the growth of B-Raf mutant lung cancer [23] and melanoma in vivo [24]. PLX8394 is currently in phase II clinical trial in B-Raf mutated cancers (clinicaltrials.gov NCT No: NCT02428712). PLX8394 inhibits V600EB-Raf in low nanomolar concentrations, but in previously published in vivo mouse experiments after oral administration of 150 mg/kg, its serum concentrations have been high (Cmax: 164 μM after 1 h; T1/2: 3.5 h) without obvious toxicity [23].
Here, we have used three-dimensional spheroid cultures of transformed RT3 keratinocytes which harbor wild-type B-Raf and hyperactive Ras, and spheroid cocultures of RT3 cells and skin fibroblasts. We show that low micromolar concentrations of PLX8394, previously shown to be non-toxic in in vivo experiments [23] inhibit phosphorylation of Smad2 and TGF-β Receptor II (TGFβRII) and p38 activation. Consequently, PLX8394 decreased MMP-1 and MMP-13 expression and laminin-332 accumulation by RT3 cells and inhibited RT3 cell invasion. Orally administered PLX8394 inhibited growth and invasion of human cSCC in vivo in xenograft model. Thus, we identify novel serine-threonine kinase targets for PLX8394 that can be blocked by low micromolar concentrations. In general, our results suggest that serine-threonine kinase inhibitors with optimal target spectrum can be potential new therapeutic compounds for advanced cSCC.
Results
Serine-threonine kinase PLX8394 inhibits laminin-332 expression in transformed keratinocytes
We have previously reported accumulation of laminin-332 in the invasive front of human cSCC tumors [8]. In this study, we used spontaneously immortalized, nontumorigenic human keratinocyte cell line (HaCaT) and its H-Ras-transformed metastatic derivative RT3. These cells have been widely used to study the progression of cSCC from benign lesions to malignant tumors [25,26,27] and they show high constitutive activation of ERK1/2 [28]. We showed that H-Ras knockdown decreases laminin-332 expression in RT3 cells when they are cultured as 3D spheroids [8]. Furthermore, in in vitro assays laminin-332 expression correlated with cell invasion [8].
Based on these previous observations, we now tested chemical inhibitors for signaling proteins in the ERK1/2 pathway, i.e. downstream effectors of Ras. Laminin-332 expression was determined by western blotting. HaCaT and RT3 cells were treated with the inhibitors for 24 h in monolayers before they were allowed to form 3D spheroids. The spheroids were allowed to grow for 72 h and DMSO treated cells were used as controls. In accordance with our previous results [8], western blotting showed that HaCaT cells do not express significant amounts of laminin-332, while this laminin is actively produced by RT3 cells (Fig. 1A). Treatment of RT3 cells with MEK inhibitors PD98059 (20 μM) and trametinib (20 nM) decreased ERK1/2 activation and consequently also laminin-332 synthesis (Fig. 1A). This is in accordance with our previous results showing that the inhibition of ERK1/2 activation decreases laminin-332 expression [8]. In accordance with previous reports [29, 30], Raf inhibitor dabrafenib (50 nM) caused paradoxical MAPK pathway activation, observed as increased (about 4-fold) ERK1/2 phosphorylation. Concomitantly, dabrafenib increased laminin-332 synthesis (Fig. 1A). Interestingly, V600EB-Raf inhibitor PLX4720 (10 μM) decreased laminin-332 γ2 chain expression (~50%) when compared to DMSO treated control sample (Fig. 1A) and a potent inhibitory effect of total laminin-332 protein (~90%) was detected with next-generation Raf inhibitor PLX8394 (10 μM). PLX8394 treatment did not lead to paradoxical increase in phosphorylated ERK1/2 level, instead we observed a minor decrease (~10%) in ERK1/2 phosphorylation, which could not explain the notable downregulation in laminin-332 synthesis (Fig. 1A).
