A noncanonical meiosis that separates chromosomes without crossovers
Drosophila males build sperm without the canonical synaptonemal complex and crossover formation used by Drosophila females and most eukaryotes. Instead, during an extended prophase I, the XY bivalent and the two autosomal bivalents resolve into three discrete chromosome territories visible by DAPI staining. This territory formation breaks pairing between heterologous chromosomes that could otherwise cause nondisjunction, so homologs still segregate accurately. The paper treats the dispersive forces behind this separation as unidentified, and asks what molecular players supply them.
The cellular setting makes the question observable. After the final gonial S-phase, diploid spermatocytes enter a roughly 90-hour growth phase and enlarge up to 25-fold in volume, becoming the largest cells in Drosophila melanogaster. Later meiotic stages pass in about 3 hours, so prophase I is where subnuclear structures can be staged and measured. The authors use that window to track two things at once: TBPH protein and long perfectly complementary dsRNA detected by the J2 antibody.
TBPH arrives first and dsRNA follows
TBPH and J2 signal both appear in primary spermatocyte nuclei at the start of prophase I, after the mitotic-to-meiotic transition, with TBPH appearing slightly earlier than dsRNA. Both increase across prophase I while mitochondria multiply and encircle the nucleus. The two signals occupy non-overlapping regions of the nucleoplasm rather than forming one merged body.
Their exits differ, which is the first clue that they do separate jobs. TBPH staining disappears at prometaphase I, coinciding with mitochondrial reorganization away from the nuclear membrane and reduced ATP5a staining. dsRNA instead peaks at prometaphase I and declines rapidly by metaphase I, a transition lasting under 10 minutes in these cells. The authors confirmed the dsRNA signal was real by treating testis squashes with dsRNA-specific RNaseIII, which abolished it relative to buffer controls at all detectable stages, and by immunoprecipitating dsRNA with J2 followed by dot blotting. Similar prophase I patterns appeared in Drosophila simulans and other species, indicating conservation across distantly related Drosophilids.
Human TDP-43 can stand in for TBPH in the testis
TBPH and human TDP-43 share homology in the N-terminal domain and RNA-recognition motifs, but the C-terminal low-complexity domain is far less conserved. To test whether that difference matters, the authors expressed human TDP-43 from a humanized endogenous tbph locus. These flies reached wild-type fertility levels and the expression fully rescued the severe locomotor deficits and shortened lifespan of tbph mutants. In humanized males, hTDP-43 and dsRNA showed spatial and temporal patterns similar to TBPH controls, and in transheterozygotes co-expressing tagged hTDP-43 and TBPH the two proteins colocalized.
Staging individual 16-cell cysts sharpened the comparison. In early prophase I both proteins sat near the nucleolus, seen as a DAPI-negative cavity; by mid-prophase I they formed a loop around the nucleolus and the XY territory; late in prophase I, as territories reached maximal separation, both signals diminished and neither was detectable by prometaphase I. The paper therefore concludes that the non-conserved low-complexity domain is not required for this meiotic localization or for dsRNA patterning.
Disease-linked mutations told a different story. Humanized males carrying heterozygous or homozygous ALS/FTD-associated Acore mutations (Q331K, M337V) showed premature dsRNA accumulation in spermatocyte cysts, ectopic prophase-I cysts appearing distally, fewer spermatid bundles, dsRNA- and ATP5a-positive inter-cyst vacuole-like structures, and disrupted chromosome territories in staged squashes. These mutant males also showed age-related fertility decline. Because these mutations sit in the conserved Acore embedded in the non-conserved low-complexity domain, the result separates two things the domain comparison alone would blur: the domain can be swapped for the human version without losing meiotic function, yet specific disease mutations inside it still compromise the testis.
