Summary: The fundamental cause of chromosomal instability in colorectal cancer remains elusive. Genetic deletions of the MACROD2 tumor suppressor could play a major role. Cancer Discov; 8(8); 921–3. ©2018 AACR.
See related article by Sakthianandeswaren et al., p. 988.
In colorectal cancer, genomic instability is classically divided into chromosomal instability (CIN) and microsatellite instability (MSI; ref. 1). CIN is a common feature of cancer and the principal type of genomic instability found in solid tumors (2). CIN is the continuous evolution of aneuploidy through gains and losses of whole or partial chromosomes. By contrast, MSI is genetic hypermutability leading to frequent single-base mutations and is named for the readily observed alterations in repetitive DNA known as microsatellites.
The origin of MSI is well understood to be inactivating mutations and silencing of genes involved in DNA mismatch repair, including MLH1, MSH2, MSH6, and PMS2. This can occur either through inherited germline mutations (Lynch syndrome) or through somatic alterations specific to the tumor. In contrast to MSI, the origins of CIN remain far more elusive. Today, decades after MSI was measured and mechanisms were established, the various types and origins of CIN remain unclear. The very measure of CIN still poses a technical challenge and has evaded standardization.
Basic research has demonstrated that CIN could arise from a large number of genetic aberrations. One key set of candidates are genes that regulate mitosis, a time in the cell cycle when genetic integrity is imperiled by the potential for chromosome missegregation. During mitosis, replicated DNA is condensed in 46 pairs of sister chromosomes, and each pair must be attached to both spindle poles before cohesion is irrevocably sundered, to distribute one sister into each new cell. Signals from kinetochores regulate this process such that a halt signal, the mitotic checkpoint, is generated until all chromosome sets are properly attached. Errors in this process produce whole-chromosome gains/losses in daughter cells (Fig. 1, top). Thus, mutation of mitotic genes was proposed as the likely origin of CIN two decades ago (3).
Indeed, in model systems, CIN is introduced by disruption of genes involved in mitotic regulation (2). Moreover, in humans a genetic disorder, mosaic variegated aneuploidy, which is characterized by CIN and enhanced cancer risk, is caused by germline mutations in the mitotic checkpoint gene BUB1B (4). However, the rarity of this disorder and the relative dearth of BUB1B mutations in sporadic cancers suggest this is a highly uncommon cause of CIN in human cancer. Moreover, large-scale genomic analyses such as The Cancer Genome Atlas have failed to identify common mutations in other mitotic regulators, and thus this mechanism appears insufficient to explain the widespread CIN observed in cancer. One proposed explanation for this difficulty is that such mutations could be distributed broadly over diverse mitotic regulators, such that no one gene is frequently mutated (5), but this idea remains to be validated. Another intriguing alternative is that CIN is nongenetic and conferred by epigenetic regulation of gene expression or cell properties such as centrosome amplification that arose in an initiating oncogenic event that is propagated to progeny. It is also possible that CIN is genetic but arises from genes that operate outside of mitosis.
Bolstering the idea that genetic alterations in nonmitotic genes contribute to CIN, cancer karyotypes are often complex, exhibiting both structural and numerical aberrations (Fig. 1, bottom). Structural aberrations arise primarily from double-stranded DNA breaks. Mutations in genes that regulate DNA repair can yield such DNA breaks and disfavor repair by homologous recombination. Nonhomologous end joining can produce dicentric or acentric chromosome fragments that are highly prone to later mitotic chromosome missegregation. Such errors could cause chromosome gains/losses through other indirect effects on mitosis, such as secondary defects of sister chromosome cohesion or failure of telomere maintenance (5). These nonmitotic mechanisms would favor complex CIN, consisting of both structural and numerical chromosome abnormalities. This complex type of CIN faithfully recapitulates the genomic structures seen in many solid tumors, arguably more than a purely mitotic defect. Many genetic alterations have been proposed as contributors to this type of CIN, including MRE11, PIGN, MEX3C, and ZNF516 (6, 7), among others, but no single gene has been identified that could explain the origin of complex CIN in a substantial fraction of colorectal cancers.
