One of the two cytoplasmic actin isoforms in Drosophila is essential

Abstract

Actin is a highly conserved protein found in all eukaryotic organisms. Most organisms have multiple cytoplasmic actin genes that encode isoforms with slightly different amino acid sequences. These different isoforms are coexpressed in many cell types. Why organisms have multiple very similar cytoplasmic actin genes is unclear. We have addressed this question with the cytoplasmic actins in Drosophila, Act5C, and Act42A. These isoforms differ by only two amino acids and both genes are expressed in all cells at all times during development. We identified P element insertions in the Act5C gene that resulted in a lethal phenotype. The lethal phenotype is rescued by a transgene with a genomic fragment that includes Act5C regulatory and amino acid coding sequences. A hybrid transgene containing the protein coding sequence for the Act42A isoform, under the control of the regulatory regions of the Act5C gene, also rescues the lethality of the Act5C mutants. Furthermore, flies that carry only one copy each of Act5C and Act42A are viable. These results suggest the amino acid differences between these two cytoplasmic actin isoforms are not important for function and the need for increased gene dosage to provide more actin is not likely to explain the existence of multiple genes. Instead, our results suggest that regulated expression of Act5C is essential to the fly.

Actin is a key component of the cytoskeleton in all eukaryotic organisms. It is critical for cell movement, determination of cell shape and cell division, and it plays important roles in many other processes, including organelle transport. Actin is a highly conserved protein, with only a few amino acid sequence differences between species as evolutionarily distant as humans and slime molds. Multicellular organisms have several isoforms of cytoplasmic actin, which are coexpressed in most cell types and have very similar sequences to each other (1).

There are several possible reasons why organisms have multiple, highly similar, actin isoforms. Organisms need a large quantity of actin, and the best way to provide enough actin may be to have multiple genes. If this were the case, the amino acid sequence differences between the isoforms would likely have no functional consequences. A related explanation for multiple isoforms is that because some cells need more actin than others, multiple genes provide a mechanism for differential regulation of actin expression. In this case, the amino acid differences would not be functionally significant either, but the regulation of the expression of the genes would be critical. Finally, the small number of amino acid sequence differences may be functionally important, allowing different isoforms to have different roles in the same cell.

There are several lines of evidence that strongly suggest that different isoforms do indeed have specialized functions (1). Myogenic cells show regulated isoform-specific expression, switching from the expression of cytoplasmic to muscle isoforms. In quiescent aortic smooth muscle cells, an abundance of α-vascular smooth muscle actin isoform is produced; but when these cells begin to proliferate and migrate in culture, they express nonmuscle actins (2). Nonmuscle vertebrate actin isoforms, β and γ, cause different phenotypes when overexpressed in tissue culture cells. For example, overexpression of β-actin in C2 myoblasts leads to increased cell surface area and loss of stress fibers, whereas overexpression of γ-actin causes the cells to round up (3). It has also been demonstrated that the actin-binding protein ezrin interacts specifically with β- but not α-actin filaments (4). Although these experiments support the idea that the sequence differences among cytoplasmic actin isoforms are functionally important, there has been no definitive test of whether different cytoplasmic isoforms can functionally replace each other. In this work, we directly test whether the cytoplasmic actins in Drosophila are functionally unique.

In Drosophila melanogaster, two isoforms each of cytoplasmic actin, larval muscle actin, and adult muscle actin have been identified, all encoded by different genes (5, 6). Experiments by Fyrberg et al. (7) have shown that specific functions do exist for muscle-specific isoforms. Null mutations in the 88F gene, encoding an adult muscle isoform, result in a flightless fly. Flight can be restored by germ-line transformation of a wild-type copy of the 88F gene. A hybrid gene, containing the coding region for the 79B isoform (the other adult muscle-specific isoform), under the regulatory control of the 88F gene also rescues the flight defect. However, substituting other actin isoforms, including the two larval muscle-specific or one of the cytoplasmic isoforms for the 88F protein, does not restore flight to the 88F mutants. A similar experiment performed by Brault et al. (8) has shown that the human β-actin isoform does not fully compensate for the endogenous 88F isoform in restoring flight to 88F mutant animals. Although they report assembly of the human β-actin isoform into thin filaments, sarcomeric organization of the indirect flight muscle is defective.

