BIOLOGICAL OVERVIEW

Soon after fertilization, pole cells bud from the posterior end of the blastula [Images], the earliest cellular phase of embryonic development. Pole cells constitute the germ line for the next generation. In males, germ line stem cells, which will eventually locate to the larval testis, are the precursors of the sperm. The genital disc, a structure derived from three abdominal segments, gives rise to the somatic cells of the testis.

As larval development in the male begins, so too does the process of spermatogenesis. At a later point in larval development, there is a switch from ß1 tubulin to ß2 isotype, but before this will be discussed in detail, a brief review of spermatogenesis is in order. At the start of spermatogenesis, only spermatogonia (the precursor cells of sperm) are observed. Approximately 28 hours after hatching both spermatogonia and primary spermatocytes (premeiotic sperm cell precursors) are observed. This condition remains unchanged for the larval testis until shortly before pupation.

The differentiation of cells from both germ line and somatic origin are required for spermatogenesis. Precursors for these cells, the germline stem cells and the somatic cyst-progenitor cells, are attached to a specialized somatically-derived structure at the tip of the testis known as the hub. At the hub, each stem cell is closely associated with two cyst progenitor cells. These three stem cells undergo unequal cell divisions: the parental stem cells remain attached to the hub, while the three daughter cells (one spermatogonium and two cyst cells) detach and initiate a program of differentiation (Castrillon, 1993).

This group of three daughter cells, one germline and two somatically derived cells, is the early cyst. The transformation of a single spermatogonium to 64 sperm occurs within a thin envelope formed by two cyst cells. Four rounds of mitotic division result in a cyst of 16 early primary spermatocytes interconnected by cytoplasmic bridges or ring canals. These spermatocytes enter a growth phase in which they undergo a 25-fold volume increase (Castrillon, 1993).

It is at this point that a switch occurs between ß1 tubulin and the ß2 isotype (Buttgereit, 1993b). In Drosophila testis stem cells, ß1 tubulin is found in mitotically active germ cells and all somatic parts of the testis, but starting with early spermatocytes, the ß1 isotype is switched off and all microtubular arrays contain ß2 tubulin. The tubulin isotype switch between ß1 and ß2 in the germ line can be attributed mostly to transcriptional control. Since ß1 mRNA and/or protein are very rapidly degraded in the transition from early to mid-spermatogenesis, factors other than transcriptional control may be involved in the isotypes switch (Buttgereit, 1993b). For example, it has been shown that an 18-bp AT-rich element, present in the 5' untranslated regions of the ß2 tubulin mRNA, is responsible for stabilizing ß2 about threefold as compared to mRNA without the element (Michiels, 1993).

During the spermatocyte growth phase a great deal of transcription takes place; the bulk of these mRNAs are stored and not translated until after meiosis. The more caudally located primary spermatocytes undergo the first mitotic division - an event not infrequently occurring in larva commencing to pupate. The late (meaning ready to undergo meiosis) primary spermatocytes enter meiosis, giving rise to 64 haploid spermatids. The nucleus of each spermatid reforms and the mitochondria fuse to form the mitochondrial derivative or nebenkern. At the onion stage, so-called because of the multilamellate appearance of the nebenkern in electron micrographs, nebenkern and nucleus are closely associated and highly uniform in size and shape. Each young spermatid will transform into a mature sperm through a complex process of cytodifferentiation. This process involves nuclear condensation as well as the dramatic elongation of the axoneme and the mitochondrial derivative to form the sperm tail. The final steps of spermatogenesis are individualization, the process by which each elongated spermatid becomes tightly invested in its own membrane, and coiling, which results in the cyst of 64 sperm being drawn to the base of the testis. During coiling, grossly abnormal sperm are segregated and subsequently degraded. The remaining sperm are transferred to the seminal vesicle for storage (Castrillon, 1993).