In this issue, Sakthianandeswaren and colleagues identify small-scale genomic microdeletions encompassing portions of the MACROD2 gene as another cause of CIN in human colorectal cancer and, less commonly, in other gastrointestinal malignancies (8). As opposed to many of the aforementioned genetic causes of CIN, MACROD2 microdeletions are common—found in nearly 1 of every 3 sporadic human colorectal cancers. The study commences with the observation of inactivating microdeletions in this gene and proceeds to comprehensively evaluate the role of MACROD2 haploinsufficiency in PARP function, CIN, and oncogenesis. MACROD2 haploinsufficiency in mice with APCMin background increases the number and size of intestinal adenomas. Mechanistically, the MACROD2 macrodomain is required for PARP1-dependent recruitment to sites of DNA damage, and the disruption of MACROD2 protein impairs activity of PARP1 transferase. MACROD2 haploinsufficiency effectively interrupts PARP1 activity, leading to altered DNA repair and increased tumor cell sensitivity to DNA damage. Additionally, MACROD2 haploinsufficiency causes CIN reminiscent of PARP inhibition, with aberrant cell divisions both producing secondary mitotic defects and resulting in structural and numerical chromosomal aberrations (Fig. 1, bottom). Thus, Sakthianandeswaren and colleagues reveal that MACROD2 is a haploinsufficient tumor suppressor that operates principally in gastrointestinal malignancies.
Because of the many prior proposed genetic causes of CIN in human cancer, it will be important to establish the relative importance of this mechanism of CIN versus those proposed in other studies. It is possible that MACROD2 haploinsufficiency, though common, may not contribute substantially to the CIN in human cancers—or that this type of CIN does not strongly promote tumor initiation and progression. Indeed, the effects of MACROD2 haploinsufficiency on oncogenesis in the APCMin mouse model was modest—increasing the number of tumors per mouse from ∼60 to ∼70 and tumor area from ∼180 to 200 mm2. The study did find concordant results in the highly distinct genetic backgrounds of mouse embryonic fibroblasts and HCT116 colorectal cancers—a major strength. Nevertheless, it will ultimately be necessary to evaluate the extent to which MACROD2 haploinsufficiency contributes to CIN in human colorectal cancer and to evaluate the relative contributions of this genetic lesion with alternate proposals for genetic and nongenetic causes of CIN.
A major barrier to this validation, and to progress in the field, is a relative lack of direct quantitative and comprehensive measures of numerical and structural CIN in human tumors. Existing direct measures of CIN such as karyotype analysis are impractical because it is necessary to culture the cancer cells, not to mention the labor involved in sampling a sufficient number of karyotypes to measure cell–cell variation. FISH analysis has been used as a more convenient alternative, but is limited to sampling of only a small number of genomic regions, and remains laborious (5). Single-cell sequencing of genomic DNA now promises to establish comprehensive cell-by-cell digital karyotypes without requiring culture of tumor cells. Once this method is routine, it will be possible to use it to establish the degree and type of CIN in human cancer, which will facilitate evaluation of the relative contribution of MACROD2 haploinsufficiency and other proposed causes.
It is instructive to contrast MSI with CIN biomarkers in tumor biology. MSI is arguably simpler—it is thought to be a dichotomous measure of the presence or absence of DNA mismatch repair, readily detectible with molecular technologies in routine use in the 1990s, and arises from defects in only four genes in human tumors. By contrast, the simplest direct measurement of CIN requires measuring the rate of chromosome gains/losses per cell division. This is in and of itself a challenge. However, given the likely diversity of mechanisms and types of CIN, it may be necessary to establish more complex measures, or at least separate measures for the rate of structural and numerical chromosome gains/losses.
It has been long proposed that CIN is a therapeutic vulnerability of cancer (9). Here, too, CIN has lagged behind MSI. MSI was recently established as a predictive biomarker for immune checkpoint therapy (10). To likewise establish CIN as a predictive biomarker, it will be critically important to establish standardized measures to dissect the relative contributions of its fundamental causes. Once this is established, it may be possible to develop rational therapeutic strategies for each particular mechanism at play. The discovery of MACROD2 microdeletions as an original cause of CIN that is quite commonly found in colorectal cancer is remarkable and an important step toward this goal.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
This work was supported in part by Carbone Cancer Center Support Grant P30 CA014520. N. Jin is supported by T32 HL007899. We thank B.A. Weaver, R.A. Denu, and K.D. Burkard for feedback on this article.