Intra- and inter-tumor heterogeneity in a vemurafenib-resistant melanoma patient and derived xenografts

Abstract

The development of targeted inhibitors, like vemurafenib, has greatly improved the clinical outcome of BRAF(V600E) metastatic melanoma. However, resistance to such compounds represents a formidable problem. Using whole-exome sequencing and functional analyses, we have investigated the nature and pleiotropy of vemurafenib resistance in a melanoma patient carrying multiple drug-resistant metastases. Resistance was caused by a plethora of mechanisms, all of which reactivated the MAPK pathway. In addition to three independent amplifications and an aberrant form of BRAF(V600E), we identified a new activating insertion in MEK1. This MEK1(T55delins) (RT) mutation could be traced back to a fraction of the pre-treatment lesion and not only provided protection against vemurafenib but also promoted local invasion of transplanted melanomas. Analysis of patient-derived xenografts (PDX) from therapy-refractory metastases revealed that multiple resistance mechanisms were present within one metastasis. This heterogeneity, both inter- and intra-tumorally, caused an incomplete capture in the PDX of the resistance mechanisms observed in the patient. In conclusion, vemurafenib resistance in a single patient can be established through distinct events, which may be preexisting. Furthermore, our results indicate that PDX may not harbor the full genetic heterogeneity seen in the patient's melanoma.

Keywords: Melanoma; drug resistance; patient‐derived xenografts; tumor heterogeneity.

Figure 1. Clinical response of patient M032 to vemurafenib

1. Treatment schedule of patient M032, who received a PET scan two weeks before the start of the treatment (purple asterisk), followed by a baseline CT scan and every two months a follow-up CT scan (red asterisks). Metastases were surgically removed either shortly before the start of the treatment or after resistance occurred (post, M032R1-R5). PR, partial response; PD, progressive disease. Locations of the removed metastases are indicated in the illustration.

2. Examples of CT scan images of several metastases (M032R1, M032R2, and M032R5).

3. Volumetric measurements based on the PET and CT scans could be generated from M032, M032R1, and M032R5. Graph illustrates that the metastases R1 and R5 initially expanded before the start of the treatment, but regressed upon vemurafenib treatment. M032R5 showed progressive disease after 4 months, whereas M032R1 still displayed stable disease. M032 was excised and no recurrence was observed.

Figure 2. Reactivation of MAPK pathway via amplification of a BRAF-encoding DNA fragment caused resistance to vemurafenib in three metastases

1. Hematoxylin–eosin (H&E) and p-ERK stainings on FFPE material of all metastases showed that all vemurafenib-resistant tumors had reactivation of the MAPK pathway. Scale bar represents 100 μm.

2. CNA profiles were generated from the WES data with germ line DNA as a reference. Colors represent segmented log2 ratio values with red for gain and blue for loss. Further inspection of chromosome 7 where BRAF is located revealed amplification (7q34) of this region in three vemurafenib-resistant metastases, namely M032R1, M032R2, and M032R5.

3. qPCR was performed on gDNA retrieved from each of the metastases, using primers for BRAF and CRAF and normalized on LINE levels. Bars represent the mean of three replicates, error bars indicate standard deviation. The results confirmed that BRAF was amplified in M032R1, M032R2, and M032R5.

4. Staining for BRAF^V600E with a mutant epitope-specific antibody confirmed the upregulation of BRAF^V600E in R1, R2, and R5. Scale bar represents 100 μm.

Figure 3. Alternative form of BRAF^V600E is correlated with p-ERK and p-MEK reactivation in M032R3

1. Immunoblotting for MAPK pathway components confirmed reactivation of p-ERK and p-MEK in all resistant metastases.

2. Immunoblotting for M032 patient’s samples using an antibody directed against either the mutant BRAF^V600E epitope or an N-terminal region of BRAF revealed the expression of a shorter variant of BRAF^V600E in M032R3, of approximately 80 kDa. This 80-kDa band could not be detected with the N-terminus-recognizing antibody.

Source data are available online for this figure.

Figure 4. T55delinsRT mutation in MEK1 is responsible for resistance to vemurafenib in M032R4

1. Somatic mutations present in each of these metastases, revealing that mutations were either shared by the pre-treatment tumor and all metastases, by some metastases, or only by single metastases.