A dominant mutation in the structural gene for ß2 tubulin results in sterile males. These males produce no motile sperm. Electron microscopic examination of such testis show both overall disorganization of spermatid components and aberrant substructure of the axonemal microtubules in developing spermatozoa. In addition, such testes exhibit profound meiotic defects indicative of failure of microtubule function, including abnormal formation or failure to form the meiotic spindle, improper or absent chromosome movement and failure to undergo cytokinesis (Kemphues, 1982 and references).

Meiotic divisions are abnormal in recessive male sterile mutations of ß2 tubulin. In addition, all microtubule-mediated events subsequent to ß2 tubulin expression are defective: meiosis, nuclear shaping and assembly of the axoneme all fail to occur. Centriolar bodies are rarely observed; when they are seein, they are usually not attached to the nucleus. Centrioles are randomly distributed; nuclei are either rounded or misshapen; endoplasmic reticulum is ofter arranged in many-layered whorles, and mitochondrial derivatives exhibit varying degrees of elongation, but no microtubules of any functional class are observed in spermatids of mutants. It is concluded that the ß2 tubulin subunit that forms the sperm axoneme is not functionally restricted, but serves multiple functions in spermatogenesis, including the assembly of tubules (Kemphues, 1982).

Mutations in ß2 tubulin result in phenotypes of varying severity. Mutations in amino acids that characterize the specific ß2 isotype disrupt axoneme formation, while mutations in highly conserved amino acids that are common to all ß tubulins disrupt microtubule assembly, a function common to several different ß tubulin isotypes. Heterozygote flies carrying both types of mutations complement each other to produce young with mild to moderate meiotic defects but they do not complement for proper axonemal morphology. (Fackenthal, 1995).

Every stage of spermatogenesis are simultaneously present in early pupa; sperm are already present 24-30 hours after the onset of pupation. At eclosion the testis contains principally transforming spermatids and maturing spermatogoa, but in the apical portion of the testis there still remain spermatogonia and a fair number of spermatocyte cysts in various stages of development. In normal flies, these cells continue their normal course of successive divisions and differentiation to form sperm, well into the adult fly stage of senile infertility (Cooper, 1950).

Axonemes are ancient organelles that mediate motility of cilia and flagella in animals, plants, and protists. The long evolutionary conservation of axoneme architecture, a cylinder of nine doublet microtubules surrounding a central pair of singlet microtubules, suggests all motile axonemes may share common assembly mechanisms. Consistent with this, alpha- and ß-tubulins utilized in motile axonemes fall among the most conserved tubulin sequences, and the ß-tubulins contain a sequence motif at the same position in the carboxyl terminus. Axoneme doublet microtubules are initiated from the corresponding triplet microtubules of the basal body, but the large macromolecular 'central apparatus' that includes the central pair microtubules and associated structures is a specialization unique to motile axonemes. In Drosophila spermatogenesis, basal bodies and axonemes utilize the same alpha-tubulin but different ß-tubulins. ß1 is utilized for the centriole/basal body, and ß2 is utilized for the motile sperm tail axoneme. ß2 contains the motile axoneme-specific sequence motif, but ß1 does not. The 'axoneme motif' specifies the central pair. ß1 can provide partial function for axoneme assembly but cannot make the central microtubules. Introducing the axoneme motif into the ß1 carboxyl terminus, a two amino acid change, confers upon ß1 the ability to assemble 9 + 2 axonemes. This finding explains the conservation of the axoneme-specific sequence motif through 1.5 billion years of evolution (Nielsen, 2001).