PLX8394 inhibits TGF-β signaling both in 3D spheroids and in 2D monolayer cultures of RT3 cells
Since the significant decrease in laminin-332 expression in RT3 cells after PLX4720 and especially after PLX8394 treatment was not associated with similar inhibition in ERK1/2 phosphorylation, we analyzed their effects on the activity of the Smad signaling pathway. To activate TGF-β signaling pathway, RT3 cells were cocultured with primary human dermal fibroblasts. RT3 cells were treated with PLX8394 (10 μM) or PLX4720 (10 μM), while dabrafenib (50 nM) was used as a control. The cells were treated with the inhibitors 24 h in monolayers before spheroid formation with primary skin fibroblasts and the spheroids were allowed to grow for 24 h and 48 h. At 48 h time point western blotting showed a decrease in the phosphorylation of Smad2 after both PLX4720 and PLX8394 treatment (Fig. 1B). However, PLX8394 was clearly more potent inhibitor of Smad2 activation and it also decreased SMAD2 phosphorylation (~80%) in 24 h time point. The most potent inhibition of Smad2 phosphorylation and laminin-332 accumulation was observed after 48 h (Fig. 1B). PLX8394 slightly decreased ERK1/2 phosphorylation at 48 h time point (Fig. 1B), but the inhibition of TGF-β signaling pathway was noted as the main mechanism, by which PLX8394 prevents laminin-332 accumulation.
To further study the effect of PLX8394 on TGF-β signaling, we treated HaCaT and RT3 cells with PLX8394 (10 μM), or with DMSO as a control, in 2D cell culture condition for 24 h. The cells were then treated with TGF-β (10 ng/ml, 30 min) and harvested for western blotting. The results showed that PLX8394 totally prevented Smad2 phosphorylation both in HaCaT and RT3 cells (Supplementary Fig. S1A). To examine whether PLX8394 could affect TGF-β induced Smad activation also in 3D spheroids, HaCaT and RT3 cells were first treated with DMSO or PLX8394 (10 μM) in 2D cell culture condition for 24 h, followed by 3D spheroid formation. Two-day-old spheroids were treated with TGF-β (10 ng/ml) for 4 h. Western blotting showed that also in spheroids, PLX8394 inhibited TGF-β induced Smad2 activation in RT3 cells (Supplementary Fig. S1B). Laminin-332 expression was decreased by PLX8394 both in 2D and 3D cultured samples (Supplementary Fig. S1A, B). To conclude, these results demonstrate that PLX8394 inhibits TGF-β signaling and concomitantly decreases laminin-332 accumulation in RT3 cells.
To verify the results obtained from RT3 cells, we treated a panel of primary and metastatic cSCC cell lines with PLX8394 for 3 days. The cells were first treated with DMSO or PLX8394 (10 μM) in 2D cell culture condition for 24 h, followed by 3D spheroid formation with primary human skin fibroblasts. The results showed that PLX8394 inhibited laminin-332 synthesis in all primary (UT-SCC-12A, UT-SCC-91A, UT-SCC-105, UT-SCC-118) and metastatic (UT-SCC-7, UT-SCC-59A, UT-SCC-115) cell lines (Fig. 1C). Phosphorylation of Smad2 was also clearly decreased in all PLX8394 treated samples compared to control, whereas phosphorylation of ERK1/2 was decreased in some of the cell lines (Fig. 1C). These results indicated that PLX8394 was a potent inhibitor of TGF-β signaling in human squamous cell carcinoma.
Next, to determine the optimal PLX8394 concentration for inhibition of Smad2 phosphorylation, RT3 cells were exposed to increasing concentrations (0–10,000 nM) of PLX8394 for 4 h or 18 h, followed by TGF-β treatment (10 ng/ml, 30 min). The results showed that 10 μM concentration of PLX8394 totally inhibited TGF-β induced Smad2 activation in RT3 cells both after 4 h (Fig. 1D) and 18 h (Fig. 1E) inhibitor treatment. Inhibition by a lower concentration of 1 μM was also evident (~10% at 4 h and ~60% at 18 h; Fig. 1D, E). In addition to RT3 cells, we tested the effects of different PLX8394 concentrations on A2058 melanoma cells which harbor V600EB-Raf mutation. Western blotting showed that also in A2058 cell line, 10 μM PLX8394 concentration blocked Smad2 activation both after 4 h (Supplementary Fig. S1C) and 18 h (Supplementary Fig. S1D) of inhibitor treatment. The efficacy of the inhibitor was verified by analyzing p-ERK1/2 levels in A2058 melanoma cells exposed to different concentrations (0–10,000 nM) of PLX8394. The results showed that 100 nM concentration of PLX8394 inhibited p-ERK1/2 levels after 4 h (Supplementary Fig. S1E) and after 18 h (Supplementary Fig. S1F) of inhibitor treatment, which is in accordance with previously published results obtained with V600EB-Raf mutated melanoma cell line [31]. Altogether, our results show that 1-10 μM PLX8394 inhibits Smad2 phosphorylation both in cells harboring wild-type B-Raf or V600EB-Raf mutation.
