Whole genome landscapes of uveal melanoma show an ultraviolet radiation signature in iris tumours

Uveal melanoma (UM) is the most common intraocular tumour in adults and despite surgical or radiation treatment of primary tumours, ~50% of patients progress to metastatic disease. Therapeutic options for metastatic UM are limited, with clinical trials having little impact. Here we perform whole-genome sequencing (WGS) of 103 UM from all sites of the uveal tract (choroid, ciliary body, iris). While most UM have low tumour mutation burden (TMB), two subsets with high TMB are seen; one driven by germline MBD4 mutation, and another by ultraviolet radiation (UVR) exposure, which is restricted to iris UM. All but one tumour have a known UM driver gene mutation (GNAQ, GNA11, BAP1, PLCB4, CYSLTR2, SF3B1, EIF1AX). We identify three other significantly mutated genes (TP53, RPL5 and CENPE).

-CENPE: the second wild-type allele is retained with all variants L130 "It is possible that disruption of this pathway is responsible for creating genomic instability allowing for chromosome aberrations to occur." The authors should have all data to answer this question. Any correlation with WGD ? with number of CNA ? The authors should take into account that Oncodrive-fm does not correct for the size of the gene, and that CENPE is a very large gene (so it is highly likely that these variants are passenger. -TP53: "Mutations in TP53 are common in many cancers but have not been described in UM to date." False: see Hajkova et al Sci Rep 2018 andref, Rodrigues et al CCR 2019. One of these cases is an hypermutated metastasis, which should be mentioned. Actually, the bottom line is that two cases are inactivated for TP53, one hypermutated metastasis (as in Rodrigues et al CCR 2019) and one primary of unknown origin.
-RPL5 : None of these mutations are associated with LOH. One case is hypermutated. L151 "RPL5 along with RPL11 and ARF are responsible for inhibiting MDM2 and stabilising p53 in response to blocks in ribosome biogenesis and nucleolar stress, often triggered by oncogene-induced translational stress. Supporting the importance of this pathway in UM, TP53 and RPL5 were altered in 42% of cases in this cohort." This is an overstatement supported by no evidence, but that the longest chromosome (chr1) is frequently altered in UM. 5) L167 "Interestingly, while iris melanomas are associated with earlier detection and favourable prognosis, 4 of 8 iris cases had M3, with two already having progressed to metastatic disease". None of these metastatic patients received immunotherapy ?
Minor comment -The first publication of of UM characterized by WGS (Furney Cancer Discov 2013) should be cited -L97 "These mutations were generally mutually exclusive except for two PLCB4 p.D630 mutations. As seen in the UM TCGA cohort, overlap occurred with the minor GNAQ p.R183Q/ GNA11 p.R183C. hotspots rather than the stronger oncogenic p.Q209 hotspot mutations." Confusing: something like "two PLCB4 p.D630 co-occurred with GNA p.R183" would be more straightforward.
-L145: "These data suggest the two mutations are biallelic". No, the authors suggest composite heterozygous mutations or mutations in trans.
Reviewer #3 (Remarks to the Author): Johansson et al describe the whole genome landscapes of 103 uveal melanoma patients, including 91 primary tumors and 12 metastases. The overall conclusions of the study are: (1) TMB is low in the vast majority of uveal melanoma patients, consistent with prior studies; but two patients with loss of function mutations in MBD4 with higher TMB. Similarly all iris-derived melanomas had high UVR damage signatures, suggesting that this subset might have unique biological properties. The finding that iris UM have higher TMB has not been reported, though the MBD4 mutations have been previously described (as cited by the authors) (2) A mutation in GNAQ pG48L appears to be a minor oncogenic hotspot mutation (3) Multiple aberrations in TP53 pathway. However, functional data for most of these mutations could be strengthen by evaluating recent, comprehensive analysis of p53 mutations (e.g. Giacornelli et al, Kotler et al, etc).
(4) RPL5 is mutated in 3 cases, which the authors suggest is related to p53, however I could not find any analysis indicating that these were mutually exclusive of TP53 mutations (or if there is any relationship at all) Overall, this is a large amount of work, but most of the conclusions are not novel and confirm prior studies. Unfortunately no functional data is provided related to the mutations observed. Finally, the discussion is somewhat superficial, and does not even mention some other whole genome sequencing manuscripts (e.g. Rover-Bertrand et al.) Given the size of this cohort, it is suggested that greater insights may be derived by combining the data in this cohort with other datasets, or at least comparing the data.
We are grateful to the reviewers for their careful consideration of this manuscript and their helpful suggestions for improving the paper. We have addressed the reviewers' comments in full below. Reviewers' questions and comments are in bold black text and the authors' responses are in blue text.