Attempts were made to identify motile axoneme-specific sequences in Drosophila ß2 tubulin by constructing chimeric ß-tubulins in which selected residues in ß1 were changed to ß2 identity. Chimeras were expressed in ß2's normal domain in the postmitotic male germ cells and tested for their ability to support axoneme assembly. Previous experiments had showen that, when tested in such conditions, ß1 can not make a motile axoneme with central pair microtubules. Instead of the canonical wild-type 9 + 2 axoneme architecture, ß1-mediated axonemes have 9 + 0 architecture. Only 25 of 446 amino acids differ between ß1 and ß2; thus, small changes in primary structure must mediate ß2's ability to make motile 9 + 2 axonemes. The last 15 amino acids in the ß-tubulin protein comprise the highly variable carboxyl terminus, which has been identified as an isotype-defining region of the molecule and is important for axoneme morphogenesis. One third of the amino acid differences between ß1 and ß2 lie within the carboxyl terminus. Moreover, in the absence of the ß-tubulin carboxyl terminus, coherent axonemes can not be initiated at the basal body at all. Therefore, in the first chimera tested (ß1-ß2i), the ß1 carboxyl terminus was replaced in its entirety with that of ß2. ß1-ß2i supports assembly of 9 + 2 axonemes, demonstrating that the ß2 carboxyl terminus carries information sufficient for assembly of the central pair microtubules. This observation led the authors to test the specific role of the axoneme motif. Remarkably, introducing the axoneme motif into the ß1 carboxyl terminus -- a two amino acid change -- allowed ß1 to make 9 + 2 axonemes (ß1-ß2ii) (Nielsen, 2001).

Although ß1-ß2i and ß1-ß2ii support assembly of 9 + 2 axonemes, neither generate functional sperm. Axonemes assembled from these chimeras fail to maintain structural integrity for the full length of the sperm tail. There are thus two separable aspects of axoneme-specific function intrinsic to the Drosophila ß2 isoform. The axoneme motif specifies the central pair, but other features of the ß2 molecule are required for distal axoneme integrity. Comparison of the overall phenotypes of ß1-ß2i and ß1-ß2ii shows that distal axoneme structure is better in ß1-ß2i males, including retention of central pairs. In middle and distal cross sections, 16 of 34 intact axonemes in males with ß1-ß2i (as the sole source of ß-tubulin) had central pairs, but, in ß1-ß2ii males, only 10 of 71 intact axonemes retained central pairs. Thus, features in the ß2 carboxyl terminus other than the axoneme motif, as well as features of the ß2 molecule other than the carboxyl terminus, are required for axoneme structural integrity (Nielsen, 2001).

Additional changes were made in ß1-ß2i to test the function of internal ß2 residues in axoneme morphogenesis. The internal variable region (amino acids 55-57) was tested because previous work showed it to be important for morphology of axoneme doublet microtubules. Amino acid 349 was tested because the tubulin crystallographic structure revealed that this residue contacts the alpha subunit and cysteine occurs at this position only in Drosophila ß2, its identical D. hydeii homolog, and in the moth Heliothis virescens testis-specific ß-tubulin. Like ß1-ß2i, chimeras carrying internal ß2-specific residues are able to make each component of the normal 9 + 2 axoneme architecture but cannot maintain a complete sperm tail (Nielsen, 2001).

All of the ß1-ß2 chimeras are compatible with axoneme motility and full male fertility when they are coexpressed with ß2 and comprise 50% or less of the postmitotic ß-tubulin pool. However, coexpression of any of the chimeric ß-tubulins at a ratio of 2:1 with endogenous ß2, causes defects in axoneme morphogenesis and male sterility. Quantitation of axoneme phenotypes in 2:1 genotypes allowed subtle differences in the functional properties of the different chimeras to be distinguised. ß1-ß2i is most effective at maintaining axoneme structure, but the other chimeras exhibit differential capacity relative to ß1 for retention of the central pair microtubules (ß1-ß2i > ß1-ß2iv > ß1-ß2iii > ß1) and maintenance of the organization of the outer nine doublet microtubules (ß1-ß2i > ß1 > ß1-ß2iv > ß1-ß2iii). Thus, despite increasing identity to ß2, the internal changes introduced into ß1-ß2i decrease functionality in axoneme assembly. In addition, like ß1, the chimeric ß-tubulins cause insertion of additional doublets into the axoneme to generate 10-doublet axonemes, as well as assembly of ectopic cytoplasmic doublet microtubules. The ß1-ß2 proteins are thus truly chimeric -- they possess ß2-specific features required for correct initiation of 9 + 2 axonemes but, nonetheless, retain essential ß1-like features incompatible with ß2 function, reflected in the capacity for de novo generation of doublet microtubules (Nielsen, 2001).

