If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Division of Molecular and Cellular Oncology, Department of Medical Oncology, Dana-Farber Cancer Institute, Boston, MassachusettsDepartment of Gastroenterology, The First Affiliated Hospital of Chongqing Medical University, Chongqing, China
Division of Molecular and Cellular Oncology, Department of Medical Oncology, Dana-Farber Cancer Institute, Boston, MassachusettsCenter for Functional Cancer Epigenetics, Dana-Farber Cancer Institute, Boston, Massachusetts
Division of Molecular and Cellular Oncology, Department of Medical Oncology, Dana-Farber Cancer Institute, Boston, MassachusettsCancer Program, Broad Institute of Massachusetts Institute of Technology (MIT) and Harvard University, Cambridge, MassachusettsDepartment of Medical Oncology, Gastrointestinal Cancer Center, Dana-Farber Cancer Institute, Boston, MassachusettsDepartment of Medicine, Brigham and Women’s Hospital and Harvard Medical School, Boston, Massachusetts
Through an integrative analysis leveraging patient-derived molecular information, we recently defined the genome-stable subtype of colorectal cancer (CRC), a previously unrecognized subgroup that lacks significant aneuploidy and elevated mutational density.
However, the functional significance of these alterations in CRC remains poorly understood. Prior studies hypothesized a gain-of-function role for the mutant SOX9 based on genomic analyses of human CRC cases.
However, to date, a direct functional analysis of the role of truncated SOX9 proteins has yet to be performed. In this research letter, we annotate SOX9 mutations in CRC, describe their transcriptional and epigenomic consequences, and postulate as to why they are selected for in genome-stable CRC.
In CRC, SOX9 alterations are predominantly nonsense/frameshift mutations that preferentially cluster in 3 functional domains found within the c-terminal half of the gene (Figure 1A and B) . We confirmed that truncated forms of SOX9 are expressed in a subset of CRC cell lines harboring endogenous mutations (Figure A1A and B), often at higher levels than endogenous wildtype (WT) SOX9. Genomic analyses of human CRC from The Cancer Genome Atlas and Cancer Cell Line Encyclopedia indicated that the majority of SOX9 mutations are heterozygous, preserving a WT copy of the gene (Figure 1C). This result raised a few possibilities as to the function of mutant truncated SOX9 proteins: (1) They carry gain-of-function properties by regulating a distinct transcriptional program, (2) they exert dominant-negative activity by inhibiting the WT SOX9 function, or (3) they behave as a null allele leading to a heterozygous state.
To investigate the transcriptional and epigenomic consequence of mutant SOX9 as well as distinguish between these possibilities, we pursued a comprehensive genome-wide molecular analysis of mutant and WT SOX9. We conditionally overexpressed 4 SOX9 protein constructs with N-terminal V5 protein tags in HT-115 CRC cells. These included 1 WT construct and 3 mutant SOX9 alleles with sequential loss of its c-terminal domains, representing the spectrum of mutations observed in patients (Figure 2A and B) . Truncated SOX9 mutants were expressed at higher levels than WT SOX9, which may reflect greater tolerance to elevated expression of truncated variants. The V5 protein tag ensured the specific assessment of the conditionally overexpressed WT and mutant SOX9 without engaging endogenous SOX9 expressed in the cells (Figure 2C). Genome-wide binding of mutant and WT SOX9 was determined using chromatin-immunoprecipitation followed by DNA sequencing (ChIP-seq) using an anti-V5 antibody. Mutant and WT SOX9 were bound to identical locations throughout the genome (∼1750 sites, Figure 2D); the motif analysis of these sites showed greatest enrichment for the native SOX9 binding sequence as determined by SeqPos (P = 1 × 10−336). Histone H3 lysine 27 acetylation (H3K27Ac) ChIP-seq, which marks active enhancers and promoters participating in transcriptional regulation, was also performed on these cell lines. All V5- and H3K27Ac-ChIP sequencing analyses were performed on 2 high-quality replicates in each experimental condition. Most of the loci identified contain a moderate, baseline level of H3K27Ac due to endogenous WT SOX9 activity, as shown by green fluorescent protein expressing control cells. However, enrichment analysis indicated that WT but not mutant SOX9 binding increased H3K27Ac and thus transcriptional activity when compared to the green fluorescent protein control (Figure 2E). Integration of V5-CHIP-seq data and messenger RNA expression profiles from RNA-sequencing using the binding and expression target analysis pipeline endorsed that only WT SOX9 is functioning as a transcriptional activator, whereas mutant SOX9 variants appear to lack transcriptional activity (Figure 2F). Global H3K27Ac and RNA-sequencing data did not show evidence of mutant SOX9-specific transcriptional activity, suggesting that a gain-of-function behavior is unlikely.
HT-115 CRC cells endogenously express WT SOX9, which enables the evaluation of whether overexpression of mutant SOX9 variants interferes with their transcriptional activity through a dominant-negative effect. Despite binding to the exact same locations, mutant SOX9 did not exert dominant-negative activity as the genes upregulated by WT SOX9 were unaffected by all 3 mutant SOX9 variants (Figure 2F; purple line would be significantly above static dotted line if dominant-negative activity was present). These results revealed that SOX9 mutations lead to expression of truncated SOX9 proteins that are incapable of canonical transcriptional activity, do not show evidence of gain-of-function properties, and do not interfere with the WT SOX9 function through dominant-negative activity. In other words, SOX9 mutations likely lead to loss of function, overexpression of nonfunctional mutant alleles, and yield hemizygous expression of WT SOX9.