Because muscle actin participates in a uniform and relatively static (as least from a polymerization standpoint) structure, it could be argued that the result that isoforms are functionally specialized in muscle is not generally applicable.

Cytoplasmic actins, in contrast to muscle actins, participate in many different structures and interact with a large number of different proteins to form these structures. Cytoplasmic actins might be thought of as more “multifunctional” than muscle isoforms, and thus need to be less specialized. However, whether individual cytoplasmic isoforms are specialized to perform subsets of the many functions required of actin remains unanswered.

In Drosophila, cytoplasmic actin is supplied by Act5C and Act42A (9, 10). There are only two amino acid differences between Act5C and Act42A (see Results). Each substitution is conservative in nature (A261S and I331V), and neither is located in a region of the actin molecule known to interact with actin-binding proteins (11). mRNAs for both Act5C and Act42A are maternally loaded and initially, transcripts for both mRNAs are evenly distributed throughout the early embryo (10). Zygotic expression of both genes occurs throughout embryonic development in all cells. The relative levels of transcripts for the two genes vary in particular tissues: Act5C mRNA levels are highest in the developing brain, whereas Act42A mRNA levels are most abundant in the developing gonad and portions of the intestine (10).

We used Drosophila as a genetically tractable model system to study the function of cytoplasmic actin. We asked whether the two cytoplasmic isoforms have different functions, and if so, were these specialized functions the result of amino acid sequence differences, the different expression patterns of the two genes, or a gene dosage effect. To differentiate between the three possibilities, we identified mutations in the Act5C gene. These mutations cause a lethal phenotype, suggesting that Act5C does have a unique function. We demonstrate that the lethal phenotype of the mutants can be rescued with a hybrid gene containing the protein-coding region of the Act42A gene under the control of the regulatory regions of the Act5C gene. This result establishes that the amino acid differences between the two isoforms are not functionally important. Manipulation of the gene dosage of both genes indicates that it is unlikely that multiple actin genes are required merely to provide enough actin globally. These results indicate that the pattern of expression of the Act5C protein is important for its essential function, and suggest that some tissues in which Act5C is expressed at high levels cannot develop or function normally with only the actin provided by Act42A.

Materials and Methods

P Element Mutagenesis to Isolate Act5C Mutations.

A local P element mobilization screen (12) was performed with fs(1)ph^8–29B flies that have a P lArB element inserted seven kb downstream of the start site of transcription of the Act5C gene (13). Fifty X chromosome lethals were isolated and four Act5C lethal alleles were identified. Those lethal mutations that were rescued by an Act5C transgene were subjected to inverse PCR to confirm the insertion of a P element near the gene.

Determination of Lethal Phase.

To determine when the Act5C mutants died, we placed each of the four mutant chromosomes in trans to an X chromosome balancer that carried a green fluorescent protein (GFP) marker. Each mutant line was allowed to lay eggs overnight, and mutant larvae were selected by lack of GFP fluorescence. The homozygous mutant larvae were transferred to new grape plates and examined each day.

Northern Analysis.

To determine the amount of residual Act5C transcript present in larvae, all four Act5C^P alleles were placed over a GFP X chromosome balancer, hemizygous mutant larvae were selected by lack of GFP fluorescence, and RNA was isolated (14) from these larvae.

To determine the amount of transcript produced by the different Act5C transgenes, the different transgenes were placed in the background of a lethal deficiency [y w l (1)Z76B] whose distal breakpoint deletes DNA containing the two most 3′ polyadenylation signals of Act5C. Males were recovered by introducing a P element, P[YES BC], which rescues the lethality caused by the Df but does not restore the missing Act5C mRNAs. These males had the following genotype: y w l (1)76B;P[Act5C];P[YES BC]. Adult males were collected and RNA-prepared.

In both cases, total RNA was electrophoresed, transferred to Nytran (Schleicher & Schuell), and the filter was hybridized to an Act5C-specific probe, pA1 (15), which recognizes all three size classes of Act5C transcripts. In addition, both filters were probed with a ³²P-labeled RP49 probe to ensure that equal amounts of RNA were loaded.

Construction of Transgenes and Generation of Transgenic Lines.

All transgenes were constructed with standard DNA molecular cloning techniques (16). In the case of the Act5C-42A transgene and the H264Q transgene, a 3.1-kb XbaI (isolated from φPCR-A) “cassette” was used as a template for in vitro mutagenesis reactions with the Chameleon kit (Stratagene) to change the appropriate amino acids present in each transgene. The changes were verified by sequencing with primers specific to the Act5C gene.