Reviewer #1
Red in Figure 2 should be used for the more conventional copy number gain rather than loss to avoid reader confusion. Figures 2 and 3 have been changed as suggested.

Reviewer #2
1) Title is not supported by evidence. If UVR signature is shown in iris UM, no causal link with the disease is provided. UVR signature also exists also in normal melanocytes or keratinocytes!
We accept that we do not show a causal link between UVR exposure and iris UM tumorigenesis, but we show for the first time molecular evidence this subset of UM have a UVR signature. We have modified the title accordingly.
2) L70 "One sample had noticeably more SVs, of which the majority (71%) were midsized (<100 kb) deletions, suggesting this was not due to chromothripsis" Limited information is provided for this interesting outlier case. Any correlation of driver mutations?
Analyses were done in relation to SVs and SNVs in the tumour and there were no notable features or driver mutations. We have now indicated this in the text.
3) L195 "In addition, two D3 tumours carried BAP1 mutations, indicating that although BAP1 inactivation typically occurs after M3, BAP1 aberration can also occur in D3 tumours, which may or may not later undergo loss of chromosome 3." These observations are not as surprising as they appear in the text. One is associated with a partial loss of chr3, which is relatively rare but previously reported, The tumour in question does not have a partial loss of chr3, it is disomy 3. and for the second case, BAP1 mutation seems to have emerged as a subclonal event (low VAF), in which the status of chr3 is yet to be determined.
It is likely correct that the nonsense mutation in this tumour is a subclonal event and we have thus added the following clarification in the main text: "Of note, one of the mutations had a low variant allele frequency (VAF = 9/80) suggesting it was only present in a subclone and as copy number tools are not as sensitive as mutation callers, it is possible that the subclone had loss of heterozygosity (LOH) that was not detected by the algorithm."

4) Evidence for new drivers in UM are scarce: -"three additional significantly mutated genes" but only one algorithm (Oncodrive-fm) shows this significance, and with a marginal p-value for CENPE.
It is not unexpected that SMGs are not always identified by all tools, since they focus on different criteria (which was our reason to run multiple tools), e.g. OncodriveCLUST is designed to find hotspot mutations, thus many tumour suppressors are missed; MutSigCV is more focused on the number of mutations, so genes with relatively few mutations are often missed; whereas Oncodrive-FM is more focused on the functional impact of mutations. These points are exemplified in our UM data by the observation that BAP1, SF3B1, and GNA11 were each only called by only 2 of the 3 tools used.

-CENPE: the second wild-type allele is retained with all variants
We acknowledge that the CENPE disruption is heterozygous in these tumours. However there is evidence to support that CENPE disruption might function in a haploinsufficient manner. The discussion has been expanded to address and clarify this issue as follows: "Studies both in cell line and mouse models have demonstrated that CENP-E/Cenp-E functions in a haploinsufficient manner, with elevated levels of chromosome mis-segregation observed in heterozygous cells and animals (PMID: 17189716, PMID: 17974949 ). Further functional work on CENPE missense mutations is required to determine their impact in UM."

L130 "It is possible that disruption of this pathway is responsible for creating genomic instability allowing for chromosome aberrations to occur." The authors should have all data to answer this question. Any correlation with WGD ? with number of CNA ?
While UM with altered CENPE do have more CNA, this cannot be taken to indicate causation, as samples with many CNAs tend to have more LOH in virtually any genejust like samples with high TMB tend to have more somatic mutations in any gene. We have added this correlation into the main text and also included the described caveat.
The authors should take into account that Oncodrive-fm does not correct for the size of the gene, and that CENPE is a very large gene (so it is highly likely that these variants are passenger).
Oncodrive-FM does correct for gene size. The algorithm computes the average FI score for all observed mutations in the gene, where the FI scores measure the functional impact of the mutations (e.g. from CADD). It next calculates the average score for the observed mutations and compares that with the distribution of averages when drawing mutations randomly. If the observed average score is more often higher than the average from randomisations, the gene is considered significantly mutated. Hajkova et al. 2018 presented two cases with germline TP53 variants (of unknown pathogenicity) proposed to be associated with Li-Fraumeni syndrome. What we report here is the common acquisition of somatic TP53 mutations, the majority having confirmed pathogenicity.