TBPH and Cocoon push dsRNA in opposite directions
Cocoon is a retrocopy of tbph, generated by a recent retroduplication, that is primarily expressed in the testis. It shares extensive identity with TBPH in the N-terminal and RNA-recognition domains but lacks the C-terminal low-complexity domain, mirroring short-processed TDP-43 isoforms upregulated in ALS/FTD patients. The Fly Cell Atlas single-nucleus RNA-seq data presented in the paper show tbph and cocoon transcripts in the male germline across spermatocyte stages.
The loss-of-function phenotypes run in opposite directions, which is the paper's central regulatory claim. tbph deficiency causes premature dsRNA accumulation and pairing defects, and human TDP-43 expression rescues those defects. Spermatocytes lacking cocoon instead show reduced dsRNA, defective chromosome territory formation, and heterologous chromosome contacts. Quantification of dsRNA fluorescence intensity from whole testes used n = 6, 6, 6, 5 individual testes across groups, and inter-territory distance in mature spermatocytes was measured in n = 43 and 46 individual spermatocytes, with group comparisons by two-tailed unpaired Welch's t test. The authors conclude that TBPH and Cocoon play opposing roles in dsRNA accumulation, so the relevant quantity is not simply "more dsRNA" but dsRNA held at the right level at the right stage.
dsRNA fills the gaps as territories separate
Staged squashes show the spatial progression. In early prophase I, when homologs are not fully paired, TBPH and dsRNA are associated with chromatin. As chromosomes begin resolving into distinct territories, dsRNA accumulates at the nuclear periphery and in loops between the separating autosomes (chrII and III), while TBPH mainly loops around the nucleolus and the associated XY territory. During early-to-mid territory separation, dsRNA extends into the XY territory and sits closely on chromatin with TBPH but shows little overt colocalization. By late prophase I, dsRNA fills the nucleoplasm between territories while being excluded from TBPH-enriched regions, and TBPH expands around the XY territory, appearing as two separate regions on either side. At prometaphase I, TBPH has largely diminished while dsRNA occupies the entire nucleoplasm between territories.
Nuclear dsRNA also colocalizes with Blanks, a dsRNA-binding protein required for male fertility. The authors' most parsimonious model places nuclear import of mitochondrial dsRNA and its association with Blanks at the center of territory formation, possibly through Blanks' reported role in forming or transporting dsRNAs. They propose that electrostatic charges, molecular crowding, and recruitment of RNA-binding proteins such as Blanks into subnuclear bodies promote physical separation of bivalents at the proper stage. They explicitly consider the alternative that dsRNA merely fills space without driving separation, and argue against it because dsRNA levels correlate with territory organization: cocoon mutants with reduced dsRNA have defective territories, while tbph mutants with premature accumulation have pairing defects.
The nuclear dsRNA comes from the mitochondrial genome
The most surprising result concerns origin rather than amount. dsRIP sequencing revealed that the dsRNAs are mitochondrial. The authors address the obvious alternative that these reads come from nuclear mitochondrial DNA segments (NUMTs) integrated into the genome: NUMT content and position vary widely across Drosophilids, yet orchestrated nuclear dsRNA accumulation during prophase I appeared consistently across numerous species. They also found similar enrichment along the mitochondrial genome in dsRIP-seq of both Drosophila melanogaster and Drosophila virilis testes. The most nuclear-enriched regions originate antisense to the 5' region of Cox I and overlap ND6 and CytB on the reverse strand, making the mitochondrial origin itself an evolutionarily conserved feature in their data.
The paper notes that mt-dsRNA accumulation has previously been reported as a pathological marker in disease states and in the cytoplasm, where export activates antiviral sensors such as PKR, RIG-I, and MDA5. Finding abundant nuclear mt-dsRNA in wild-type testis is therefore a departure from that framing. Because the dsRNA appears in the nucleus from early prophase I while the nuclear membrane is intact, the authors propose active import of mt-dsRNA, or of its sense and antisense precursors, across two mitochondrial and two nuclear lipid bilayers. They connect the timing to the dramatic mitochondrial restructuring in spermatocytes, when mitochondria encircle the nucleus, and note that mitochondrial-nucleus contacts have been shown to allow noncanonical translocation of mitochondrial protein components into the nucleus. This import step remains a proposal, not a measured transport mechanism.