All transgene constructs were injected into w;[Δ2-3 ry⁺]Sb ry/TM3 flies (17), injected survivors were mated to yw67c23 flies, and transformants in the resulting progeny were identified by red eye color. Stable transformed lines were made by removing the [Δ2-3 ry⁺]Sb ry chromosome and using appropriate balancers for each transformed line.

Genomic DNA and Reverse Transcription (RT)-PCR Sequencing.

Genomic DNA or total RNA was isolated from either IsoA or OreR adult flies as described by Jowett (14). Genomic DNA was used as a template for amplifying the coding regions of either Act5C or Act42A genes. The RNA was used to generate cDNAs with an RT-PCR kit (Stratagene). These single-stranded cDNAs were the templates in PCR designed to amplify either Act5C or Act42A cDNAs by using primers specific for each gene. The amplified products were electrophoresed, gel purified with the Qiaquick extraction kit (Qiagen, Chatsworth, CA), and sequenced directly with gene-specific primers.

Rescue Analysis.

Males carrying the different transgenes were crossed to females of the genotype ywAct5C^P/FM6;+/+;+/+. Rescue was determined by the presence of non-Bar, red-eyed male progeny. All four Act5C P element mutations were tested for rescue with the different transgenes.

Gene Dosage Studies.

Females with the genotype ywAct5C^P* were mated to Df(2R)nap1/In(2LR)Gla Dp(2:2)BGwg^Gla-1 or Df(2R)nap9/In(2LR)Gla Dp(2:2)BGwg^Gla-1 males. Both are heterozygous for the Act42A gene. Each Act5C P element allele was mated to each deficiency. The number of ywAct5C^P*/+;+/ Df(2R)nap1 or ywAct5C^P*/+;+/Df(2R)nap9 female progeny was compared with the number of ywAct5C^P*/+;+/In(2LR)Gla Dp(2:2)BGwg^Gla-1 female progeny.

Results

P Element Insertions Upstream of the Act5C Transcription Start Site Are Lethal.

To obtain Act5C mutations that would permit us to examine the reasons that animals have multiple actin genes, we performed a local mobilization screen to obtain P element insertions within the Act5C gene. Our parental line, fs (1)ph^8–29B, was one in which a P element insertion resided ≈7.5 kb downstream of the transcription start site of the Act5C gene (Fig. 1). The presence of this P element, fs (1)ph^8–29B, causes a female sterile phenotype (13), but Act5C mRNA levels are unaffected (C.R.W., data not shown). We mobilized the parental P element and, because Act5C is on the X chromosome, we collected flies that had a lethal phenotype with respect to the X chromosome. Females heterozygous for the lethal chromosome in trans to an X chromosome balancer were collected, DNA was isolated, and genomic Southern analysis was performed to identify insertions in the Act5C region (data not shown). We identified four such lines, each with a P element insertion within a 200-bp region ≈3 kb upstream of the start site of transcription of the Act5C gene (Fig. 1). All four lines also carry a copy of the original P element insert, and one line has a total of three P element insertions (data not shown).

Figure 1.

Map of the Act5C region. Shown are the Act5C transcription unit, the location of P element insertions (original and Act5C), DNA included in transgene constructs, and an Act5C-specific probe, pA1. The 6.4-kb XbaI-XhoI DNA fragment was used as a probe for the Southern blots. All four Act5C P elements map within the ∼200-by EcoRI-BglII fragment.

All four Act5C P element mutant alleles die as first instar larvae. Shortly after hatching the mutant larvae are indistinguishable from wild type with regards to movement, but they soon appear noticeably sluggish and die within 4 days after egg hatching without growing substantially or molting.

Act5C mRNA Levels Are Affected by the Presence of the P Element Insertions Upstream of the Act5C Gene.

We collected hemizygous mutant larvae and measured Act5C mRNA abundance in each of the P element lines as compared with wild type. As shown by Northern analysis (Fig. 2), levels of Act5C mRNAs are severely reduced in all four of the P element insertion lines. Act5C mRNA is maternally loaded into the oocyte and we believe the small amount of Act5C mRNA seen in the P element insertion lines reflects residual maternally loaded Act5C mRNA.

Figure 2.