and ref, Rodrigues et al CCR 2019.
We have referenced the one somatic TP53 mutation described in Rodrigues et al. as well as a second case in the MSK-IMPACT data set; these observations were, however, single instances in cohorts of 42 and 43, respectively. Here we report for the first time that TP53 is a statistically significantly mutated gene in UM. We have changed the text to: "Somatic mutations in TP53 have been described in two UM. A hotspot mutation p.R175H (n = 1216 in IARC TP53 database) was observed in a hypermutated metastatic UM with deficient MBD4 and another hotspot mutation p.M237I (n = 196 in IARC TP53 database) was observed in a UM in a pan-cancer study of metastatic tumours. Here we identified TP53 as an SMG and report six somatic TP53 mutations across four tumours in addition to eight cases of LOH" -RPL5 : None of these mutations are associated with LOH. One case is hypermutated.
We acknowledge that none of the UM has a double hit of RPL5, which is expected as RPL5 has been described as a haploinsufficient tumour suppressor (Fancello et al. Oncotarget, 2017). We have extended the discussion on this point to make it clearer for the reader.
L151 "RPL5 along with RPL11 and ARF are responsible for inhibiting MDM2 and stabilising p53 in response to blocks in ribosome biogenesis and nucleolar stress, often triggered by oncogene-induced translational stress. Supporting the importance of this pathway in UM, TP53 and RPL5 were altered in 42% of cases in this cohort." This is an overstatement supported by no evidence, but that the longest chromosome (chr1) is frequently altered in UM.
The discussion on RPL5 has been rewritten and the text above has been deleted. 5) L167 "Interestingly, while iris melanomas are associated with earlier detection and favourable prognosis, 4 of 8 iris cases had M3, with two already having progressed to metastatic disease". None of these metastatic patients received immunotherapy?
Unfortunately, neither of the two patients in the iris UM cohort who have developed metastases received immunotherapythus we cannot correlate response to treatment. We hope our report may lead to the analysis of previous and future metastatic iris UM patients who receive immunotherapy to determine if a correlation with improved response rates exists.

Minor comment -The first publication of UM characterized by WGS (Furney Cancer Discov 2013) should be cited
We have added the Furney et al. reference.
-L97 "These mutations were generally mutually exclusive except for two PLCB4 p.D630 mutations. As seen in the UM TCGA cohort, overlap occurred with the minor GNAQ p.R183Q/ GNA11 p.R183C. hotspots rather than the stronger oncogenic p.Q209 hotspot mutations." Confusing: something like "two PLCB4 p.D630 co-occurred with GNA p.R183" would be more straightforward.
Thanks for the suggestion. We have changed the main text to: "These mutations were generally mutually exclusive except for two PLCB4 p.D630 mutations that co-occurred with GNAQ/GNA11 p.R183H mutations. This co-occurrence between PLCB4 mutation and the minor hotspot p.R183, rather than the stronger oncogenic p.Q209 hotspot mutations, has previously been described in the UM TCGA data." -L145: "These data suggest the two mutations are biallelic". No, the authors suggest composite heterozygous mutations or mutations in trans.
We have changed the sentence to: "These data suggest the two mutations occurred on different alleles, with the majority of the transcripts from the p.R342* allele undergoing nonsense mediated decay."

Reviewer #3
Multiple aberrations in TP53 pathway……... functional data for most of these mutations could be strengthen by evaluating recent, comprehensive analysis of p53 mutations (e.g.

Giacornelli et al, Kotler et al, etc).
We thank the reviewer for this suggestion and have added the results from these characterisation studies into the discussion of our observed TP53 mutations. For further clarity, we have also added a summary of these data together with data from the IARC TP53 database and the scores from the FATHMM prediction tool.

(4) RPL5 is mutated in 3 cases, which the authors suggest is related to p53, however I could not find any analysis indicating that these were mutually exclusive of TP53 mutations (or if there is any relationship at all).
We have added the following sentence about RPL5 and TP53 mutations: "Given the link between RPL5 and p53, we tested for an association between aberration in RPL5 and TP53. Indeed, mutations in these genes were mutually exclusive, but when also considering copy-number loss there was no association".

