1. Introduction
Melanoma, the deadliest form of human skin cancer, originates from melanocytes and exhibits a high tendency to metastasize, accounting for approximately 90% of skin cancer-related mortality [
1]. Melanoma represents 1.7% of all global cancer diagnoses and ranks as the fifth most common cancer in the United States [
2]. While melanoma is highly treatable when confined to its primary site, metastatic melanoma presents a grim outlook, with a median survival of only approximately six months [
3,
4,
5]. Furthermore, systemic therapies currently employed in patients with metastatic melanoma exhibit varying response rates, with rapid development of tumor resistance observed in the majority of patients [
3,
6,
7]. While surgical intervention remains the primary treatment for melanoma, recent breakthroughs in immunotherapy and targeted molecular therapies for metastatic melanoma offer significant potential [
8].
Melanoma treatment, incorporating surgery, chemotherapy, immunotherapy, and radiotherapy, faces challenges due to melanoma resistance, primarily attributed to melanin production. Recent focus has been on targeting the mitogen-activated protein kinases (MAPK) pathway to overcome resistance and improve therapeutic outcomes [
9,
10]. Inhibiting the BRAF or MEK portions of the MAPK pathway exerts primarily cytostatic effects compared to chemotherapy. MEK inhibition in BRAF mutant cells may induce apoptotic and cytostatic impact [
11]. Early noninvasive biomarkers are crucial to assess target modulation and treatment efficacy due to the potential absence of tumor shrinkage. BRAF, activated by RAS, is highly expressed in melanocytes, neural tissue, testes, and hematopoietic cells. Phosphorylated BRAF activates MEK (a kinase component of the MAPK pathway), which in turn activates ERK (MAPK) by phosphorylation and thus stimulates growth and transformation [
12]. Unfortunately, melanoma cells have a hypermutable genome leading to tumor resistance by responding to the blockage of the MAPK pathway by rerouting to an alternate path. Therefore, most patients develop resistance to targeted therapy within weeks to months of initiation of treatment.
Melanoma often exhibits a glycolytic phenotype driven by the constitutive activation of the
BRAF gene mutation (
Figure 1) [
13]. This activation increases glucose uptake and glycolysis, mediated by the MAPK pathway and its induction of hypoxia-inducible factor 1α (HIF-1α), a key regulator of glycolytic activity [
14,
15]. In response to varying energy demands and environmental cues, melanomas demonstrate metabolic plasticity, adjusting between glycolysis and oxidative phosphorylation (OXPHOS) [
16]. Glutamine could also become a primary energy source, facilitated by heightened glutaminolysis and upregulation of glutamine transporters. Furthermore, resistance to BRAF and MEK inhibition primarily arises from metabolic adaptations within the MAPK signaling pathway [
16,
17].
Published research, including ours, shows that magnetic resonance spectroscopy (MRS) is a noninvasive avenue for monitoring metabolic changes in melanoma models following diverse anticancer treatments [
10,
11]. We demonstrated that MRS discerns significant changes in metabolic signatures in successfully treated melanoma models with dabrafenib, a BRAF inhibitor, showcasing early therapy-related changes in metabolomics, pH, and bioenergetics [
10]. This elucidation underscores the potential of MRS in delineating early-treatment metabolic responses to targeted therapies. Notably, the MRS availability in clinical MRI systems guarantees the translation of the metabolic signatures as possible biomarkers of treatment efficacy, supporting personalized therapeutic approaches in clinical melanoma management.
This report examines four human melanoma cell lines characterized for their BRAF sensitivity: a wild-type (WM3918), a BRAF resistance mutant-type (WM983BR), and two BRAF-sensitive mutant-types (WM983B and DB-1). These studies tested the cell lines’ response to trametinib, a MEK kinase inhibitor targeting the MAPK signaling pathway (
Figure 1). Inhibition of the MEK kinase is known to elicit treatment responses in melanoma models, as the hyperactive BRAF precedes MEK in the MAPK pathway. To discern trametinib-induced metabolic changes, we grew the melanoma cell lines in athymic nude mice xenografts and cell cultures. We employed in vivo
1H and
31P MRS noninvasively in the xenografts alongside other in vitro analytical methods in the cell cultures to assess the metabolic response of human melanoma.
