HLA class I downregulation is associated with enhanced NK‐cell killing of melanoma cells with acquired drug resistance to BRAF inhibitors

The frequent development of drug resistance to targeted therapies in cancer patients has stimulated interest in strategies counteracting resistance. Combining immunotherapies with targeted therapies is one such strategy. In this context, we asked whether human NK cells can target melanoma cells that have acquired resistance to selective inhibitors targeting activating mutants of the B‐Raf kinase (BRAF inhibitors, BRAFi). We generated drug‐resistant cell variants in vitro from human BRAF‐mutant melanoma cell lines MEL‐HO, COLO‐38, SK‐MEL‐37, 1520 and from primary melanoma cells freshly isolated from two patients. All drug‐resistant cell variants remained susceptible to lysis by IL‐2‐activated NK cells; and two BRAFi‐resistant lines (BRAFi‐R) became significantly more susceptible to NK‐cell lysis than their parental lines. This was associated with significant HLA class I antigen downregulation and PD‐L1 upregulation on the drug‐resistant lines. Although blocking HLA class I enhanced the extent of lysis of both BRAFi‐R and parental cells to NK‐cell‐mediated lysis, antibody‐mediated inhibition of PD1–PD‐L1 interactions had no detectable effect. HLA class I antigen expression on BRAFi‐R melanoma variants thus appears to play a major role in their susceptibility to NK‐cell cytotoxicity. These findings suggest that NK‐cell‐based immunotherapy may be a viable approach to treat melanoma patients with acquired resistance to BRAF inhibitors.


Peer review correspondence
of increased susceptibility to NK cell killing (or resistance for the 1520 line) or link between BRAFi resistance changes in NK cell ligands. Reviewer: 2 Comments to the Author The current study attempts to demonstrate the susceptibility of drug-resistant tumor lines to NK cell killing in comparison with untreated parental lines. They demonstrate one line (Mel Ho) has increased killing opf the resistant line whereas another line 1520 actually has lesser killing. Furthermore, other lines assessed had equivocal effects. They attempt to discern the mechanism underlying these effects with surface expression of various targets/ligands with subtle changes being observed but no real clear indication as to why differential killing effects are occurring. While potentially of interest, the variability (even opposite effects) observed with the cell lines make definitive conclusions difficult to extrapolate as may be cell line contingent. The effects in Figure 4 are also difficult to correlate as definitive blocking studies are needed to show they do play a role. It is unclear as to why higher killing in the resistant Mel Ho line is associated with lower MIC-A and why this was not assessed in 1520 or other lines. Figure 3 is key yet hard to discern based on significance and variability.

Reviewer: 3
Comments to the Author Summary: The article by Tan et al. evaluates the ability of IL-2 activated NK cells to respond to BRAF-inhibitor resistant melanoma cell lines. The potential for these studies has a high level of clinical relevance since resistance to BRAF inhibition in patients is a common long-term problem. The authors nicely demonstrated the generation of multiple resistant cell lines, and validated the V600E mutation that existed in one of the parent lines was maintained in the resistant cell line. The purpose of the report was to determine if the BRAF-inhibitor resistant cell lines would be sensitive to IL-2 activated NK cell killing. While the data presented has important clinical ramifications, additional studies are needed to confirm the ability of IL-2 activated NK cells to work in the pre-clinical setting.
Specific comments: 1. The methods section indicates that both the BRAF resistant and sensitive cell lines were cultured for two weeks prior to functional studies in inhibitor-free media. Did the tumor cells maintain their resistance after two weeks without exposure to the inhibitor? If all of the data presented are from after the two-week wash out period, it would be interesting to determine if the effect on NK cell sensitivity would be stronger Peer review correspondence immediately following BRAF-inhibitor treatment. As seen with other agents, continued treatment with the inhibitor while ineffective at controlling tumor growth directly, may continue to sensitize the tumor cells to NK cell mediated killing.
2. Studies evaluating NK ligand expression on the parent and resistant cell lines indicated a potential mechanism for the differences in susceptibility to NK killing. These studies should be supported by additional in vitro confirmatory studies with blocking antibodies to determine the ligand interaction(s) that are responsible.
3. All of these studies were conducted with tumor cell lines, and while this suggests NK cells would still recognize and kill tumor targets, there is no primary data to support this. The manuscript can be strengthened by the addition of primary samples (BRAF sensitive and resistant) to validate they are sensitive to IL-2 activated NK killing.
4. There is no flow cytometry data in the main paper demonstrating the gating strategy for the functional studies. This is shown in Supplementary Figure 3. This figure should be included in the manuscript, along with the bar graph already depicted of the results.
5. The use of NKp46 as the single defining NK-specific marker is not standard. Including a flow cytometry stain of the IL-2 activated NK cells that includes CD56, CD3, CD94, and CD16 at a minimum would be helpful to evaluate the phenotype of the NK cells post culture, at the time of the functional assays. The new experimental evidence confirms the inverse correlation between MHC class I expression at the cell surface and susceptibility to NK-cell mediated killing.