The discussion is somewhat superficial, and does not even mention some other whole genome sequencing manuscripts (e.g. Rover-Bertrand et al.) Given the size of this cohort, it is suggested that greater insights may be derived by combining the data in this cohort with other datasets, or at least comparing the data.
We have extended the discussion substantially, particularly discussion the new SMGs. We have also extended the description of UM categories, how they correlate with the common UM mutations, and have compared our data with previous large cohorts such as TCGA and Royer-Bertrand et al.
Johansson and Coll provide an analysis by WGS of the largest series of uveal melanoma (103 UM cases) described so far. In this revised manuscript, they propose some improvements and some modifications of their first manuscript, without providing any new key information. This analysis largely confirms previously published results, and further shows the presence of UVR mutational signature in iris UM. They demonstrate the rare but significant role of TP53 in this disease, which was previously reported. Evidence for RLP5 and CENPE as new drivers are minimal and unconvincing. It is difficult to understand the interest of using WGS rather than WES, (UVR role in iris melanoma is quite obvious using WES) when the authors are not exploiting their WGS data to extract additional information such as structural variants or alternations of none coding regions, such as promoter regions. 1) Title is not supported by evidence. If UVR signature is shown in iris UM, no causal link with the disease is provided. UVR signature also exists also in normal melanocytes or keratinocytes! We accept that we do not show a causal link between UVR exposure and iris UM tumorigenesis, but we show for the first time molecular evidence this subset of UM have a UVR signature. We have modified the title accordingly. → Thank you 2 L70 "One sample had noticeably more SVs, of which the majority (71%) were midsized (<100 kb) deletions, suggesting this was not due to chromothripsis" Limited information is provided for this interesting outlier case. Any correlation of driver mutations? Analyses were done in relation to SVs and SNVs in the tumour and there were no notable features or driver mutations. We have now indicated this in the text. → Thank you 3) L95 "In addition, two D3 tumours carried BAP1 mutations, indicating that although BAP1 inactivation typically occurs after M3, BAP1 aberration can also occur in D3 tumours, which may or may not later undergo loss of chromosome 3." These observations are not as surprising as they appear in the text. One is associated with a partial loss of chr3, which is relatively rare but previously reported, The tumour in question does not have a partial loss of chr3, it is disomy 3. → Indeed you classified MELA_0800 as disomic. However, this tumor has quite a low tumor content as estimated using GNAQ VAF (18%), so the chromosome status is questionable without further analyses. BAP1 VAF (11%) is thus compatible with a partial loss of chr3. Actually, genomic profiles of all tumors (such as FACETS) should be provided to clarify such outlier cases. and for the second case, BAP1 mutation seems to have emerged as a subclonal event (low VAF), in which the status of chr3 is yet to be determined.
It is likely correct that the nonsense mutation in this tumour is a subclonal event and we have thus added the following clarification in the main text: "Of note, one of the mutations had a low variant allele frequency (VAF = 9/80) suggesting it was only present in a subclone and as copy number tools are not as sensitive as mutation callers, it is possible that the subclone had loss of heterozygosity (LOH) that was not detected by the algorithm." → I do not share this analysis. The variant is moderate and no LOH is present in this tumor (BAP1 VAF 40% GNAQ 37%). So it is most likely a passenger mutation. 4) Evidence for new drivers in UM are scarce: -"three additional significantly mutated genes" but only one algorithm (Oncodrive-fm) shows this significance, and with a marginal p-value for CENPE. It is not unexpected that SMGs are not always identified by all tools, since they focus on different criteria (which was our reason to run multiple tools), e.g. OncodriveCLUST is designed to find hotspot mutations, thus many tumour suppressors are missed; MutSigCV is more focused on the number of mutations, so genes with relatively few mutations are often missed; whereas Oncodrive-FM is more focused on the functional impact of mutations. These points are exemplified in our UM data by the observation that BAP1, SF3B1, and GNA11 were each only called by only 2 of the 3 tools used.
-CENPE: the second wild-type allele is retained with all variants We acknowledge that the CENPE disruption is heterozygous in these tumours. However there is evidence to support that CENPE disruption might function in a haploinsufficient manner. The discussion has been expanded to address and clarify this issue as follows: "Studies both in cell line and mouse models have demonstrated that CENP-E/Cenp-E functions in a haploinsufficient manner, with elevated levels of chromosome mis-segregation observed in heterozygous cells and animals (PMID: 17189716, PMID: 17974949 ). Further functional work on CENPE missense mutations is required to determine their impact in UM." L130 "It is possible that disruption of this pathway is responsible for creating genomic instability allowing for chromosome aberrations to occur." The authors should have all data to answer this question. Any correlation with WGD ? with number of CNA ? While UM with altered CENPE do have more CNA, this cannot be taken to indicate causation, as samples with many CNAs tend to have more LOH in virtually any gene -just like samples with high TMB tend to have more somatic mutations in any gene. We have added this correlation into the main text and also included the described caveat.
