Abstract
The Arp2/3 complex nucleates essential branched actin networks in most eukaryotes. Interestingly, the subunit Arp2 encodes two splice variants that differ merely by five amino acids in the D-loop, which is critical for actin polymerization. However, it is unknown if this alternative exon, or "microexon," impacts Arp2/3 function or is even expressed. Here, we found that the Arp2 microexon has been evolutionarily retained for over 600 million years yet varies in sequence. We investigated the unique microexon in Drosophila Arp2 and found that the long splice variant encoding the microexon ("Arp2L") is expressed, though not as highly as the shorter variant ("Arp2s") in some tissues. We purified recombinant Drosophila Arp2/3 containing Arp2s or Arp2L and found no differences in actin polymerization rates in vitro. To test for functional divergence in vivo, we replaced Arp2 in D. melanogaster with the coding sequence of either Arp2s or Arp2L. Both splice variants fully rescue the Arp2-knockout lethality phenotype, yet they functionally diverge in sperm development, in which Arp2L-expressing flies exhibit defects in the alignment and motility of actin cones, structures that separate syncytial sperm. The microexon sequence, rather than the increased length of the D-loop, is responsible for cone defects. Despite these fitness costs, our evolutionary experiment suggests that encoding Arp2L provides an overall fitness advantage. These findings reveal that despite not exhibiting intrinsic differences in vitro, the Drosophila Arp2 splice variants are non-redundant in vivo, and the microexon sequence is likely specialized for tissue-specific roles.