CSI Genetics. A headline in the recent issue of Nature Genetics caught my eye the other day: “Hereditary mixed polyposis syndrome is caused by a 40-kb upstream duplication […]”. Two thoughts went through my head when I read this headline. First, I was wondering how the authors actually found this upstream duplication (one of the last things you look for). Secondly, I realised that demonstrating causality of this duplication must have been very, very difficult. In fact, the paper by Jaeger and colleagues reads like a good detective story and tells several lessons that might apply to EuroEPINOMICS.
The defendant. Hereditary mixed polyposis syndrome (HMPS) is an autosomal dominant genetic disorder that leads to various kinds of polyps in the gastrointestinal tract. Polyposis, i.e. the presence of multiple polyps are medically relevant, since the risk for colorectal carcinoma is markedly increased. Between various polyposis syndromes, HMPS stands out as an entity that presents with various, unrelated kinds of polyps. Ever since the identification of a large Ashkenazi Jewish family with HMPS, this condition was assumed to be monogenic, and by including a second Ashkenazi family, a locus could be identified on 15q13.3 (the next door neighbour of the 15q13.3 microdeletion, so to speak). Other families with HMPS helped strengthen the evidence for this locus – and all families carried the same haplotype. For the scientific detective teams, this was evidence beyond reasonable doubt that a founder mutation was at play. But at this point, they didn’t yet know what they had gotten themselves into.
The smoking gun. Three genes were possible culprits for HMPS: GREM1, SCG5 and FMN1. Despite various attempts, no mutations were found in these genes. However, the authors realised something unusual that raised their suspicion: HMPS is rarely seen outside the Ashkenazi population. Therefore, could it be possible that the causative mutation is due to an event that is very specific and very unlikely to happen spontaneously? This is when they turned towards copy-number changes as possible causes. Using a high-resolution custom array-CGH, they searched the region for duplications and deletions – and they eventually found a 40 kB duplication spanning the end of SCG5 upstream of GREM1. But duplications at this range are frequent and often of unknown significance. Could a variant like this really cause disease? They would need some more evidence to make this claim.
Strengthening the case. The authors went on to test 40 patients with HMPS and 50 unaffected individuals. All patients with HMPS had this duplication, but not a single control individual. However, the authors needed more support for their case. They investigated existing data for a genome-wide association study for colorectal cancer with ~700 cases and ~900 controls. None of the controls carried this duplication, but a single case was positive. Jaeger and colleagues pressed further and found out that this patient also was of Ashkenazi descent and had HMPS. At this point, the duplication had little chance of escaping, but could there be anything else on the haplotype that might simply be linked? Using a tour de force, the authors sequenced the entire haplotype with Sanger sequencing, which resulted in ~450 amplicon (PCR reactions for sequencing). They did not find other mutations. The 40 kB duplication was the culprit. But how could a variant like this cause disease?
Reconstructing the crime scene. The three genes in this region GREM1, SCG5 and FMN1 had something to do with the way that this duplication causes HMPS. But what exactly happened? The authors looked at gene expression of these three genes in gut tissue (colonic crypts) and found an interesting result. SCG5, the gene affected by the duplication is not affected by this duplication at all. Neither is FMN1. However, GREM1 is highly upregulated and ectopically expressed. The duplication does something to the GREM1 promotor that causes upregulation of this gene and expression in tissues where it is otherwise silent (ectopic expression). GREM1, coding for the gremlin 1 protein, is thought to be an antogonist to signalling of bone morphogenic proteins (BMPs). In fact, some evidence points to a role of BMPs in the generation of polyps, suggesting a possible mechanism of how excessive gremlin 1 protein results in polyposis.
A precedent for other monogenic disorders. In 2003, Page and colleagues summed up the criteria in genetic research required for showing causation in complex genetic disorders. Interestingly, even though a monogenic disorder is involved, the unusual mutation mechanism evokes similar questions: when is the evidence sufficient to believe that this duplication is causative? On a genetic level, the authors set a clear precendent of what would be expected from other monogenic disorders with similarly complicated mutation mechanisms: (a) a clear association with the phenotype, (b) absence in a large control cohort, (c) identification of the same mutation in additional cases (not necessarily unrelated for founder mutations). Given the phenotypic complexity of the epilepsies and the difficulties in accessing brain tissue for gene expression studies, making the case for a comparable mutation in seizure disorders will be challenging.