Meiotic recombination: hundreds of deliberate DNA breaks per cell, made by a topoisomerase relative, at sites chosen by a single zinc-finger protein

Budding yeast; the mammalian requirement was established subsequently by knockout mice not cited here.

• 'Catalysed by Spo11' names one subunit of the enzyme. Type IIB topoisomerases are obligate heterodimers of an A subunit (the Spo11-like catalytic subunit) and a B subunit, and the B subunit went unrecognised in meiosis for nearly twenty years: a TopoVIB-Like protein family was identified as an essential, conserved partner required for meiotic double-strand break formation in mouse and plants. The 1997 sequence-homology assignment could not see this, so the covalent-intermediate chemistry is right but the enzyme it belongs to was only half described. It matters for the TOP2 analogy this graph draws, because it means the meiotic machine has a dedicated partner-subunit requirement that TOP2 does not. Robert T et al. 2016

Published back-to-back with E3 and should be cited as a pair; PRDM9 explains a large fraction but not all of hotspot placement, and PRDM9 is pseudogenised in some vertebrates that still recombine.

• The knockout falsifies the strong form of the claim. Prdm9-null mice are proficient at initiating recombination: hotspots are still observed, still at H3K4me3 marks, but relocated to promoters and other sites of PRDM9-independent H3K4 trimethylation that are rarely used in wild-type animals. The authors' conclusion is that PRDM9's role is as much to SEQUESTER the recombination machinery AWAY from gene promoters and functional elements as to create hotspots. 'Determines the positions of hotspots' and 'is required for hotspots' are different statements and only the first survives. Brick K et al. 2012
• A mammalian lineage disproves necessity outright. Canids lack a functional PRDM9, yet a fine-scale linkage-disequilibrium map from 51 free-ranging dogs shows elevated recombination near CpG-rich regions including gene promoters, with little association with spermatocyte H3K4 trimethylation. Recombination is therefore hotspot-structured without PRDM9 at all, by a mechanism the PRDM9 account does not describe. Auton A et al. 2013

Population-genetic inference rather than direct biochemistry; the causal demonstration comes from E2 and E4 together.

• Population-genetic drive inference cannot see what direct break mapping sees. High-resolution maps of meiotic double-strand breaks in individual human genomes show that, along with DNA binding by PRDM9, ADDITIONAL factors dictate the efficiency of break formation - so between-individual differences in hotspot usage are not read out from PRDM9 genotype alone. A motif-drive signal in allele-frequency data is consistent with PRDM9 determining hotspots and also with PRDM9 being one of several determinants whose binding sites happen to be the ones under conversion bias. Pratto F et al. 2014

Sperm-typing in a limited number of donors; effect sizes vary by hotspot and by allele.

• Generalisability bound that sperm-typing cannot address: these are male gametes. Mapping meiotic double-strand break hotspots in male and female mice found grossly different recombination landscapes arising from up to fifteen-fold differences in hotspot USAGE between the sexes, with most recombination occurring at sex-biased hotspots, partly determined by chromosome structure and by DNA methylation, which is absent in females at the onset of meiosis. An allele's measured hotspot activity in sperm is therefore not that allele's activity in oocytes, and the paper explicitly notes this contradicts past findings in humans. Brick K et al. 2018

Mouse; nucleosome mapping is population-averaged.

• A symmetric nucleosome-depleted region centred on the motif does not establish that PRDM9 binding suffices to make a break. Humanising the PRDM9 DNA-binding domain in mice showed that altering one Prdm9 allele changes the behaviour of breaks controlled by the OTHER allele at chromosome-wide scales, that these effects track the degree to which a variant binds BOTH homologues at the sites it controls, and that this binding symmetry acts downstream in the recombination process. Population-averaged nucleosome maps at motif-centred sites average over exactly the per-homologue asymmetry that decides which bound sites proceed. Davies B et al. 2016

One genomic disorder; NAHR is the dominant but not universal mechanism for recurrent genomic rearrangements, and counter-examples exist.

• PLP1 duplications causing Pelizaeus-Merzbacher disease do not follow the canonical clustered-NAHR model, supporting an alternative mechanism. Lee JA et al. 2006

Inference of mitotic origin rests on breakpoint distribution and parental-origin data rather than direct observation.

• The mitotic-origin inference predicts post-zygotic mosaicism, and non-mosaic type-2 NF1 deletions exist. Two patients with non-mosaic (germline) type-2 NF1 deletions have been characterised clinically, with a phenotype matching germline type-1 deletion patients. A germline type-2 deletion cannot have arisen mitotically in the patient, so 'type-2 deletions arise by mitotic intrachromosomal recombination' describes a tendency in the reported series rather than a mechanism that excludes a meiotic or parental-germline route. Vogt J et al. 2011