PLX8394 inhibits TGF-β signaling by affecting TGF-β type II receptor kinase activity
A dimeric TGF-β molecule activates its downstream signaling cascades by binding to two TGF-β receptor type II proteins (TGFβRII). Consequently, TGFβRIIs activate type I receptors (TGFβRI). Both TGFβRIs and TGFβRIIs are protein serine-threonine kinases [32]. To further investigate the mechanism by which PLX8394 decreases Smad2 phosphorylation, we treated RT3 cells with increasing concentrations (0–10,000 nM) of PLX8394 for 24 h, followed by spheroid formation with primary human skin fibroblasts. The spheroids were allowed to grow for 72 h before harvesting for western blotting. The analyses showed a slight (~20%) decrease in the phosphorylation of TGFβRII after 10 μM PLX8394 treatment, along with completely inhibited Smad2 activation and laminin-332 expression (Fig. 2A). At this time point (72 h) 10 μM PLX8394 also decreased the amount of phosphorylated ERK1/2 (Fig. 2A).
To study the time-dependency of PLX8394 action, RT3 cells were treated with 10 μM PLX8394 for 24 h in monolayer culture before the formation of 3D spheroids with primary skin fibroblasts. The spheroids were grown from 1 to 5 days before harvesting for western blotting. DMSO treated control samples showed strong laminin-332 expression starting from day two (Fig. 2B). PLX8394 prevented laminin-332 expression from day 2 to 5 (Fig. 2B and Supplementary Fig. S2A). The presence of phosphorylated Smad2 was only detected in 1-day-old PLX8394 treated samples, whereas control samples showed continuous presence of phosphorylated Smad2 during the first 4 days. The experiment was repeated three times and the statistical significance of the differences was tested. The effect of PLX8394 on Smad2 phosphorylation was most evident in early time points and statistically significant on days 1 and 2 (Fig. 2B and Supplementary Fig. S2A). In accordance with the experiment shown in Fig. 2A, PLX8394 inhibited TGFβRII phosphorylation in all time points starting from day 1 (Fig. 2B and Supplementary Fig. S2A). To conclude, the results shown in Figs. 1,2, in RT3/fibroblast coculture spheroids PLX8394 completely inhibits laminin-332 production from day 2. On day 1, it is already possible to see inhibition in the phosphorylation of Smad2 and TGFβRII, indicating that in our 3D coculture model, inhibition of TGF-β signaling results in prevention of laminin-332 synthesis after 24 h of PLX8394 treatment.
Binding of TGF-β to TGFβRII leads to the recruitment and activation of TGFβRI (ALK-5), which then phosphorylates Smad2 [32]. To examine the putative role of TGFβRI in the PLX8394-related inhibition of the TGF-β signaling pathway, we used adenoviral vector coding for constitutively active mutant of ALK5 (RAdCA-ALK5) or empty vector (RAd66) as a control. RT3 cells were either left uninfected or infected for 48 h with RAdCA-ALK5 (100 MOI) or with RAd66 (100 MOI) as a control. The cells were then treated with 10 μM PLX8394 for 24 h and harvested for western blotting. As expected, RadCA-ALK5 infection increased Smad2 phosphorylation (Fig. 2C). In RadCA-ALK5-infected cells PLX8394 did not decrease p-Smad2 levels (Fig. 2C). When the cells were treated with TGF-β, it was possible to detect additional increase in Smad2 phosphorylation that could be inhibited by PLX8394 treatment (Supplementary Fig. S2B). These observations suggest that constitutively active TGFβRI (ALK-5) is not a target of PLX8394. However, we cannot exclude the possibility that PLX8394 still inhibits wild-type TGFβRI.