The proportion of axoneme defects increases with distance from the basal body, indicating either that axonemes become progressively defective as they grow or that distal structure fails to be maintained after assembly. The possibility that ß2-containing dimers might be preferentially incorporated early in axoneme morphogenesis was considered. In this model, an increasing gradient of chimeric ß1-ß2 dimers would be generated along the length of the axoneme, corresponding to the progressive loss of distal structure observed. As the most stringent test of this hypothesis, immunolocalization of ß1 along the length of axonemes was examined in males with two copies of ß1 and one of ß2. ß1 is uniformly distributed. Thus, loss of distal axoneme integrity does not result from differential usage of tubulin heterodimers during assembly but more likely reflects accumulation of slight differences in axoneme geometry over the long length of the Drosophila sperm tail (Nielsen, 2001).

The axoneme motif most likely mediates central pair assembly through isotype-specific interactions with other proteins. The carboxyl terminus is a surface feature both in the dimer and in microtubules; thus, sequence changes are unlikely to influence tubulin function by changes in the 3D structure. The carboxyl terminus is a site both for MAP binding and for posttranslational modifications. For example, ß-tubulin carboxyl terminus polyglycylation is necessary for motility in Tetrahymena. The observation that altering internal ß1 residues to ß2 identity can decrease functionality argues that their normal function requires amino acid interactions that obtain only in the ß2 protein. Amino acid residues 55-57 as well as 349 are involved in interprotofilament contacts; an alteration in these contacts could potentially affect the geometry of the entire axoneme (Nielsen, 2001).

The data presented here show that specific changes in ß-tubulin residues produce discrete effects on axoneme morphogenesis. A deeply conserved feature of ß-tubulins used in motile axonemes, the axoneme motif, specifies an equally ancient structural feature, the central pair microtubules. In contrast, sequences in variable regions and other internal residues affect an evolutionarily labile feature, the length of the axoneme, and may have coevolved in ß2 to support the exceptionally long sperm tails within the genus Drosophila: from 2 mm in D. melanogaster up to the giant 5.8 cm in D. bifurca (Nielsen, 2001).

Axoneme beta-tubulin sequence determines attachment of outer dynein arms

Axonemes of motile eukaryotic cilia and flagella have a conserved structure of nine doublet microtubules surrounding a central pair of microtubules. Outer and inner dynein arms on the doublets mediate axoneme motility. Outer dynein arms (ODAs) attach to the doublets at specific interfaces. However, the molecular contacts of ODA-associated proteins with tubulins of the doublet microtubules are not known. This study reports that attachment of ODAs requires glycine 56 in the β-tubulin internal variable region (IVR). In Drosophila spermatogenesis, a single amino acid change at this position results in sperm axonemes markedly deficient in ODAs. Moreover, it was found that axonemal β-tubulins throughout phylogeny have invariant glycine 56 and a strongly conserved IVR, whereas nonaxonemal β-tubulins vary widely in IVR sequences. These data reveal a deeply conserved physical requirement for assembly of the macromolecular architecture of the motile axoneme. Amino acid 56 projects into the microtubule lumen. Imaging studies of axonemes indicate that several proteins may interact with the doublet-microtubule lumen. This region of β-tubulin may determine the conformation necessary for correct attachment of ODAs, or there may be sequence-specific interaction between βtubulin and a protein involved in ODA attachment or stabilization (Raff, 2008).

Axoneme bending, and thus flagellar beating, occurs when dynein arms move along the adjacent doublet. Dynein-mediated motility requires ATP-dependent interactions of the dynein heavy-chain motor domain with C-terminal regions of the β-tubulin component of the α,β-tubulin dimers of the B tubule. At the other end, the outer dynein arms (ODAs) are anchored to the doublet A tubule via a suite of proteins associated with the base of the arm. The discovery of a specific sequence requirement in axonemal β-tubulin for attachment of ODAs provides the first information about molecular interactions between tubulins of the A tubule and proteins involved in attaching ODAs (Raff, 2008).

