HSP90α plays an important role in piRNA biogenesis and retrotransposon repression in mouse

HSP90, found in all kingdoms of life, is a major chaperone protein regulating many client proteins. We demonstrated that HSP90α, one of two paralogs duplicated in vertebrates, plays an important role in the biogenesis of fetal PIWI-interacting RNAs (piRNA), which act against the transposon activities, in mouse male germ cells. The knockout mutation of Hsp90α resulted in a large reduction in the expression of primary and secondary piRNAs and mislocalization of MIWI2, a PIWI homolog. Whereas the mutation in Fkbp6 encoding a co-chaperone reduced piRNAs of 28–32 nucleotides in length, the Hsp90α mutation reduced piRNAs of 24–32 nucleotides, suggesting the presence of both FKBP6-dependent and -independent actions of HSP90α. Although DNA methylation and mRNA levels of L1 retrotransposon were largely unchanged in the Hsp90α mutant testes, the L1-encoded protein was increased, suggesting the presence of post-transcriptional regulation. This study revealed the specialized function of the HSP90α isofom in the piRNA biogenesis and repression of retrotransposons during the development of male germ cells in mammals.


Figure S2 Length distributions of transposon-derived small RNAs and configuration of 16-and
19-nt small RNAs.
(A, B, C) Small RNAs in Fkbp6,Tdrd12,and Hsp90α KO libraries (red) and their WT controls (blue) were mapped to transposon consensus sequences, and mapped reads of indicated lengths were counted.
The read counts were normalized by miRNA levels in the respective libraries as reads per million miRNA reads. (D) The reduction by the KO mutations was calculated for RNAs of the indicated lengths.
The Fkbp6 and Tdrd12 mutants exhibited a severe reduction in piRNAs of 28 nt or longer (typically MIWI2-bound), whereas the Hsp90a mutants displayed a reduction in piRNAs of 24-27 bp (typically MILI-bound), in addition to piRNAs of 28 nt or longer. In the Tdrd12 mutant, MIWI2 does not bind small RNAs, and the slicer activity of MILI is proposed to be affected (Pandey et al. 2013, Proc. Natl. Acad. Sci. USA 110:16492-16497). Our analysis revealed that 16-and 19-nt RNAs were diminished in the Tdrd12 mutant, which is consistent with the inference that these 16-and 19-nt RNAs are byproducts of the slicing reactions to generate secondary piRNAs.
(E, F) Mapping positions of 16-and 19-nt RNAs derived from sense (E) and antisense (F) strands of the L1 retrotransposon were compared with those of other small RNAs. Configuration of 16-and 19-nt small RNAs suggests they are byproducts of RNA cleavage during the ping-pong cycle. The blue and red bars denote the exact read counts for the indicated RNA in the indicated configurations. (1) Sense-strand 19-nt RNAs whose 3′-ends are at the same position as the 3′-end of at least one sense-strand 16-nt RNA.
(3) Sense-strand 19-nt RNAs whose 5′-ends are at the same position as the 3′-end of at least one antisense-strand 29-nt RNA. (4) Sense-strand 16-nt RNAs whose 3′-ends are at the same position as the 3′-end of at least one sense-strand 19-nt RNA. (5) Sense-strand 16-nt RNAs whose 3′-ends are 1 bp upstream of the 5′-end of at least one sense-strand 24-33-nt RNA. (6) Sense-strand 16-nt RNAs whose 5′-ends are at the same position as the 3′-end of at least one antisense-strand 26-nt RNA. (7) Antisense-strand 19-nt RNAs whose 3′-ends are at the same position as the 3′-end of at least one antisense-strand 16-nt RNA. (8) Antisense-strand 19-nt RNAs whose 3′-ends are 1 bp upstream of the 5′-end of at least one antisense-strand 24-33-nt RNA. (9) Antisense-strand 19-nt RNAs whose 5′-ends are at the same position as the 3′-end of at least one sense-strand 29-nt RNA. (10) Antisense-strand 16-nt RNAs whose 3′-ends are at the same position as the 3′-end of at least one antisense-strand 19-nt RNA. (11) Antisense-strand 16-nt RNAs whose 3′-ends are 1 bp upstream of the 5′-end of at least one antisense-strand 24-33-nt RNA. (12) Antisense-strand 16-nt RNAs whose 5′-ends are at the same position as the 3′-end of at least one sense-strand 26-nt RNA. If a sense RNA satisfies (2) and (3), then the sense 24-33-nt RNA and antisense 29-nt RNA are involved in the ping-pong cycle, as shown in Fig. 3E. Expected values (gray) were the averages calculated from 1000 simulated datasets in which the mapping positions of 16-or 19-nt RNAs were randomized. Error bars indicate standard deviations. The p-values were calculated from the aforementioned simulations, and statistical significance is indicated by asterisks (*** p < 0.001, ** p < 0.01, * p < 0.05). Table   Table S1. Oligonucleotides used in this study