Reviewer: 2
Comments to the Author The current study attempts to demonstrate the susceptibility of drug-resistant tumor lines to NK cell killing in comparison with untreated parental lines. They demonstrate one line (Mel Ho) has increased killing opf the resistant line whereas another line 1520 actually has lesser killing. Furthermore, other lines assessed had equivocal effects. They attempt to discern the mechanism underlying these effects with surface expression of various targets/ligands with subtle changes being observed but no real clear indication as to why differential killing effects are occurring. While potentially of interest, the variability (even opposite effects) observed with the cell lines make definitive conclusions difficult to extrapolate as may be cell line contingent. The effects in Figure 4 are also difficult to correlate as definitive blocking studies are needed to show they do play a role. It is unclear as to why higher killing in the resistant Mel Ho line is associated with lower MIC-A and why this was not assessed in 1520 or other lines. Figure 3 is key yet hard to discern based on significance and variability.
As suggested by this reviewer we performed blocking studies to determine the molecular mechanism regulating the NK cytotoxic recognition of BRAFi resistant melanoma variants. As reported in the new figure 5 (refers to paragraph "Immunomodulatory effects of BRAFi and MHC class I molecules on NK cell cytotoxicity", page 6 lines 22-26 and page 7 lines 1-2) we blocked the interaction between CD155 and CD112 with their receptor (DNAM-1) and the MIC A/B, ULBP 1-2-3-4 receptor (NKG2D) using monoclonal antibodies. Our data clearly showed that these ligands are not involved in the recognition of 1520R.
Instead by masking MHC class I molecules we had the formal proof that MHC class I is the key regulatory mechanism in the regulation of the NK cell recognition of BRAFi resistant melanoma variants. Our new Peer review correspondence data clearly indicate that the reduced expression of MHC class I on MEL-HO cells is the mechanism by which these cells have higher sensitivity to NK cell mediated killing after acquisition of drug resistance.

Reviewer: 3
Comments to the Author 2. Studies evaluating NK ligand expression on the parent and resistant cell lines indicated a potential mechanism for the differences in susceptibility to NK killing. These studies should be supported by additional in vitro confirmatory studies with blocking antibodies to determine the ligand interaction(s) that are responsible.
3. All of these studies were conducted with tumor cell lines, and while this suggests NK cells would still recognize and kill tumor targets, there is no primary data to support this.
The manuscript can be strengthened by the addition of primary samples (BRAF sensitive and resistant) to validate they are sensitive to IL-2 activated NK killing.
4. There is no flow cytometry data in the main paper demonstrating the gating strategy for the functional studies. This is shown in Supplementary Figure 3. This figure should be included in the manuscript, along with the bar graph already depicted of the results.
Peer review correspondence 5. The use of NKp46 as the single defining NK-specific marker is not standard. Including a flow cytometry stain of the IL-2 activated NK cells that includes CD56, CD3, CD94, and CD16 at a minimum would be helpful to evaluate the phenotype of the NK cells post culture, at the time of the functional assays.
1. The primary cells used were maintained in the presence of the inhibitors without wash out period after generation of the resistant variants. For the specific methods refer to paragraph "Generation of BRAFiresistant cells" page 14, lines 1-4. As suggested by this reviewer the effect on NK cell sensitivity immediately following BRAF-inhibitor treatment has been analysed. The new data are included in the paragraph "Susceptibility of primary melanoma cells to NK cell cytotoxicity upon short-term BRAFi treatment and upon established BRAFi resistance" page 7-8, figure 6,7,8,9 and Supplementary 4. We