Distinct β-tubulin isotype classes with conserved expression patterns were identified in vertebrate β-tubulin families, based on sequences in two variable domains, the final C-terminal tail (CTT) and a smaller internal variable region (IVR) comprising amino acids 55-57. Other β-tubulin families are not homologous to the vertebrate families, but a role for the variable regions in functional specialization is conserved. Using Drosophila spermatogenesis as a model, specific β-tubulin CTT requirements have been identified for motile axonemes, including an axoneme motif shared by all axonemal β-tubulins. The sequence requirement for ODAs demonstrated in this study is the first identification of a distinct function for the IVR (Raff, 2008).

Many β-tubulins were examined for their ability to support axoneme assembly in Drosophila spermatogenesis, and it was observed that in all but two of them, ODAs are reliably present in most cross-sections of axonemes. ODAs were deficient in axonemes utilizing β-tubulins with variant IVRs, but even substantial changes in C-terminal sequences did not affect ODA attachment. The most elegant example is the B2t⁶ mutation in the Drosophila testis-specific β2-tubulin; this mutation has a single change at amino acid 56 in the IVR. B2t⁶ mutant males make intact but nonmotile sperm axonemes in which the majority of doublets lack ODAs. Attachment of inner dynein arms (IDAs) is unaffected. The B2t⁶ mutation is fully recessive. Heterozygous males incorporate equal amounts of wild-type β2 and B2t⁶ protein into sperm; axonemes have normal morphology, and males are fully fertile. Thus, functional axoneme architecture can accommodate 50% of the B2t⁶ protein (Raff, 2008).

The B2t⁶ mutation also has another phenotype. In addition to the classic '9 + 2' pattern, insect sperm axonemes have an outer ring of nine accessory microtubules, each associated with one of the doublets. Accessory and central-pair microtubules contain a luminal filament, appearing as a 'dot' in cross-section. In B2t⁶ males, the A tubules of most doublets also contain luminal filaments (Fuller, 1988). In Drosophila spermatogenesis, addition of dynein arms precedes the appearance of luminal filaments in the central-pair and accessory microtubules. Therefore, it was unlikely that the failure to add ODAs is secondary to adding luminal filaments to doublet microtubules. It was not possible to clearly demonstrate that the two phenotypes are independent, separable events by using males coexpressing the chimeric protein β3β2C along with wild-type β2. β3β2C, which has the β3 IVR containing nine instead of three amino acids, has a phenotype very similar to that of B2t⁶, but in contrast to the recessive point mutation, β3β2C is strongly dominant (Hoyle, 1995). The ratio of the two proteins present in the male germline was controlled by generating males expressing different copy numbers of β3β2C and β2. Strikingly, this allowed for titration of ODAs. Addition of ODAs depended on the percentage of β3β2C in the total β2-tubulin pool, confirming the role of the IVR in determining ODA addition. Moreover, axoneme morphology in β3β2C-expressing males conclusively showed that addition of ODAs and of luminal filaments in the doublet microtubules occur independently. In axonemes of males with 33% β3β2C, some doublets lack ODAs and some doublets contain luminal filaments, but the two phenotypes are not linked (Raff, 2008).

ODA addition is independent of CTT sequence. Both B2t⁶ and β3β2C have the wild-type β2 C terminus. ODAs were examined in males that coexpressed the parent β3 molecule along with β2. Sperm axonemes of these males exhibited the same ODA phenotype as the chimeric β3β2C protein. Whereas addition of ODAs is independent of C-terminal sequences, other aspects of axoneme functionality, reflected in male fertility, strongly depend on the CTT (Raff, 2008).