The authors should take into account that Oncodrive-fm does not correct for the size of the gene, and that CENPE is a very large gene (so it is highly likely that these variants are passenger). Oncodrive-FM does correct for gene size. The algorithm computes the average FI score for all observed mutations in the gene, where the FI scores measure the functional impact of the mutations (e.g. from CADD). It next calculates the average score for the observed mutations and compares that with the distribution of averages when drawing mutations randomly. If the observed average score is more often higher than the average from randomisations, the gene is considered significantly mutated.
→ Actually you are right. OncodriveFM does not correct for gene size but the use of FI score does the job. However, your p_value is very marginally significant (0.02) and for only one algorithm, and you are not taking into account that you tested all mutants (more than 3000 genes / more than 3000 tests) without any correction for multiple testing. Without any biological validation such as aneuploidy, your results are not convincing at all.
-TP53: "Mutations in TP53 are common in many cancers but have not been described in UM to date." False: see Hajkova et al Sci Rep 2018Hajkova et al. 2018 presented two cases with germline TP53 variants (of unknown pathogenicity) proposed to be associated with Li-Fraumeni syndrome. What we report here is the common acquisition of somatic TP53 mutations, the majority having confirmed pathogenicity. and ref, Rodrigues et al CCR 2019. We have referenced the one somatic TP53 mutation described in Rodrigues et al. as well as a second case in the MSK-IMPACT data set; these observations were, however, single instances in cohorts of 42 and 43, respectively. Here we report for the first time that TP53 is a statistically significantly mutated gene in UM. We have changed the text to: "Somatic mutations in TP53 have been described in two UM. A hotspot mutation p.R175H (n = 1216 in IARC TP53 database) was observed in a hypermutated metastatic UM with deficient MBD4 and another hotspot mutation p.M237I (n = 196 in IARC TP53 database) was observed in a UM in a pan-cancer study of metastatic tumours. Here we identified TP53 as an SMG and report six somatic TP53 mutations across four tumours in addition to eight cases of LOH" → OK -RPL5 : None of these mutations are associated with LOH. One case is hypermutated. We acknowledge that none of the UM has a double hit of RPL5, which is expected as RPL5 has been described as a haploinsufficient tumour suppressor (Fancello et al. Oncotarget, 2017). We have extended the discussion on this point to make it clearer for the reader. → OK L151 "RPL5 along with RPL11 and ARF are responsible for inhibiting MDM2 and stabilising p53 in response to blocks in ribosome biogenesis and nucleolar stress, often triggered by oncogene-induced translational stress. Supporting the importance of this pathway in UM, TP53 and RPL5 were altered in 42% of cases in this cohort." This is an overstatement supported by no evidence, but that the longest chromosome (chr1) is frequently altered in UM. The discussion on RPL5 has been rewritten and the text above has been deleted.
→ As for CENPE, your p_value for RPL5 is very marginally significant (0.02) and for only one algorithm, and you are not taking into account that you tested all mutants (more than 3000 genes / more than 3000 tests) without any correction for multiple testing. Without any biological validation such as aneuploidy, your results are not convincing at all. 5) L167 "Interestingly, while iris melanomas are associated with earlier detection and favourable prognosis, 4 of 8 iris cases had M3, with two already having progressed to metastatic disease". None of these metastatic patients received immunotherapy? Unfortunately, neither of the two patients in the iris UM cohort who have developed metastases received immunotherapy -thus we cannot correlate response to treatment. We hope our report may lead to the analysis of previous and future metastatic iris UM patients who receive immunotherapy to determine if a correlation with improved response rates exists. → OK Minor comment -The first publication of UM characterized by WGS (Furney Cancer Discov 2013) should be cited We have added the Furney et al. reference. → Thank you -L97 "These mutations were generally mutually exclusive except for two PLCB4 p.D630 mutations. As seen in the UM TCGA cohort, overlap occurred with the minor GNAQ p.R183Q/ GNA11 p.R183C. hotspots rather than the stronger oncogenic p.Q209 hotspot mutations." Confusing: something like "two PLCB4 p.D630 co-occurred with GNA p.R183" would be more straightforward. Thanks for the suggestion. We have changed the main text to: "These mutations were generally mutually exclusive except for two PLCB4 p.D630 mutations that co-occurred with GNAQ/GNA11 p.R183H mutations. This co-occurrence between PLCB4 mutation and the minor hotspot p.R183, rather than the stronger oncogenic p.Q209 hotspot mutations, has previously been described in the UM TCGA data." → Thank you -L145: "These data suggest the two mutations are biallelic". No, the authors suggest composite heterozygous mutations or mutations in trans.