Northern blot analysis of Act5C mutants. Mutant alleles were placed in trans to a GFP X chromosome balancer, and fluorescent and nonfluorescent first instar larvae were collected separately. Total RNA was isolated from pools for each mutant line. To determine the amount of Act5C transcript, an Act5C-specific probe, pA1, was used. RP49 is a loading control and demonstrates approximately equal loading in all lanes.

The Lethal Phenotype of the P Element Insertions Can Be Rescued by Introducing a Wild-Type Copy of the Act5C Gene.

To demonstrate that the lethal phenotype of these P element insertions is the result of loss of Act5C function, we introduced a wild-type copy of the Act5C gene into the mutant background and assayed for rescue of the lethality.

Two different rescue constructs, 9.4EcoRI Act5C and 10.5XhoI Act5C (Fig. 1), were transformed into flies by means of P element-mediated transformation and stably transformed lines were obtained. We assayed for Act5C expression from the transgenes by crossing the transgenes into a line of flies that express only the 1.8-kb transcripts of Act5C (for details on the strain, see Materials and Methods). As shown in Fig. 3 (lanes 3–6), three independent lines carrying the 10.5XhoI Act5C transgene and one line carrying the 9.4EcoRI Act5C transgene express the 2.0-kb and to a lesser extent, the 2.3-kb transcripts of the Act5C gene. These transcripts are the result of Act5C transcription from the transgene.

Figure 3.

Northern blot analysis to determine whether Act5C transgenes are expressed. Transgenes were placed in a genetic background [y w l (1)76Z; YES BC] in which only the 1.8-kb message of Act5C is made (lane 2). Males with the correct genotype were collected and total RNA was isolated. The Act5C-specific probe, pA1, was used to detect expression of the transgenes by the presence of the 2.0-kb message. Genotypes of the flies used to isolate RNA were as follows: lane 1, wild type; lane 2, y w l(1)76Z; YES BC; lane 3, 9.4EcoRI Act5C (Act5C wild-type transgene); lane 4, 10.5XhoI Act5C-D (Act5C wild-type transgene); lane 5, 10.5XhoI Act5C-G (Act5C wild-type transgene); lane 6, 10.5XhoI Act5C-H (Act5C wild-type transgene); lane 7, Act5C-42A-D₂; lane 8, Act5C-42A-K₁; lane 9, Act5C-42A-JJ₁; lane 10, Act5C-42A-AA₃; lanes 11–13, three different BglII transgenes with a premature stop at amino acid 90 in the Act5C protein. RP49 was used as a control to demonstrate that the lanes were loaded similarly (data not shown). Note that the total amount of Act5C mRNA is unaffected by the deletion of the 3′ sequences (compare lanes 1 and 2).

We then tested the ability of these four transgene insertions to rescue the lethal phenotype of the four Act5C P element mutant alleles. All four Act5C transgene insertions rescued the lethal phenotype of the Act5C P element mutant alleles. Quantitative data for rescue efficiency of two mutant lines are shown in Table 1. If the wild-type transgene fully rescues the lethality of the mutants, we would expect the number of males carrying an X chromosome with the Act5C P element mutation should equal the number of males carrying nonmutant X chromosomes (balancer). We expected 50% of all males to be rescued (our values range from 46 to 64%). Despite the fact that the rescue efficiency was somewhat variable for different transgenes rescuing different mutant lines, the transgenes have strong rescuing ability. This rescue supports the conclusion that the mutant alleles are indeed Act5C mutations.

Table 1.

Rescue of two Act5C mutant alleles by using various transgenes

	ywact5C3447	ywact5C3627
yw;10.5XhoI Act5C-D	54% (188)	46% (348)
yw;9.4EcoRI Act5C	61% (141)	64% (173)
yw; Act5C-42A-D2	55% (200)	54% (368)
yw; Act5C-42A-AA3	0.03% (58)	0.018% (56)
yw;H264Q #9-3	63% (65)	47% (92)
yw;H264Q #16	53% (94)	34% (83)

The number shown is the percent of males carrying an Act5C mutant X chromosome out of total males, shown in parentheses. The transgenic lines were homozygous, therefore every male offspring should receive a copy of the transgene. 10.5XhoI Act5C-D and 9.4EcoRI Act5C are wild-type transgenes; Act5C-42A-D₂ (expressed) and Act5C-42A-AA₃ (not expressed, see text) are transgenes with Act5C control regions and Act42A protein sequences. H264Q #9-3 and H264Q #16 are transgenes with the H at 264 in φPCR-A corrected to Q. 

