We found that the D. melanogaster ferlin gene misfire is expressed in the testis and ovaries of adult flies, and has tissue-specific transcription initiation sites, alternatively spliced mRNAs, and multiple developmental functions. Our analysis of cDNAs, while not exhaustive, has provided a framework for predicting potential Mfr isoforms and for interpreting the differential effects of mutations on male and female fertility.
Thus far, ferlins have been described as having multiple C2 domains, typically four to six, and a TM domain at the C-terminus. Our cDNA study confirmed this ferlin structure for the predicted protein isoform T1, which we suggest from genetic data is the Mfr protein required for male fertility. However, we also isolated cDNAs that predict shorter testis and ovary isoforms. Alternative splicing appears to be common in the ferlin gene family [2, 13, 14, 28–30]. For human OTOF, Northern blot analysis shows that both short and long transcripts are produced . The functional significance of the short transcript is unknown, but its structure is strikingly similar to that predicted for Mfr isoforms T2 and T3. Also, a subset of mfr cDNAs (T4, O1, and O3) predicts isoforms that contain one to three C2 domains, but lack the TM domain. cDNAs that predict ferlin isoforms lacking TM domains have also been reported for human MYOF and C. elegans fer-1 [2, 30]. Further biochemical studies are needed to determine which ferlin isoforms are produced and whether variation in C2 domains or presence of soluble versus membrane forms contributes to function.
Ferlin proteins are known to have a role in Ca2+-dependent membrane-membrane interactions in mammalian muscle cells, myoblasts, and inner ear hair cells [7–10], and C. elegans spermatids [2, 15]. Therefore, Mfr may also mediate Ca2+-dependent interactions between membranes. In males, Mfr's time of action may be during spermatogenesis to affect sperm function during fertilization. However, its multiple effects on PMBD and later stages of sperm activation suggest activity of the protein during fertilization. After insemination, Mfr could facilitate interactions between the acrosome and sperm plasma membrane to elicit PMBD. These interactions may also involve Snky, an acrosomal membrane protein that like Mfr, we propose is acting as a signaling protein for PMBD . Previously it was proposed that PMBD might occasionally occur spontaneously, allowing sperm produced by snky mutant males to effectively bypass the requirement for Snky and produce a few progeny . Here, we show that a subset of sperm lacking Mfr function progress past PMBD to achieve nuclear decondensation and pronuclear apposition, but produce no progeny. Later events of sperm function, perhaps those associated with nuclear envelope dynamics or pronuclear apposition, may also require one or more Mfr isoforms. Possible Ca2+ cues for Mfr activation may come from an intracellular source such as the acrosome, which is a Ca2+-storage vesicle in sperm of some species [31, 32], or from cytoplasmic sources in the activated D. melanogaster egg .
Only three of the male sterile mutations, which are all located downstream of the C2D encoding region, induce detectable effects on female fertility. The locations of these mutations are consistent with the idea that a short ovarian isoform, which includes the C2E and C2F domains, is important for Gurken localization and consequently egg patterning. However, the incomplete penetrance of the mfr phenotype even with presumed null alleles mfrZ0695, mfrZ1386, and mfrZ4070 suggests that other proteins can at least partially compensate for loss of Mfr. In addition, embryos produced by mfrZ4070/mfrZ1386 mothers show developmental delay and/or arrest during the earliest mitotic divisions. During these early mitotic divisions, D. melanogaster embryos are rapidly dividing in a syncytium and do not undergo transcription. Consequently, the early mitotic cycles are under maternal genetic control. The defects observed in embryos produced from mfrZ4070/mfrZ1386 mothers before cycle 13 suggest that mfr mRNA and/or proteins are maternally deposited into the embryo and used during the early mitotic divisions. Microdomains of Ca2+ have been identified in D. melanogaster syncytial embryos undergoing mitosis  and Mfr may associate with these transient Ca2+ signals that are critical for cell division.
For ferlin proteins, the contribution of individual C2 domains to ferlin function remains largely an unanswered question. In addition to biochemical analyses to address this question [7, 9, 10], these studies and those of Washington et al. using C. elegans  show that genetic analysis provides a complimentary approach. For the C. elegans fer-1 gene, a predominance (8/10) of missense mutations were recovered. The location of these mutations within the C2C-, C2E-, and C2F-encoding domains suggested that each of these domains is important for fer-1 function, and that there is little functional redundancy among the C2 domains. In contrast, we recovered an unusually high proportion of mfr nonsense mutations (8/11), which may reflect functional redundancy among C2 domains within the Mfr protein. Alternatively, there may be functional redundancy among protein isoforms expressed in the same tissue, with the introduction of premature stop codons expected to have more general effects on isoform production. However, one critical difference between the C. elegans and D. melanogaster studies is that the majority of the C. elegans mutations were selected as temperature sensitive male sterile alleles, which precludes the recovery of nonsense mutations. For both organisms, now that the mutant phenotypes and gene structures are known, the function of C2 and other ferlin domains can be systematically tested. For instance, the effects of targeted disruptions of individual domains on phenotypes or the ability of individual isoforms to rescue mutant phenotypes can be evaluated.
Finally, we note that while ferlin genes are expressed in the gonads and required for fertility in both D. melanogaster and C. elegans [2, 15], this gene family has not yet been implicated in fertility in mammals. To date, studies of mammalian ferlins have focused largely on tissues that show a disease phenotype caused by ferlin mutations and a role in the gonad has not been explored. However, ferlin transcripts have been detected in the testis and male germline of colts  and mice , and an antibody that recognizes mouse OTOF shows that this protein is located in the testis . It will be interesting to determine if one or more of the mammalian ferlin proteins, like those in D. melanogaster and C. elegans, play a role in fertility. Alternatively, ferlins whose original function in animal evolution may have been specific for the gonad, may have been co-opted for entirely new uses in mammals.