We next asked whether a truncating mutation can be introduced into WT SOX9 CRC cell lines using CRISPR/Cas9. Directing Cas9 to the c-terminus of SOX9 using sgRNA#9 in 3 CRC cell lines led to protein expression of a truncated variant at the expected size (Figure A2A). Despite the constitutive expression of the CRISPR/Cas9 machinery, WT SOX9 expression was found when analyzing protein from a pooled population in each of the cell lines. By analyzing single-cell clones (Figure A2B and C), it was clear that homozygous SOX9 alterations were infrequent. Two CRC cell lines did not tolerate homozygous mutations in any surviving clones, and a third, COLO-205, produced a minority of clones containing a homozygous mutation (3/24). These results suggest that, if mutated, SOX9 is preferentially altered in a heterozygous fashion. Notably, these data also raised the possibility that WT SOX9 is required for CRC as homozygous alterations occur infrequently if at all. These findings were followed up in a recently published companion manuscript
focused on the requirement for WT SOX9 in CRC. Given that ∼10% of CRCs harbor a heterozygous SOX9 mutation that preserves a WT copy, it opens further areas of investigation into the advantage provided by a haploinsufficient SOX9 state.
Here, we provide evidence that an important downstream mediator of WNT signaling, SOX9, is preferentially mutated in a heterozygous fashion in CRC. We additionally found that SOX9 mutations cluster in the c-terminal half of the gene, consisting mostly of nonsense and frameshift mutations, and lead to overexpression of truncated forms of SOX9 that lack gain-of-function, dominant-negative, or canonical transcriptional activity. Early studies of SOX9 assigned a potential gain of function to these mutations given their predilection for the c-terminus, leading to overexpression of truncated variants in human tumors.
Gain-of-function mutations have been characterized in a number of cancer genes including TP53; the pathogenic significance, however, is now under debate with recent evidence showing that mutational processes may shape the disproportionate frequency of hotspot mutations
When evaluating the functional significance of SOX9 mutations, we set a high threshold for gain-of-function properties, scrutinizing any evidence of novel DNA-binding, noncanonical gene expression changes, or phenotypic advantages. Ultimately, it appears that truncating mutations lead to a nonfunctional allele, which raises new questions about potential advantages of single-copy gene expression in cancer.
TCF7L2, which encodes the essential WNT pathway cofactor TCF4, is also recurrently mutated in a heterozygous fashion in CRC (Figure A2D and Table). The prevalence of heterozygous, inactivating mutations in another essential WNT pathway component suggests a functional cancer-specific advantage to this mutational pattern (Table). Losing 1 copy of a pathway regulator could relieve negative feedback on WNT while still maintaining moderate expression of genes that are essential for preventing terminal differentiation of intestinal cells, such as SOX9.
For example, overexpression of WT but not truncated mutant SOX9 blocks WNT signaling as indicated by reporter activity in 293T cells (Figure A2E). Recently, the inhibitory activity of SOX9 on the WNT pathway was demonstrated to be mediated by MAML2.
Another potential advantage of heterozygous mutations is supported by the dose-dependent nature of SOX9 function, which is exemplified in normal intestinal stem cells where high SOX9 expression promotes quiescence.
This would support the hypothesis that a heterozygous mutation attenuates SOX9 expression while retaining the necessary functions as a WNT mediator and intestinal stem cell factor. A third possibility is that heterozygous SOX9 mutations confer phenotypic plasticity, defined as the ability to change cell states, which is increasingly being recognized for its contributions to cancer initiation, progression, metastasis, and drug resistance.
We speculate that the acquisition of plasticity may underlie selection of heterozygous null SOX9 mutations in CRC. This “goldilocks” level of SOX9 protein expression may enable the flexibility to achieve different cellular states when facing selective pressures of neoplasia. Even with the pieces of evidence presented in this study, which was limited by reliance on cancer cell lines, better model systems and more systematic evaluation are required to rigorously examine whether heterozygous SOX9 mutations confer plasticity. These experiments will be critical for determining the mechanism by which heterozygous SOX9 mutations serve a functional advantage in CRC.
Conflicts of Interest: This author discloses the following: N.S. is a consultant for Astrin Biosciences. The remaining authors disclose no conflicts.
Funding: This work was funded by the Claudia Barr Award and Virtual Scholar Award from the Department of Defense (CA201084) to N.S.
Ethical Statement: The corresponding author, on behalf of all authors, jointly and severally, certifies that their institution has approved the protocol for any investigation involving humans or animals and that all experimentation was conducted in conformity with ethical and humane principles of research.
Data Transparency Statement: All RNA-sequencing and ChIP-sequencing data and generated reagents (eg, organoids, cell lines, and/or plasmids) will be shared either by depositing in public domains or through material transfer agreements in compliance with our institution.