The phenotype of sperm axonemes in males that coexpress the moth Heliothis virescens β2 homolog along with Drosophila β2 provides another indicator of the specificity of the IVR's influence on ODA addition. H. virescens β2 causes even stronger dominant disruption of Drosophila spermatogenesis than β3β2C or β3 and thereby severely disrupts axonemes if it makes up 10% or more of the male germline β-tubulin pool. In sperm axonemes of such males, the presence of moth β2 strikingly confers moth-like morphology to the accessory microtubules. However, H. virescens β2 has glycine 56 in the IVR, and even fragmented or partial axonemes have ODAs on the doublets (Raff, 2008).

Sequence comparisons revealed that glycine 56 is invariant in all known axonemal β-tubulins. The data thus strongly support the hypothesis that stable attachment of outer arms depends on the sequence in the β-tubulin internal variable region in all motile axonemes with ODAs. Moreover, the entire axonemal β-tubulin IVR is conserved, with a consensus sequence of TGG. In contrast, nonaxonemal β-tubulins exhibit a range of amino acids at all positions in the IVR (Raff, 2008).

Electron tomography indicates that ODAs contact the A tubule through at least three interfaces, two of which involve proteins identified as part of the ODA docking complex. Other components at the base of the arm have been shown to contact tubulins, but the specific tubulin residues involved have not been previously known. The demonstration of the crucial role for β-tubulin glycine 56 and the conserved axonemal IVR provides the first identification of a specific tubulin sequence required for association of ODAs with axoneme doublets. β-tubulin amino acid 56 lies on the inner microtubule surface facing the lumen. Specific interactions with this region of tubulin are not defined. However, the open nature of the microtubule lattice means that the microtubule lumen is accessible to proteins outside the microtubule. Recent studies of doublets in sea urchin sperm and Chlamydomonas flagella indicate several structures inside the microtubule luminal surface, whose positions suggest they may help to stabilize the doublet or anchor proteins that attach to the outside of the microtubules. The presence of glycine at position 56 of the β-tubulin IVR may determine the conformation necessary to permit these associations. Alternatively, there may be sequence-specific interactions between glycine 56 and protein(s) involved in ODA attachment. Because some ODAs remain present even in B2t⁶ mutant males, glycine 56-mediated interactions may play a role in the efficiency or stability of ODA attachment. The remarkable conservation of the IVR region in all axonemal β-tubulins suggests that the mechanism for ODA attachment is also conserved, consistent with the conservation of some of the proteins of the outer-arm docking complex. Several luminal structures, including the luminal filaments in Drosophila sperm axoneme central-pair and accessory microtubules, are added after microtubule construction. In B2t⁶ and β3β2C males, mislocation or absence of some of the normal luminal proteins in the doublet microtubules might be a factor in the incorrect addition of luminal filaments into doublet microtubules (Raff, 2008).

A key question is whether the conserved sequence of the axonemal β-tubulin IVR reflects its importance for any other aspect of axoneme structure or function. Motile axonemes lacking outer dynein arms occur naturally in sperm of eels; in some insect sperm, including caddisflies and mayflies; and in the flagellated sperm of basal nonangiosperm land plants, including gingko, cycads, ferns, lycophytes, and bryophytes. Outer-arm dynein genes are not present in the genomes of the fern Marsilea and the moss Physcomitrella. The evolutionary loss of ODAs in diverse phylogenetic groups potentially provides a test of whether the IVR sequence remains constrained in the absence of the need for attachment of the ODAs. There are as-yet-insufficient data. However, the IVR is SGG in the two plant species with motile gametes for which complete presumptive axoneme β-tubulin sequences are available, suggestive of the possibility of evolutionary constraint on the IVR sequence even when ODAs have been lost (Raff, 2008).