2. A 3-bp in-frame insertion in MEK1 (T55delinsRT) was identified in M032R4. This mutation was located at the base of the α-helix at the N-terminal site of MEK1.

3. Validation of T55delinsRT was done by PCR using insertion-specific primers. As a control, primers amplifying exon 2 of MEK1^WT were used.

4. Transfection of pQXCIP-GFP, MEK1-WT, or MEK1-T55delinsRT into HEK293T cells showed that this mutation induces p-ERK and p-RSK.

5. Validation of expression of pQXCIP-GFP, MEK1^WT, or MEK1^T55delinsRT in A375 melanoma cells by immunoblotting.

6. Dose–response curves for A375 melanoma cells expressing GFP, MEK1^WT, or MEK1^T55delinsRT with indicated doses of BRAF inhibitor vemurafenib, ERK inhibitor SCH772984, or MEK inhibitor trametinib. Error bars indicate standard deviation.

7. Colony formation assays with A375 melanoma cells, infected with GFP, MEK1^WT, or MEK1^T55delinsRT-encoding lentivirus, and treated with indicated doses and inhibitors.

8. Treatment of A375 MEK1^T55delinsRT melanoma cells with DMSO (−), 250 nM dabrafenib (D), 10 nM trametinib (T), or a combination (D+T).

Source data are available online for this figure.

Figure 5. T55delinsRT mutation in MEK1 confers resistance to BRAF inhibition and promotes local invasion in vivo

1. A375 melanoma cells expressing GFP, MEK1^WT, or MEK1^T55delinsRT were injected into immune-deficient mice (n = 8 per group), and after the tumor size of ˜100 mm³ was reached, mice were treated with 30 mg/kg dabrafenib or vehicle. Graphs represent fold change in tumor volume normalized on the tumor volume on the day of the start of the treatment. Error bars indicate standard error of the mean.

2. Average tumor weight of all groups at the end of the experiment described in (A). Error bars indicate standard deviation. Unpaired t-test was used to calculate statistical significance. P-values are indicated in the graph.

3. Immunoblotting for MAPK pathway components performed on tumors of all groups.

4. Immunohistochemistry for Sirius Red (staining for collagen) and H&E on tumors from all groups, to identify grade of demarcation and local invasion. Scale bar represents 500 (upper row) and 100 (lower row) μm.

Source data are available online for this figure.

Figure 6. MEK1^T55delinsRT is a preexisting mutation

gDNA was isolated from another fragment of the tumor resection, and the PCR with the insertion-specific primers was repeated. Primers amplifying exon 2 of MEK1^WT were used as an input control. Source data are available online for this figure.

Figure 7. Patient-derived xenografts display similar resistance mechanisms as the parental metastases

1. Staining for hematoxylin–eosin (H&E) and p-ERK on the PDX samples M032.X1, M032R1.X1, M032R2.X1, M032R4.X1, and M032R5.X1. Stainings showed that p-ERK is higher in the PDX derived from the resistant metastases. Scale bar represents 100 μm.

2. Immunoblotting for components of the MAPK pathway confirmed reactivation of p-ERK in the PDX derived from the vemurafenib-resistant metastases, although the p-ERK signal is heterogeneous in the pre-treatment PDX.

3. Validation of the BRAF amplification was performed by qPCR on gDNA derived from the PDX, using primers for BRAF and CRAF and normalized on LINE expression. Bars represent the mean of three replicates, error bars indicate standard deviation. The results confirmed that BRAF was amplified in PDX derived from M032R2 and M032R5, but not from M032R1.

4. Presence of the MEK1^T55delinsRT was analyzed by PCR using insertion-specific primers. As a control, primers amplifying exon 2 of MEK1^WT were used. The insertion was found in M032R1.X1 and M032R4.X1.

Source data are available online for this figure.

Figure 8. Overview of all identified vemurafenib resistance mechanisms in patient and PDX samples

Resistance mechanisms discovered in each of the lesions have been identified: a) in patients samples by WES (BRAF^V600E amplification in M032R1, R2 and R5; MEK1^T55delinsRT in M032R4 and M032) and immunoblotting (of alternative form of BRAF in M032R3); and b) in PDX by IHC and qPCR (BRAF^V600E amplification in M032R2 and R5) and PCR analysis (MEK1^T55delinsRT in M032R1 and R4). Gray arrows indicate that the lesion was at the back side. ampl., amplification; alt., alternative.