We also constructed a transgene, H264Q, in which we “corrected” the amino acid histidine at position 264 to glutamine. This change corrects an apparent cloning artifact in the Act5C φDNA clone isolated previously (see next section). As seen in Table 1, two independent transgene insertions (H264Q #9-3 and H264Q #6) rescue the Act5C mutants but not significantly better than the transgenes that encode histidine at position 264.

To be sure that expression of Act5C protein was required, the 10.5XhoI Act5C construct was modified to introduce a premature stop codon at amino acid 95 of the Act5C protein (ΔBglII transgene). Three stably transformed lines were obtained and were tested for their Act5C expression (Fig. 3, lanes 11–13). In contrast to our result with the 9.4EcoRI Act5C and 10.5XhoI Act5C transgenic lines, we do not see expression of the 2.0- or 2.3-kb transcripts in these lines. It is likely that because the construct introduces a premature stop codon, the transcript is unstable. The ΔBglII transgenes were tested for their ability to rescue the lethal phenotype of the Act5C P element mutant alleles and all three failed to rescue the lethal phenotype of the P element insertion lines.

Because this region of the genome is very tightly packed with ORFs, we wanted to be sure that the observed rescue was the result of Act5C expression and not expression of other ORFs on our transgenes. A transgene containing a 6.4-kb XhoI-XbaI fragment, including Act5C upstream regulatory regions and the first two Act5C exons, was constructed (Fig. 1). Six transgenic chromosomes carrying this construct were tested for their ability to rescue the Act5C P element mutant alleles and all six failed to rescue the lethal phenotype of the P element insertion lines.

A gene encoding a protein with homology to the Male-sterility 2 protein in Arabidopsis (18) lies just 3′ of Act5C (unpublished results). To exclude the possibility that the P element insertions affect this gene and are responsible for the lethal phenotype, a transgene containing a 3.3-kb SalI-BamHI fragment was constructed. This 3.3-kb SalI-BamHI transgene rescues flies with lethal mutations (generated in the screen in which the Act5C mutations were isolated) that fall into in a different complementation group. Five independent insertions of this transgene failed to rescue the Act5C P element mutant alleles. These data indicate that alterations in the gene just 3′ to Act5C are not responsible for the lethal defects in the Act5C P element insertion alleles.

In sum, the Act5C lethal P element insertion lines fail to accumulate Act5C mRNA, are rescued by transgenes carrying a functional Act5C gene, and are not rescued by other transgenes that include all of the sequences from the region but do not express Act5C. These data argue that we have identified mutations in the Act5C gene and that loss of Act5C function is lethal to the fly.

The Lethal Phenotype of the P Element Insertions Can Be Rescued by Introducing a Hybrid Act5C-42A Gene.

We wanted to determine whether the amino acid differences between the Act5C protein and the Act42A protein were essential for the Act5C protein function. We constructed a hybrid transgene that produces a protein identical to Act42A gene but under the control of the regulatory regions of the Act5C gene. We obtained four independent transformed lines carrying this hybrid construct. We assayed flies carrying these transgenes for transgene expression in the background where only a 1.8-kb transcript is seen from the endogenous Act5C gene (Fig. 3, lanes 7–10). In three of the lines we detected a 2.0-kb and to a lesser extent, a 2.3-kb transcript, which are characteristic of the Act5C gene that must be derived from the transgene. One line did not show the expression of these transcripts, indicating the transgene sequences were not expressed. We then tested the ability of these four independent transgene insertions to rescue the lethal phenotype of the four Act5C P element mutant alleles. The three transgene insertions that restored expression of the three Act5C transcript classes rescued the lethal phenotype of the Act5C P element mutant alleles. The one transgene insertion (Act5C-42A_AA3), in which the Act5C expression pattern was not restored, did not rescue the lethal phenotype (Table 1).

It was possible that, although the Act42A isoform rescued lethality to some extent, this isoform did not function as well as the Act5C isoform. To test this hypothesis, we counted the number of rescued animals with each independent insertion of each transgene. Although there was some variation among different lines in rescue efficiency, the rescue of lethality was quantitatively similar with transgenes carrying Act5C or Act42A coding sequences (Table 1). These data indicate that the amino acid differences between the two cytoplasmic isoforms in Drosophila are not functionally important.