GENE STRUCTURE

Exons - 4

PROTEIN STRUCTURE

Amino Acids - 446

Structural Domains and Evolutionary Homologs

The three ß tubulins of Drosophila exhibit an 87% homology in overall amino acid sequence. The first and second introns of ß3 tubulin occur in positions identical to those of the first two introns in the vertebrate ß tubulin genes. The fly ß3 tubulin contains an additional six amino acids not present in any other ß tubulin. The position of the third intron differs from that of the vertebrate ß tubulin genes, but is identical in position to the testis-specific ß2 isoform. The testis-specific isoform and the ubiquitously expressed ß1 isoform are highly homologous both to one another and to the vertebrate major neural form. The ß3 isoform is more divergent. It reveals an ancient divergence in ß tubulin sequences that must have occurred at or before the split between lines leading to insects and vertebrates. For ß tubulins there appear to be two major regions that differ in their secondary structures. A predicted alpha-helical region is present in the carboxyl third of the molecule, and a predicted ß-pleated sheet region falls primarily in the amino-terminal two-thirds of the molecule (Rudolph, 1987).

The ß1, ß2 and ß3 tubulin-specific sequences in the chromosomes of Drosophila auraria are found in the same polytene band in region 32C of the 2L polytene chromosome. In contrast, the three beta tubulin genes in D. melanogaster are not closely linked (Scouras, 1994).

As a test of whether orthologous ß-tubulins from different species are functionally equivalent, the moth Heliothis virescens ß2 homolog was expressed in Drosophila testes. Usually Drosophila microtubules exhibit a 13 protofilament number in the accessory microtubules, found associated with the highly conserved pattern of nine doublet microtubules surrounding a central pair of two singlet microtubules. When coexpressed with Drosophila ß2, the moth isoform imposes the 16-protofilament structures, characteristic of that found in the moth, on the corresponding subset of Drosophila microtubules. There is a profound failure of cytoplasmic microtubule function and axoneme assembly resulting from the coexpression of Hvß2 and Drosophila ß2. Thus, the architecture of the microtubule cytoskeleton can be directed by a component of ß-tubulin. There is no previous demonstration of microtubule architecture being specific to a particular tubulin. In Drosophila, incorporation of ß-tubulins into microtubules of different protofilament number depends on cellular context; specialized 15 protofilament microtubules that function in wing maturation contain ß1 and ß3, isoforms that in other cells give rise to the more typical 13 protofilament microtubules (Raff, 1997).

Functional constraint underlies 60 million year stasis of Dipteran testis-specific ß-tubulin

How do proteins evolve while maintaining their function? Previous studies find a highly stringent structure/function relationship between the Drosophila melanogaster testis-specific tubulin β2 (βTub85D) and the spermtail axoneme, such that small changes in the β2 protein render it unable to generate a motile axoneme. This raises the question, how does β2 evolve while maintaining its function? To address this question full- and partial-length β2 sequences were cloned from 17 species of Drosophila and Hirtodrosophila flies spanning 60 Myr of evolution. Not a single amino acid difference is coded among them -- β2 maintains its function by not evolving. Gene genealogical analyses was performed to determine ortholog/paralog relationships among insect tubulins. The Lepidopteran and Dipteran testis-specific β-tubulins are likely orthologs, and surprisingly, despite functioning in the same structure, the Lepidopteran orthologs are evolving rapidly. It is argued that differences in tubulin isoform use in the testes cause the Dipteran axoneme to be less evolvable than the Lepidopteran axoneme, which has facilitated the evolution of a unique amino acid synergism in Drosophila and Hirtodrosophilaβ2 that is resistant to change, contributing to its evolutionary stasis (Nielsen, 2006).

The evolutionary stasis of the Drosophila+Hirtodrosophilaβ2-tubulin protein seems best explained by functional constraint: stringency in the structure/function relationship between β2 and the spermtail axoneme does not permit viable variation in the protein. The complete and long stasis of Drosophila+Hirtodrosophilaβ2-tubulin indicates that there are special constraints on its function that are unique to this clade (Nielsen, 2006).