Mutant testes shift immune and metabolic gene expression
Bulk RNA sequencing of testes from tbph mutants and two independent cocoon mutants identified differentially expressed genes, with hits defined as absolute log2 fold change of at least 1 and adjusted P value below 0.05, tested by DESeq2 Wald test. The paper highlights a subset of top genes with immune or metabolic functions, and KEGG analysis indicates enrichment of detoxification and nutrient metabolism pathways, with the number of differentially expressed genes per pathway shown in each bar. The authors also report that genes altered in tbph and cocoon deletion mutants are highly enriched in neurons and include known neurodegenerative disease genes, which they use to argue that spermatogenesis is a relevant model for studying human neurodegenerative disease proteins.
These expression data broaden the phenotype but do not establish that the immune or metabolic changes cause the meiotic defects. The causal chain the paper actually supports runs through dsRNA level and territory organization: tbph loss raises dsRNA prematurely and produces pairing defects rescued by human TDP-43, cocoon loss lowers dsRNA and produces defective territories with heterologous contacts, and nuclear dsRNA colocalizes with Blanks. The gene-expression shifts sit alongside that chain as correlated consequences of the same mutations.
What the model explains and what it leaves open
The paper's own framing separates observation from mechanism. The regulated accumulation of nuclear mt-dsRNA during prophase I, its mitochondrial origin, the opposing effects of TBPH and Cocoon, and the rescue by human TDP-43 are supported by the imaging, genetic, and sequencing data presented. The claim that mt-dsRNA actively drives chromosome territory formation through electrostatic, crowding, or Blanks-mediated forces is presented as the most parsimonious model, and the authors acknowledge the alternative that dsRNA passively fills inter-territorial space. The import route into the nucleus is likewise proposed rather than traced.
Scope is set by the system. The work is in Drosophila male meiosis, a noncanonical process lacking the synaptonemal complex and crossovers of canonical meiosis, so the territory-formation readout does not directly transfer to organisms that use those structures. The authors note that murine TDP-43 has been reported to be required for homolog pairing in mouse prophase I, which they cite as support for a broader role, but the mechanistic details in canonical meiosis remain for future work. They also raise, as a possibility rather than a finding, that fertility or gametogenesis deficits could precede and serve as a biomarker for TDP-43-mediated neurodegenerative disease, and that mt-dsRNA may contribute to the etiology of TDP-43-related disease.
The decisive variable is dsRNA level, not its presence
The comparison that should guide how readers use this paper is between the two loss-of-function directions. Losing tbph pushes dsRNA up prematurely and produces pairing defects; losing cocoon pushes dsRNA down and produces defective chromosome territories with heterologous chromosome contacts. Both mutants are subfertile, but they fail from opposite sides of the same quantity. That makes the regulated level and timing of nuclear mt-dsRNA, rather than its mere presence, the variable the evidence supports as functionally relevant, and it is why the authors argue against a purely passive filling role.
A second comparison sharpens the disease angle. Human TDP-43 expressed from the endogenous tbph locus restores wild-type fertility and meiotic dsRNA patterning despite the poorly conserved low-complexity domain, yet the ALS/FTD-associated Acore mutations Q331K and M337V inside that domain still cause premature dsRNA accumulation, ectopic prophase-I cysts, disrupted territories, and age-related fertility decline. For readers weighing TDP-43 variants, the paper's priority is therefore the conserved Acore and the dsRNA set point it controls, with Drosophila male meiosis as the assay that makes both the level and the timing visible.
Sources used in this answer
TDP-43 regulates mitochondrial dsRNA accumulation and chromosome territory formation during prophase I of Drosophila male meiosis
TBPH and its retrocopy Cocoon regulate nuclear accumulation of mitochondrial-origin dsRNA during Drosophila male meiosis prophase I, with tbph loss causing premature dsRNA and pairing defects and cocoon loss reducing dsRNA and disrupting chromosome territory formation.