The Act5C and Act42A Proteins Differ by Two Amino Acids.

In the process of sequencing the various transgenes we constructed, we noticed that the amino acid sequence in the clone (φPCR-A) we used in all our experiments differed from those published in the literature. Published sequence data reported differences of either five or seven amino acids between the two cytoplasmic actins (6, 7, 15). Because the actual isoform amino acid sequence differences were critical for our experiments, we resequenced relevant regions of the genomic DNA and cDNAs for both Act5C and Act42A from two Drosophila strains (OreR and IsoA). We detected no strain differences, and the sequences we obtained were identical for each gene whether we used genomic DNA or cDNAs. We observed only two amino acid differences between Act5C and Act42A: an alanine (5C) or serine (42A) at amino acid 261 and an isoleucine (5C) or valine (42A) at amino acid 331 (Fig. 4). In the φPCR-A clone (5C) there is a histidine at amino acid 264. This residue is a glutamine in the sequence from both genomic DNA and cDNA, suggesting the histidine in the cloned DNA (φPCR) resulted from a cloning artifact. A glutamine at amino acid 264 is present in all of the other actin isoforms in Drosophila. Our sequence data for the Act5C and Act42A genes are in complete agreement with the recent BDGP sequences for these two genes (19).

Figure 4.

Alignment of the Drosophila Act5C, Drosophila Act42A, human β, and human γ protein sequences. The consensus sequence is the Drosophila Act5C protein, and differences in amino acids as compared to Act5C are shown.

The Expression of Act5C Is Crucial to the Fly.

Because the Act5C-42A transgene rescued the lethal phenotype of the Act5C mutants, we know it is not the difference in amino acid sequences between the two cytoplasmic actin isoforms that is essential to the fly. However, two possibilities exist as to why viability is restored. The fly may need expression of Act5C in specific cells at specific times during development. Alternatively, it is possible that the total gene dosage of cytoplasmic actin is important. In the latter case, flies mutant for Act5C would die, because they have only two copies of cytoplasmic actin genes (two copies of Act42A) and the level of actin expression is simply below a critical threshold. The flies carrying the Act5C transgenes might live, because by supplying cytoplasmic actin from the transgene, a critical level of total cytoplasmic actin is once again present. To test for this sort of gene dosage effect, we constructed flies that had only one copy of Act5C and one copy of Act42A. We reasoned that if the regulated expression of the Act5C gene is essential to the fly, flies carrying one copy of each cytoplasmic actin gene should live, even though they have only two cytoplasmic actin genes. On the other hand, if a critical supply of cytoplasmic actin is required, we predicted that flies carrying only one copy of each cytoplasmic actin gene (i.e., two total copies, like the Act5C mutants) should not make enough actin and, therefore, die.

We performed a genetic cross that would allow us to obtain flies that carry one copy of Act5C and one copy of Act42A to test whether this combination was viable (see Materials and Methods for details). In all crosses to generate such animals, the number of flies doubly hemizygous for Act5C and Act42A was not significantly different from those only hemizygous for Act42A. Because we were able to obtain doubly hemizygous flies, reduced cytoplasmic actin gene dosage, per se, is not likely to be the reason that Act5C mutants die.

Discussion

We have generated mutations in one of the two cytoplasmic actin genes in Drosophila, Act5C. Mutations in this gene are lethal, indicating that this gene's function is not redundant with that of the other cytoplasmic isoform, Act42A, with which it coexpressed in all cell types. Our substitution of Act42A amino acid coding sequences for Act5C in the rescuing transgene conclusively demonstrates that amino acid sequence differences between the two isoforms are not functionally important. Rather, our results are most consistent with a requirement for Act5C gene expression in particular cells to supply enough actin for their normal function. In other words, it is the regulated temporal and spatial pattern of expression of the Act5C gene that is important for its unique role in flies.