Peer review correspondence
Executive Editor was sought. All opinions have now been assessed and the comments of the referees are included at the bottom of this letter.
You will see that Referee 2 was not satisfied with the revisions made; this referee felt that the subtle differences in some of your figures and the lack of a clear mechanism from this phenomenon warranted rejection. However the Executive Editor feels that a revision is feasible, and asks that you experimentally address the concerns of Referees 1 and 3 and discuss the weakness/limitations of your work in response to Referee 2 in your next revision. Please note that the journal does not encourage multiple rounds of revision and you should fully address the concerns of the referee in this final round of revision. Should you disagree with any of the referees' concerns, you should address this in your point-by-point response and provide solid scientific reasons for why you will not make the requested changes.
You should also pay close attention to the editorial comments included below. *In particular, please edit your figure legends to follow Journal standards as outlined in the editorial comments. Failure to do this will result in delays in the re-review process.* Please note that submitting a revision of your manuscript does not guarantee eventual acceptance, and that your revision will be re-reviewed by the referees before a decision is rendered.
If the revision of the paper is expected to take more than three months, please inform the editorial office.
Revisions taking longer than six months may be assessed by new referee(s) to ensure the relevance and timeliness of the data.
Once again, thank you for submitting your manuscript to European Journal of Immunology and we look forward to receiving your revision. Specific comments: 1. The authors show that inhibition of Class I on 1520-R cells increases susceptibility to NK cell killing.
However, it is not clear that this is the only mechanism of resistance to NK cell killing since the parental line also becomes more NK-sensitive (as would be expected with blocking class I). Is there a significant difference between % lysis of parental and 1520-R cells with anti-HLA-I ( Figure 4B), as there appears to be? If sothe authors should discuss that this might not be the sole mechanism of NK resistance in this cell line.
2. On page 32, the authors state "Similar results were obtained using dabrafenib (Supplement 2)", but are discussing increased MHC-I expression. Supplemental Figure 2 shows killing, but no data for Class I expressiondo the authors have this data and if so it should be shown. If this data is not available the discussion should be changed accordingly.
3. The statistics for MHC-I expression in Mel30-R and Mel35-R cells show a "significant" difference, but do the authors think this is a biologically significant difference? The results are based on MFI which are shown as % of control, but the histograms in Figure 6 show very little difference (especially when compared with changes in Class I in other figures, such as Figure 5B). There are large differences in NK cell killing of Mel30 versus Mel30-R cells, and based on the flow shown it is hard to believe that this is due to Class I alone. This is worth a discussion and a more fair/balanced discussion of the data.

Reviewer: 2
Comments to the Author This is a revised study attempting to address issues related to drug resistance in melanoma as it relates to NK cell killing capability. A primary issue is the marginal differences observed both in MFI changes and killing extent in the lines, inconsistency with regard to overall outcome and lack of clear mechanism due to variable and partial effects especially in figure 5 and 6.

Peer review correspondence
Comments to the Author The revisions made thus far to the manuscript by Tan et al. have improved both the quality and the relevance of the data with regards to the effectiveness of NK cell-based therapy against BRAF resistant melanoma cells. However, the authors failed to evaluate a critical interaction that was found to be altered in the sensitive and resistant cell lines. Although the importance of PD-L1 expression was discussed in the paper, the authors failed to attempt blocking this interaction in the resistant lines to determine if this ligand also played a role in sensitizing BRAF resistant cells to NK killing, although it appears to be the most significantly dysregulated ligand. The potential impact of the manuscript on designing effective immunotherapies against resistant melanoma samples would be enhanced if these studies were added. In addition, the primary samples should be evaluated for PD-L1 expression as well given the results with the melanoma cell lines.

Second revision -authors' response -22 September 2015
We agree with the comments and, accordingly, we now present what we feel is a more balanced view. Specifically, we have: 1) Changed the title by shifting the emphasis on the main discovery of the study, which is the susceptibility of both BRAFi-responsive and -resistant melanoma cells to NK cell cytotoxicity.
2) Discussed the role of MHC class I in the context of other potential interpretations.
Moreover, as suggested by Reviewer 1 and 3 and by the Editor, we have extended our analysis of the mechanisms to the PD1/PD1L pathway. The new data included in the revised manuscript clearly rule out this pathway in regulating the NKmelanoma cells cross talk. We now state this in the text of the manuscript.

Third Editorial Decision -7 October 2015
Dear Dr. Carbone, It is a pleasure to provisionally accept your manuscript entitled "NK cell targeting of melanoma cells with acquired drug resistance to BRAF inhibitors" for publication in the European Journal of Immunology. For final acceptance, please follow the instructions below and return the requested items as soon as possible as we cannot process your manuscript further until all items listed below are dealt with.
Please note that EJI articles are now published online a few days after final acceptance (see Accepted Articles: http://onlinelibrary.wiley.com/journal/10.1002/(ISSN)1521-4141/accepted). The files used for the