Reciprocal constraint couples architectural protein abundance and pericentromeric satellite expansion

Long-read sequencing has enabled precise measurements of highly repetitive centromeric satellites and their rapid divergence between species^1-7. Large satellite arrays emerge from libraries of shorter arrays via stochastic expansions^8,9, but understanding the selective pressures constraining such expansions remains a major challenge. Here, using the mouse "major" satellite as a model, we reveal reciprocal functional constraints between increasing satellite copy number and abundance of a conserved architectural protein in female meiosis. We show that HMGA2 (high mobility group AT-hook 2) is enriched at major satellite, and its expression correlates with major satellite copy number: both are high in Mus musculus compared to the closely related Mus spretus. To test functional constraints, we modulated HMGA2 abundance by depletion or overexpression and used a musculus/spretus hybrid to generate oocytes with intermediate HMGA2 expression and major satellite copy number. We find that HMGA2 depletion disrupts major satellite packaging in major satellite-rich musculus but not hybrid oocytes, indicating that increasing copy number requires high HMGA2 expression. Conversely, HMGA2 overexpression disrupts chromosome segregation in major satellite-poor spretus but not hybrid oocytes, indicating that high HMGA2 expression requires expanded major satellite arrays. Based on these results, we propose a co-evolution model in which satellite expansion is constrained by architectural protein abundance, whereas protein abundance is constrained reciprocally by satellite array size.