4. Discussion
BRAF/MEK inhibition has become a standard-of-care option for
BRAFV600-mutated melanoma. Dabrafenib, a BRAF inhibitor, and trametinib, a MEK inhibitor, have received approval from the U.S. Food and Drug Administration for treating BRAF-driven melanoma [
21]. BRAF and MEK are crucial components of the MAPK signaling pathway, which regulates various essential cell functions, including cell growth and apoptosis [
26,
27,
28]. We previously reported significant metabolic changes in preclinical melanoma models with mutated
BRAF dependency sensitive to dabrafenib [
10]. In the present work, we expanded these observations, demonstrating again the significant metabolomic phenotype variability in melanomas with high dependency on mutated
BRAF [
10,
29,
30]. However, different from our previous work inhibiting the altered BRAF protein directly using dabrafenib [
10], in the present work, we use trametinib to inhibit MEK, which is a subsequent step after BRAF in the MAPK pathway (
Figure 1). Thus, we reduced MAPK signaling by indirectly affecting the hyperactive BRAF protein, inhibiting MEK. Our findings strongly suggest a tight relationship between altered tumor metabolism and MAPK inhibition. Molecularly targeted agents inducing selective MEK inhibition play a crucial role in melanoma by inhibiting the abnormal MAPK signaling pathway and restoring apoptosis [
31,
32,
33]. Hence, maximizing the benefits of inhibiting the MAPK pathway is crucial to achieving a high objective therapy response. Our noninvasive metabolic measurements (i.e.,
1H and
31P MRS), which could be translatable to the clinical arena, may be valuable in determining the metabolic effect of BRAF and MEK inhibitors and, thus, help as clinical biomarkers of response to these therapies.
We evaluated glucose uptake by isolated tumor cells, considering the correlation of this uptake with lactate production in cancer cells due to the Warburg effect [
34]. Our results in
Figure 2 demonstrate that trametinib reduces glucose uptake in all melanoma cell lines. However, the two sensitive cell lines showed reduced lactate production with trametinib, while the resistant ones significantly increased it. The decreased glucose uptake with increased lactate production may be due to a different energy source the resistant cell lines utilize besides glucose (e.g., glutaminolysis, lipolysis, etc.). As an aside, these results could explain, at least in part, the difficulties of predicting response by positron emission tomography (PET) [
35,
36,
37]. PET is the modality for staging patients with solid malignancies, including melanoma [
37,
38,
39]. For these purposes, the tumor uptake of the radiotracer
18F-2-fluoro-2-deoxy glucose (FDG) is followed by PET. FDG is an analog of glucose that cannot be metabolized further than the first step of glycolysis. As trametinib reduces the glucose uptake in all melanoma cell lines tested (
Figure 2), an FDG-PET exam might not be able to demonstrate differences to predict response. In contrast,
1H MRS shows lactate and alanine differences (
Figure 5) associated with response and may be assessed noninvasively in melanoma patients.
Our results with the OCR and ECAR measurements (
Figure 3 and
Table 1) demonstrate that the wild-type cell line WM3918 under trametinib treatment showed no significant differences in OCR and the smallest, although statistically significant, decrease in ECAR of all cell lines compared to its control group. As expected, the OCR/ECAR did not change with trametinib therapy in WM3918. In comparison, the remaining cell lines showed significant but variable changes in OCR and ECAR. Trametinib proportionally increased both parameters in the resistant mutant WM983BR cell line, maintaining the OCR/ECAR constant at the lowest level of all cell lines. Both mutant cell lines, resistant (WM983BR) and sensitive (WM983B), had an increased OCAR but differed in ECAR. These changes demonstrate an upregulation of respiration in both cell lines, with an increase in glycolytic activity for WM983BR and a decrease in WM983B. Finally, the three parameters decreased in the remaining sensitive mutant cell line, DB-1. The trametinib-related changes in DB-1 demonstrate a decrease in respiratory and glycolytic activities, with a reduction in overall energy production.
Furthermore, WM983B and DB-1 cell lines are BRAF-positive mutants, showing significant shifts in cell energy phenotype. These shifts reveal that MEK inhibition via trametinib impacts glycolytic activity in WM983B and DB-1. The trametinib-related decrease in ECAR in the sensitive mutant cell lines WM983B and DB-1 established them as less glycolytic (
Figure 3).
