A third of untreated colorectal cancers (COAD) at first diagnosis show focal evidence of activation of the DNA damage machinery, as shown by colocalization of multiple independent members of the repair complex. This phenotype represents an all-or-nothing feature and not a continuum among all cases. There are no distinct histopathologic, immunologic or clinicopathologic characteristics associated with this phenotype. The outcome after the post-surgery adjuvant chemotherapy treatment of these patients with DNA-damaging drugs (5FU etc.) was not different from the rest of the patients.
This phenotype was previously reported.
Oka et al. 31 reported a gradual increase in phosphorylation of ATM, H2AX and CKH2 from normal tissue, to adenoma and to COAD in 55 patients. Oxidative, replicative and mechanical stress was the suggested cause for DDR activation in non-apoptotic cells during cancer progression. Takabayashi et al. 32 found variable co-localization of ℽH2AX and 53BP1 in foci in 33.9% of 56 COAD, but did not elaborated about the cause for this phenotype. Neither manuscript provided additional phenotypic or clinicopathologic data.
DDR is a restraining checkpoint on the road to cancer, as shown in four seminal papers dating 2005 and 2006 7, 9, 11, 33. Since an intact ATM response is required to assembly the DDR foci 9, we assume that in the DDR + group of COAD this arm of the response must be preserved, at least in an hemizygous state, as shown by the co-assembly of several of its members (ℽH2AX, pATM, pCHK2, 53BP11, pNBS1) (31, 32 and this report). Indeed, ATM inactivating mutations have been found in only 7% of non-hypermutated, TP53 wild-type COADs 34.
Immunostaining on serial sections suggested a coexpression of proliferation and DDR markers 7, 33 and a dissociation in cancer between response scores and senescence 9.
Our high-plex IF data show a different scenario, made of a majority of COADs which are non-informative because of lack of DDR phenotype, apoptosis or senescence, and a substantial minority which displays a DDR phenotype which is at variance to the expectation.
These latter COAD cases contain DDR + glands which are non-proliferating, negative for senescence markers (p16, p21) and in which TP53 expression is irrelevant or diminished, compared to the surrounding tissue.
To this point, a more nuanced picture has emerged over the years, adding up to the findings published in 2005 and 2006.
Individual lesions of members of the DDR (ATM, ATR, CHK1, CHK2), in the presence of replication stress response, cause an incomplete escape from quiescence or apoptosis, allowing the cells to resume growth 35. While ℽH2AX and pCHK2 remain active, p21/CDKN1A is post-transcriptionally downregulated via MDM2 in a MEK/ERK dependent, TP53 independent fashion 35.
This phenotype, which we describe for the first time in vivo, may allow the progression to cancer and may be retained in a third of full-blown COAD, while is being changed to a DDRnot phenotype in the others. Worth noting that cancers stem cells from continuously growing cell lines seems to show this behavior 35.
What is the cause of the phenotype? DNA replication stress 2 seems the unifying motif for the many aspects which may lead to ℽH2AX phosphorylation (the broadest and least specific hallmark of a constellation of mechanisms) and the loading of DDR molecules into the nucleus. DNA replication stress may lead to multiple abnormalities which include replication forks stalling, DNA breaks, chromosome missegregation, immune activation and cell cycle arrest. This latter aspect which we observed in our COAD cases may derive by multiple actors such as availability of metabolic constituent of the DNA replication 29, 36 or out-titration of essential cofactors 2. Furthermore, asynchrony of the causes and effects of the replication stress have been observed and may cause the microheterogeneity of the phenotype we have described.
We did not look for genetic lesions underlying the DDR phenotype; worth noting that no single genetic abnormality alone may identify a third of COADs. It is however conceivable that multiple separate lesions may contribute to cause DNA replication stress 2, 12.
Consistent with published data, 12% of our COAD cohort had a MSI-H phenotype and in the majority of the cases (87%) MLH1 was absent. The MSI-H phenotype was equally represented in the DDR + and DDRnot groups, however none of the MSI-H DDRℽH2AXHi was MLH1 protein negative. The presence of alternative MSI gene abnormalities was over-represented in the DDRℽH2AXLo phenotype.
MSH6, but not MLH1 neither PMS2, is part of the DDR complex loaded on the replication stress sites 8. Besides this, there is little evidence for a role of the MLH1 in the DDR complex, thus the suggestion of the requirement of an intact MLH1 to express the full DDR phenotype remains unexplained.
We reasoned that the DDR + tumor areas might be the site of punctuated karyotipic evolution 37 in COAD with ongoing copy number variation. To prove that, we used a FISH probe for the most represented chromosome alteration in COAD, chr20qAmp 27 and we targeted the analysis to the DDR + phenotype-containing areas (and controls).
For the first time, ch20qAmp has been assessed by multi-region FISH analysis, which provides spatial dimension to the chromosome status, as well as a distinction between subchromosomal amplification and polysomy. In only one case of the tested, 20qAmp was confirmed in a subclone, the remaining being polysomic.
Different high-throughput technologies (ArrayCGH, SNP Microarray, NGS) have been used to study chromosomal copy number aberrations in colorectal cancer and chromosome 20q gain/amplification is the recurrent anomaly in microsatellite stable subtype. FISH, a low- throughput molecular cytogenetic test developed in the 1980s, remains the only test capable of discriminating between amplification and polysomy.
Unexpectedly, we found uniformity in chr20q asset across all areas of all tumors, without any evidence of variegation or clonal evolution, at least with the probes we used. These data are confirmed by Bollen et al. 37, where the chr20qAmp seeding core for a progeny of organoids did not modify the chr20q status over multiple rounds of divisions and punctuated karyotypic evolution of other segments. It is possible that clonal aneuploidies, such as chr20q, represent the fixed outcome of a selection process 38 which is not subject to further evolution. Private subclonal copy number variations or rearrangements in a fully established tumor may require single cell in-situ genomics 39 in order to investigate the occurrence of punctuated evolution at DDR + sites.
Worth considering that both in the senescence model 40 and in the experiments showing a mitosis-associated ℽH2AX increase 26, no other DDR proteins are colocalized.
Evidence of a type I interferon response or accumulation of intratumoral lymphocytes remained elusive, as well as linking a DDR phenotype with response to DNA damaging therapy. Regarding this latter aspect, personalized therapy and basket innovative trials have produced remarkable therapeutic advances in subsets of cancers otherwise chemotherapy-resistant 41, thus the evidence of a substantial number of potentially treatable COAD cases 3, 14, 15 should prompt additional translational work.
Our work shows a robust assay for this subpopulation of cancers, which can be determined on routinely processed material.
This study has limitations. An IHC assay on routinely processed material or acquisition of bidimensional IF images at relatively low numerical objective aperture (NA 0.75) do not reach the resolution required to resolve the subcellular aggregates of DDR members, because of the combination of optical constraints, antigen availability and image quality (e.g. zeta stacks, confocal images). The analytical balance of this work is tilted toward a robust assay, which can be applied in routine diagnostics, at expense of greater analytical sensitivity and a detailed, more complex set of data.
Another limitation is the inability to reproduce data obtained with the same or equivalent reagents, however this falls in the problematic of science reproducibility, particularly evident with antibody based assays 42.
Lastly, assessing the DDR within a complete multi-omics picture would require not only single cell genomics and epigenomics but single cell proteomics, to address post-translational changes of labile regulatory proteins such as the ones involved in DNA damage response.
The identification of a novel COAD subgroup, based on a post-transcriptional DDR + phenotype, detectable by routine diagnostics methods, is of biological interest and may be susceptible of therapeutic targeting if the molecular mechanism is further explored in detail.