Tardigrades have a complete condensin II: two divergent CAP-H2 kleisins in three eutardigrade genomes

Abstract

Condensin II shapes interphase genome architecture: species that lack it tend to fold their chromosomes in a Rabl-like, type-I configuration, while species that have it form chromosome territories (type II). In the study that established this, the tardigrade Hypsibius dujardini (now H. exemplaris) was the only species of 24 in which neither condensin's accessory subunits could be found. It was counted among the type-I species that lack condensin II, and the authors left open whether the subunits had diverged beyond recognition. Here we searched the annotated proteomes of three eutardigrades and found both complexes complete. Reciprocal BLAST recovers condensin I and the condensin II HEAT subunits CAP-D3 and CAP-G2. A profile hidden Markov model built from 33 animal CAP-H2 sequences finds the missing kleisin: each species has two CAP-H2 genes. Both carry the Pfam CNDH2_N and CNDH2_C domains, score against CAP-H2 and not CAP-H, and form a single tardigrade clade within CAP-H2 in a maximum-likelihood kleisin tree (ultrafast bootstrap 94 for the placement, 100 for the clade). Conserved gene order dates the duplication to before the split between Hypsibius and Paramacrobiotus. Copy a sits at an orthologous locus in all three species, and the b copies of Hypsibius and Paramacrobiotus share a conserved neighbourhood. In H. exemplaris, CAP-H2 is co-expressed with CAP-D3, CAP-G2, SMC2 and SMC4 across 48 developmental and adult samples; its correlations with them rank in the top 0.1–0.3% of 14,851 expressed genes. In single embryos, all condensin II subunits are expressed through the first ~46 hours after laying and fall to near zero thereafter, while CAP-H persists. Structure predictions (Boltz-2) place the C-terminal region of five of the six tardigrade CAP-H2 proteins on their own species' SMC4 head with confidence comparable to human CAP-H2 (ipTM 0.76–0.81 against 0.83), while wrong-partner and scrambled controls score 0.19–0.59. With a longer SMC2 head-and-neck construct, the N-terminal regions of the b copies of H. exemplaris and R. varieornatus are placed on the SMC2 neck in all five samples (ipTM 0.62–0.69, against 0.57–0.61 for human CAP-H2 and sample medians of 0.11–0.19 for scrambled controls), on the same stretch of the neck as the human kleisin. In these two species copy b is therefore predicted to bind both SMC heads, in separately predicted interfaces. The two Ramazzottius varieornatus CAP-H2 genes were already listed in a pan-eukaryotic survey (van Hooff et al. 2025); the H. exemplaris and Paramacrobiotus genes, the correction of the species record behind the architecture study, the paralog history (from phylogeny and gene order) and the expression data are new here. With the tardigrade reclassified, the condensin II–architecture association still holds (Fisher's exact p = 0.019, from 0.007). The tardigrade becomes the seventh species with a complete condensin II and a type-I call, a call that rests on a single manually annotated feature in whole-animal Hi-C. Early embryos, where condensin II is expressed and chromosomes remain individualised through interphase, are the place to test whether tardigrade genome architecture changes with developmental stage. Version 2 added structure predictions (Boltz-2) of the CAP-H2 regions bound to the condensin ring, with controls (Fig. 4). Version 3 adds the N-terminal regions predicted with a longer SMC2 construct (Fig. 4c): the b copies bind the SMC2 neck like human CAP-H2, so the earlier 'no signal' reflected the construct. Version 4 corrects version 3: SMC2 contact residue numbers (off by the construct linker), narrower wording on ring closure and on one C-terminal interface, and the controls' limits, after an independent recomputation of all structure numbers (Codex, OpenAI); all ipTM values, ranges, medians and interface counts were reproduced.