It is known that mutations in the single actin gene in budding yeast are lethal (20), but this result is expected, because there is only one gene encoding actin in the yeast genome and actin is known to play many important roles in cells. However, multicellular organisms have multiple actin genes. Some of these genes are known to be specific to certain tissues, like muscle, and when mutated, cause defects in those tissues. However, in every organism where it has been studied, there are also several cytoplasmic actin isoforms and these isoforms are coexpressed in many different cell types of the organism. Because there have been no previous reports of cytoplasmic actin mutations in an animal, it has been unclear, until this work, whether mutations in one of the coexpressed genes would cause a phenotype. It seemed reasonable to expect that loss of the expression of one gene might have no detrimental effect because of the redundant expression of the remaining gene(s). The isolation of Act5C lethal mutations demonstrates that, at least in Drosophila, the cytoplasmic isoform genes are not redundant. However, our data that Act42A amino acid coding sequences expressed in the Act5C pattern rescue Act5C mutants suggest that it is the expression pattern of Act5C, rather than a unique function of the Act5C amino acid sequence, that is important.

Given the similarity of the Drosophila cytoplasmic isoforms, it is perhaps not surprising that Act42A coding sequences can replace Act5C coding sequences in rescuing the lethality of Act5C mutants and provide full function. We do not see any apparent morphological defects in Act5C mutants rescued with the hybrid transgene, although it is possible there are subtle phenotypes we have not yet detected.

By contrast, the two cytoplasmic actin isoforms in vertebrates β and γ, which differ by only four amino acids (Fig. 4), do not seem to be functionally identical. Overexpression of each isoform in tissue culture cells results in distinct changes to their cell morphology or cytoarchitecture (3). Further, biochemical interactions of the two isoforms with particular actin binding proteins are also different (4). These data have been interpreted to mean that the isoforms have distinct functions caused by the amino acid differences between them. N-terminal residues 1–4, where the differences between β and γ are found, participate in interactions with actin-binding proteins (11). Amino acid differences in this region are more likely to affect function than those seen in the Drosophila isoforms, which fall into regions that have not been shown to participate in interactions with ABP. Thus, vertebrate isoforms may be functionally specialized even though those in Drosophila are not. However, the methods of functional assay used in the vertebrate experiments are quite different from mutational/substitution experiments such as we have performed. The phenotype of mutants of the vertebrate cytoplasmic actin genes might prove helpful in understanding the true differences among highly similar isoforms.

In these experiments we have also tested whether increased gene dosage is an important reason that there are multiple actin genes in multicellular organisms. The viability of flies carrying only one copy each of Act5C and Act42A strongly suggests that this is not the case. Our data that actin gene dosage is not critical might be explained by the idea that actin protein levels are maintained through autoregulation of actin gene transcription (21). If the levels of monomeric pools of G-actin are high, actin mRNA is degraded. If monomeric pools of G-actin are low, actin mRNA synthesis is increased (21). We see evidence of such autoregulation in flies whose two most 3′ polyadenylation signals of Act5C are deleted. In these flies, only the 1.8-kb-size class of Act5C mRNA is produced and its level is comparable to the levels of the 2.0-kb-size class seen in wild-type flies (see Fig. 3, lanes 1 and 2). Unfortunately, we cannot determine how much of the total cytoplasmic actin is either Act5C or Act42A, because no isoform-specific Abs are available.

The requirement for flies to have at least one copy of Act5C to survive might be explained by differential mRNA stability. Act42A transcripts are turned over at a higher rate than Act5C transcripts (S. Tobin, personal communication), and therefore, the amount of transcript from each gene that accumulates is different. Achieving a sufficiently high level of accumulated transcript to provide enough actin in some tissues may be impossible with only Act42A transcripts. However, whether it is transcription pattern or message stability that is critical, our conclusion remains the same: there are at least some (and maybe all) tissues that require the expression of the Act5C gene.

One aspect of differential “regulation” that we have not tested in our experiments is the possible importance of the 3′ untranslated region (UTR). Some actin messages contain “addresses” or “zip codes” for localization within the cell, which are important for actin function (22, 23). It is possible that Act5C messages, through their 3′ UTRs, are targeted to a different subcellular location than are the Act42A messages and that this subcellular localization is critical. This phenomenon falls into the category of “regulated expression,” not isoform-specific functions or gene dosage effects.

These actin mutants establish that the two cytoplasmic actin isoforms in Drosophila are not redundant and provide us with the opportunity to study actin cytoskeletal function in an animal with many different cell types and actin structures with a genetic approach. In addition, further studies of actin isoform specificity with transgenes encoding muscle-specific isoforms and isoforms from evolutionarily distant organisms will reveal what sequence changes can be tolerated and still permit normal actin interactions and functions in the many different cell types in multicellular organisms.

Abbreviation

GFP

green fluorescent protein