Figure 3 also shows that WM983B cells relied more on respiration with trametinib treatment and had the second-largest change in energy phenotype. DB-1, the most energetic melanoma line, was the most significantly affected and had the most prominent energy phenotype shift with treatment (
Figure 3 and
Table 1). These facts were corroborated by the decreased significantly glucose uptake and lactate production shown in
Figure 2 and the reduced tumor lactate values in vitro and in vivo depicted in
Figure 4 and
Figure 5, respectively. However, given that WM983B and WM983BR have a higher OCR with trametinib, they differ from DB-1, which has a lower OCR. Although we expected to see a reflection of these results in the bioenergetic status of the cell lines, the xenografts of these three cell lines had a trametinib-related increase of β-NTP/Pi (
Figure 6). We theorize that despite the decreased cellular respiration found in DB-1 in the controlled in vitro OCR studies (
Figure 3), its increased bioenergetic status in vivo could be due to the inhibition of the hyperactive MAPK pathway and the concomitant reduction of the Warburg effect. Under these in vivo conditions, instead of synthesizing lactate, glycolytic intermediates, mainly pyruvate, could be available for cellular respiration, albeit the fact that respiration seems to be deficient in DB-1 in the in vitro studies (
Figure 3). The inhibition of the Warburg effect is supported by the substantial reduction in lactate production shown in
Figure 2 and the reduction of tumor lactate content depicted in
Figure 4 and
Figure 5 for DB-1 and WM983B. However, it is also possible that the DB-1 inconsistencies may be due to the different experimental conditions of cultured tumor cells vs. in vivo xenografts.
The present results complement the ones we obtained on the effect of BRAF inhibition on the same melanoma models [
10]. WM3918, the wild-type melanoma cell line, and WM983BR, the BRAF inhibitor-resistant mutant line, showed no significant shifts in OCR/ECAR ratio with trametinib treatment, indicating that the effect of trametinib was minimal. Other studies on MEK signaling inhibition in wild-type melanoma have shown similar findings [
40]. However, WM3918 did show a significant drop in ECAR, which suggests that trametinib had some effect on glycolytic activity in the wild-type line. WM983BR showed small but significant increases in OCR and ECAR with trametinib treatment. Still, the OCR/ECAR ratio shift was not significant, probably due to high measurement deviations. Other studies on BRAF/MEK inhibition found that some BRAF mutants rapidly resisted BRAF and MEK inhibitors through genetic or epigenetic alterations [
41]. The in vitro resistance can also be seen in our results in the dabrafenib-resistant WM983BR. However, more work is needed to elucidate the discrepancies in the response of WM983BR to trametinib.
Our present results show that the cellular effects of a positive response to MEK inhibition were accompanied by a reduction of lactate production (
Figure 2) and a time-dependent reduction in tumor lactate levels (
Figure 4 and
Figure 5). These results corroborate the report by Falck Miniotis et al. [
11]. However, our methodology also allowed us to demonstrate changes in alanine, bioenergetics, and pH. As shown in
Figure 5, the lactate changes matched alanine’s in WM983B and DB-1. However, reduced tumor lactate levels were found on Day 5, but alanine was not significantly reduced on that day in the WM983BR cell line. These data match the fact that WM983BR has a decreased but not significant tumor volume with trametinib treatment during our observation period, showing only a statistical trend (
p = 0.06,
Figure 7). This suboptimal response matches with a significant tumor lactate reduction but not with one for alanine, suggesting that WM983BR has a unique metabolic phenotype responding differently to trametinib than the two fully responding cell lines. Therefore, finding ways to coax an alanine reduction in WM983BR (e.g., using different therapies or combining other drugs with trametinib) may help improve its response. In comparison, WM983B and DB-1, the responding cell lines, showed the expected decrease in tumor growth while on trametinib therapy (
Figure 7), correlating with significant lactate and alanine changes (
Figure 5).
An essential objective of this work was to explore the clinical translatability of our standardized noninvasive tumor metabolic assessment. Experimental measurements using
1H MRS have revealed that the reduction in lactate observed after BRAF/MEK inhibition was more pronounced in cells containing the oncogenic
BRAFV600E mutants than non–BRAF-driven cells [
11]. These metabolic differences suggest that BRAF-independent cells do not experience a robust metabolic response to treatment, as lactate levels remained unchanged. However, a reduction in alanine was observed following MEK inhibition in the cell lines containing the oncogenic
BRAFV600E mutation and in non–BRAF-driven cells (
Figure 5). Alanine, a product of pyruvate amination and an essential component of the glutaminolysis pathway, plays a crucial role in cellular metabolism. Our findings in
Figure 4 and
Figure 5 demonstrate decreased lactate and alanine concentration with MEK inhibition, as observed by in vitro and in vivo
1H MRS of preclinical melanoma models.
