Knockout! Knockout! Who’s Not There?

The deeper we look at human genomes, the more unexpected things we find. And increasingly we are surprised by what each of us lacks. In particular, in recent years we have learned that naturally occurring human gene knockouts are prevalent. The results of the Exome Aggregation Consortium (ExAC) now paint a rich and detailed picture of gene redundancy and essentiality in humans (Lek et al., 2016xLek, M., Karczewski, K.J., Minikel, E.V., Samocha, K.E., Banks, E., Fennell, T., O’Donnell-Luria, A.H., Ware, J.S., Hill, A.J., Cummings, B.B...., and Exome Aggregation Consortium. Nature. 2016; 536: 285–291Crossref | PubMed | Scopus (915)See all ReferencesLek et al., 2016).Protein-truncating variants are surprisingly common in human genomes. Image from iStock.com/adempercem.View Large Image | View Hi-Res Image | Download PowerPoint SlideThe breadth of individuals who had their protein-coding regions of their genomes sequenced by ExAC is impressive, with the effort cataloging variation in more than 60,000 individuals from diverse ancestries. Within this cohort, more than 7 million individual variants are reported, which averages out to a variant occurring every eight base pairs throughout protein-coding regions. Just more than half of these are observed only once—that is, in a single individual—and the vast majority of these were not observed in previous sequencing efforts, such as the 1000 Genomes Project. This suggests that many more variants will be uncovered by sequencing yet more people, but the sample size is sufficient to begin to approach saturation for some kinds of mutation, including the most common type, transition variants at CpG sites. For example, a majority of all of the theoretically possible variants are observed synonymous mutations at CpG sites. With this depth of variant identification, we now have a clear understanding of which individual genes and gene classes are more commonly the target of protein truncating variants. The frequency or infrequency of such protein-truncating variants tells us which are so critical as to never tolerate disruption and those that can seemingly be lost without repercussion. Reassuringly, the gene classes most intolerant of loss-of-function (LOF) protein-truncating variants are core to the basic workings of the cell (components of the ribosome, proteasome, spliceosome, etc.), whereas the gene family that is least constrained by mutational tinkering is olfactory receptors. The grouping of genes least tolerant of protein-truncating variants also contains nearly all known haploinsufficient human disease genes, and the authors speculate that this group is likely to yield numerous other genes with a discernible impact on fitness.How often are protein-truncating variants found in humans? The findings of Lek et al. suggest that there is an average of 35 genes homozygous for protein truncating variation per person, which is in the ballpark of previous estimates (MacArthur et al., 2012xMacArthur, D.G., Balasubramanian, S., Frankish, A., Huang, N., Morris, J., Walter, K., Jostins, L., Habegger, L., Pickrell, J.K., Montgomery, S.B...., and 1000 Genomes Project Consortium. Science. 2012; 335: 823–828Crossref | PubMed | Scopus (555)See all ReferencesMacArthur et al., 2012). This is to say that most of us manage just fine with three dozen genes out of action, and this doesn’t include potential variation in non-coding regions that might also impact a gene’s expression. As mentioned before, not all genes are created equal in terms of fitness costs and benefits, but of the ∼20,000 protein coding genes in humans, the ExAC effort reveals 1,775 genes among those with predicted homozygous LOF genotypes. It remains to be seen how many of these come with discernable phenotypes on close examination. For anyone who has scratched their head at a knockout mouse with seemingly no discernible phenotype—the bane of many a grad student project—the existence of this degree of diversity of human gene knockouts may resonate as much as it shocks. Regardless, these are striking numbers to think about, and they hammer home the notion that, with this richer understanding of our genetic diversity, humans soon will be (or already are) the model organism of choice for the study of genetics.Rare homozygous knockouts in particular should be informative about gene function. In a study reported earlier this year, the exomes of 3,222 British adults of Pakistani heritage were analyzed and used to identify individuals with rare homozygous loss-of-function variants (Narasimhan et al., 2016xNarasimhan, V.M., Hunt, K.A., Mason, D., Baker, C.L., Karczewski, K.J., Barnes, M.R., Barnett, A.H., Bates, C., Bellary, S., Bockett, N.A. et al. Science. 2016; 352: 474–477Crossref | PubMed | Scopus (49)See all ReferencesNarasimhan et al., 2016). This population was chosen for its higher rate of marriage between first cousins, which would be expected to enrich for homozygous gene knockouts. Additionally, the exome analysis could readily be linked to health records to draw potential links with disease. Although the individuals identified harboring rare gene knockouts did not present as unusually unhealthy in this study, as indicated by their rates of medical consultation and drug prescription, there is likely much information yet to be gleaned. And as with the findings of Mongkol et al., the phenotypic effects of homozygous loss-of-function genotypes need to be explored individually.But this begs the question, what kind of things can we learn from “knockout humans”? Narasimhan et al. provide a nice example through the characterization of an individual identified with LOF variants in PRDM9, a histone methyltransferase that influences locations chosen for meiotic recombination. Analysis of the individual and her child shows that sites of meiotic recombination were indeed localized away from PRDM9-dependent hotspots. The fact that she had a child might not have been predicted from the existing knockout mice, which are infertile. Neither do the resulting changes in recombination sites match the overall pattern observed in dogs, which lack PRDM9 altogether, suggesting that there are yet unexplored mechanisms for directing crossover events in humans. It seems that what we lack or can do without is fundamental to understanding our uniquely human biology.