One such constraint could result from a unique amino acid synergism in the Drosophila+Hirtodrosophila β2-protein. Of the more than 1000 β-tubulin sequences found on GenBank spanning 1.5 Byr of eukaryotic evolution, only Drosophila+Hirtodrosophilaβ2 has Cys29 (all but a few are Gly29), Thr55, and Ala57 (no other β-tubulin is Ala57, most are Gly57). These residues are in contact in the folded protein, forming the neck of an amino acid loop structure involved in inter-protofilament contacts. Functional tests reveal that these amino acids participate in a synergism. In tests of Dmβ2 amino acid function, β2 Thr55 and Ala57 residues reduced β1's ability to support a full-length axoneme, which indicates that their proper function depends on amino acid interactions only present in Dmβ2-a synergism. Synergism raises the problem of path dependence in evolution, as the viability of a substitution at these sites depends on whether a previous, accommodating substitution occurred in the protein. This could have the effect that only a subset of the possible evolutionary paths between ancestral and Dmβ2 identity maintains a functional tubulin. This would significantly reduce the rate of evolution at synergistic sites relative to sites showing additive behavior, such as the carboxy-terminus amino acids. Synergism, in addition to possible constraints resulting from having to support such a long axoneme, may act to keep the entire protein from evolving (Nielsen, 2006).

β2 has evolved in Dipterans since mosquitoes and fruit flies shared an ancestor, though more slowly than in Lepidopterans. This rate difference could be explained by the absence and presence, respectively, of a testis-specific α-tubulin. Drosophila use a major α-isoform (α84B) to support both somatic and axoneme microtubule function, whereas Bombyx divide these functions between a major isoform (BmTUA4) that supports somatic function, and a testis-specific α-isoform (BmTUA6) that supports the axoneme. The major α-isoforms BmTUA4 and Dmα84B are 99% identical; however, BmTUA6 and Dmα84B, both of which support motile axonemes, are only 78% identical. This indicates that motile axonemes can be supported by a wider range of α-tubulin structures than somatic function. If maintaining somatic function constrains major α-tubulin evolution, use of the major α-isoform in the Dipteran axoneme may constrain the evolution of the Dipteran β2 protein, slowing its evolution relative to Lepidopteran β2. Conversely, in Lepidopterans, the presence of highly divergent, testis-specific isoforms (Hvβ2 and BmTUA6) suggests that release of somatic constraint via use of testis-specific α-and β-isoforms has released testis tubulin evolution; this division of labor possibly allows the axoneme to explore more functional space, to be more evolvable, than possible when one member of the dimer has multiple responsibilities (Nielsen, 2006).

This difference in tubulin isoform use may have facilitated the evolution of the unique β2 amino acid synergism in a Dipteran, as opposed to a Lepidopteran lineage. If use of a major α-isoform and testis-specific β-isoform in the axoneme is conserved among Dipterans, then Dipteran testis-specific β-tubulin has evolved with essentially the same α-tubulin for 265 Myr. This provides a great amount of time to test different β-tubulin structures on the same α-tubulin substrate, thus increasing the chance that a "difficult to evolve" protein configuration, that is, the synergism found in Drosophila+Hirtodrosophilaβ2, would evolve in a Dipteran lineage. When the Drosophila+Hirtodrosophila β2 allele did appear approximately 60 Mya, it would also need to have a selective advantage to persist. Flies with long spermtails outcompete flies with shorter spermtails in Drosophila cage experiments, and given the reduced-length effects of mutations at amino acid sites 55 and 57, there is a basis for natural selection to promote the Drosophila+Hirtodrosophilaβ2 allele (Nielsen, 2006).

This potentially fundamental role of evolutionary history and opportunity in the evolvability of these insect axonemes reveals an important limitation on the power of natural selection to shape phenotypes. Rare events that release or result in functional constraint, rather than selection on allelic variation, may be a distinctive feature of macroevolutionary change in general, lending support to the Steven J. Gould notion that the tape of evolution would play differently each time it is rewound (Nielsen, 2006).

date revised:  25 November 2008  

The Interactive Fly resides on the
Society for Developmental Biology's Web server.