To demonstrate prediction by lactate and alanine, we adapted the clinically used RECIST (response evaluation criteria in solid tumors) [
42,
43] to determine response using the mean group value of tumor volume on Day 5 of the trametinib treatment xenografts depicted in
Figure 7. On Day 5, WM3918 had a 50% increase in tumor volume; thus, RECIST classifies it as a progressive disease (PD). In comparison, WM983BR had a tumor increase below 20%, and WM983B had a decrease of less than 30% in tumor volume, classifying both as stable diseases (SD). Finally, DB-1 shows a 40% decrease, so RECIST considers it a partial response (PR). Following this convention, the only xenograft with a positive response on Day 5 is DB-1 (PR). However, continued trametinib therapy could have brought some of these tumors to achieve a complete response (CR). The linear regressions for WM983B and DB-1 in
Figure 7 predict that CR (i.e., a complete absence of tumor) could have been achieved by Day 24 and Day 14, respectively. Notably, these two xenografts have significantly and sustainably reduced lactate and alanine since Day 2. In comparison, the time course of the tumor volume of WM983BR predicts that CR can never be reached. WM983BR had a tumor lactate value significantly reduced by Day 5, but the alanine value did not change (
Figure 5). These analyses suggest that early and sustained reductions of tumor lactate and alanine values predict a subsequent CR response in melanoma. They also suggest that a delayed reduction of lactate without changes in alanine (e.g., WM983BR) or no changes in both metabolites (e.g., WM3918) predict a negative response.
Using
31P MRS, we found that higher bioenergetic levels (i.e., β-NTP/Pi) correlate with an increased response to trametinib in responsive melanoma lines (
Figure 6). However, only the DB-1 sensitive line showed a significant increase in extracellular pH (pHe). Tumors with higher glycolytic or energetic profiles, such as WM983B and DB-1, show better responsiveness to trametinib than wild-type (WM3918) and BRAF-resistant (WM983BR) melanoma cell lines. The reason for this correlation between cellular energy state and glycolytic capacity remains unknown. However, this selective tumor de-energization could enhance the tumor’s response to therapeutic agents like trametinib.
Although we expected metabolic responses similar to trametinib compared to dabrafenib, as both affect the same regulatory pathway, we found subtle differences amongst the cell lines. For example, dabrafenib increased OCR significantly in WM983B, while the OCR change was insignificant with trametinib. Furthermore, when comparing control with trametinib therapy, WM983BR has reduced (or inverted) time-related slopes for lactate and alanine in the xenograft studies. In comparison, both slopes are positive and not significantly different when using dabrafenib. The same WM983BR has a much steeper slope for the b-NTP/Pi increase in response to trametinib than dabrafenib [
10].
Finally, the differences between the lactate levels before therapy could also be important as potential biomarkers of response in patients with melanoma. The in vivo value of lactate in the untreated (control) DB-1 xenografts (the best responder to trametinib) is significantly larger in comparison to the rest of the xenografts (
Figure 5) We hypothesize that finding a large tumor lactate value in melanoma patients before therapy could be predictive of a positive response with trametinib.
Author Contributions
Conceptualization, K.N., P.K.G., and D.S.N.; methodology, K.N. P.K.G., and D.S.N.; software, P.K.G., K.N., and D.S.N., ; validation, P.K.G., K.N., F.A.M., D.S.N., and S.O.; formal analysis, P.K.G., F.A.M., K.N., D.S.N., and S.O.; investigation, P.K.G., K.N., D.S.N., and S.O.; resources, K.N., D.S.N., P.K.G., and S.O.; data curation, P.K.G., F.A.M., K.N., D.S.N., and S.O.; writing—original draft preparation, P.K.G., K.N., D.S.N., and S.O.; writing—review and editing, P.K.G., F.A.M., K.N., D.S.N. and S.O.; visualization, P.K.G., F.A.M., K.N., D.S.N., and S.O.; supervision, K.N. P.K.G and D.S.N.; project administration, K.N., and D.S.N.; funding acquisition, K.N., and D.S.N. All authors have read and agreed to the published version of the manuscript.