Finally, we analyzed the expression of Smad7, an inhibitory Smad, that attenuates intracellular TGF-β signaling through physical interaction with activated TGF-β receptor I and by blocking phosphorylation and activation of Smad2 [33]. In control samples, the strongest Smad7 expression was detected without TGF-β treatment and again in 4-day old TGF-β treated samples, which was opposite to p-Smad2 levels in control samples (Fig. 2D). In PLX8394 treated samples, Smad7 expression was weak and did not change over time. Thus, the results clearly disprove the idea that PLX8394 could regulate Smad2 phosphorylation via Smad7 activation.
p38 MAPK regulates laminin-332 expression in RT3 cells
Smad proteins are the main mediators of TGF-β related effects. However, in addition to the Smads, TGF-βs also activate non-Smad signaling pathways, including p38 MAPKs [13, 34]. To test whether p38 MAPK pathway participates in laminin-332 synthesis in RT3 cells, we treated cells with p38 inhibitors SB203580 or BIRB796 for 24 h in monolayer cultures before they were allowed to form 3D spheroids with primary human skin fibroblasts. The spheroids were grown for 72 h. Western blotting showed that both inhibitors significantly decreased the expression of β3- and γ2-chains of laminin-332 (Fig. 3A and Supplementary Fig. S3A). Thus, in our model system laminin-332 synthesis is also regulated by p38 MAPK pathway.
We next tested whether PLX8394 is able to inhibit p38 pathway. In this purpose, we first analyzed the phosphorylation of p38. RT3 cells were treated with 10 μM PLX8394 for 24 h in monolayer culture prior to formation of 3D spheroids with primary human skin fibroblasts. The western blot results demonstrated that PLX8394 decreased phosphorylation of p38 in 3-day-old spheroids (Fig. 3B and Supplementary Fig. S3B). To further analyze PLX8394 participation in p38 signaling inhibition, we examined the phosphorylation of CREB, a down-stream target of p38. Cocultured spheroids were grown for 1–5 days before harvesting for western blotting. Spheroids grown for 3 days showed a slight decrease in CREB phosphorylation, and the phosphorylation was significantly decreased at day 5 (Fig. 3C and Supplementary Fig. S3C). This confirms that PLX8394 inhibits p38 pathway starting from day 3. To conclude, p38 pathway is yet another regulator of laminin-332 expression and the inhibition of this signaling mechanism by PLX8394 may affect laminin-332 accumulation in RT3/fibroblast spheroid model, at least in late time points.
To further study PLX8394 as an inhibitor of p38 pathway, we treated the cells with inflammatory cytokines and osmotic stress, all known activators of p38. Monolayer cultures of RT3 cells were treated with tumor necrosis factor α (TNF-α, 10 ng/ml, 30 min), interleukin 1β (IL-1β, 10 ng/ml, 30 min), and sorbitol (400 mM, 1 h). The activity of p38 was analyzed by western blotting of p-CREB (Fig. 3D). All three treatments increased CREB phosphorylation, but PLX8394 could only slightly inhibit p38 activation after sorbitol treatment and had no effect on CREB activation by TNF-α or IL-1β (Fig. 3D). Thus, PLX8394 does not appear to be a universal inhibitor of p38 pathway.
PLX8394 inhibits RT3 cell invasion through collagen I
To test whether PLX8394 affects cell invasion, we treated RT3 cells with 10 μM PLX8394 for 24 h prior to spheroid formation with skin primary fibroblasts. Three-day-old spheroids were embedded in collagen I and the invasion was followed by confocal microscope every 24 h during 5 days. Confocal images of the spheroids showed that RT3 cells treated with PLX8394 were not able to invade as efficiently as control cells that were treated with DMSO (Fig. 4A). Quantification showed that treatment with PLX8394 significantly decreased RT3 cell invasion out of coculture spheroids already after 48 h when compared to DMSO treated control samples (Fig. 4B). Fibroblast invasion from the same spheroids was not affected by PLX8394 treatment (Supplementary Fig. S4A). Western blotting of the same samples that were used in invasion assay showed that the inhibitor treatment significantly reduced p-Smad2 levels and laminin α3, β3, and γ2 chain synthesis (Supplementary Fig. S4B and C)…
