Ovarian- and somatic-specific transcripts of the mosquito clathrin heavy chain gene generated by alternative 5'-exon splicing and polyadenylation.

Insect oocytes are extraordinarily specialized for receptor-mediated endocytosis of yolk protein precursors. The clathrin heavy chain (CHC) is the major structural protein of coated vesicles, the principal organelles of receptor-mediated endocytosis. To understand the role of clathrin in the development of the oocyte's powerful endocytotic machinery we determined the structure of the mosquito chc gene. The gene spans approximately 45 kilobases and its coding region is divided into seven exons, five of which encode the protein. Three distinct mature transcripts of this gene were identified in mosquito tissues. Two of them code isoforms of the CHC polypeptide differing in their NH2-terminal sequences, and are specifically expressed in female germ-line cells. The third transcript has a 3′-untranslated region about 1 kilobase longer than the other variants, and is found only in the somatic cells. Tissue-specific 5′-exon splicing and alternative polyadenylation of the primary transcript combine to give rise to these mRNAs. We identified two alternative promoters, distal and proximal, separated by approximately 10 kilobases involved in tissue-specific regulation of mosquito chc gene expression. Our data provide the first molecular evidence for complex structure and regulation of a chc gene, in this case occurring at both the transcriptional and post-transcriptional levels.

Insect oocytes are extraordinarily specialized for receptor-mediated endocytosis of yolk protein precursors. The clathrin heavy chain (CHC) is the major structural protein of coated vesicles, the principal organelles of receptor-mediated endocytosis. To understand the role of clathrin in the development of the oocyte's powerful endocytotic machinery we determined the structure of the mosquito chc gene. The gene spans approximately 45 kilobases and its coding region is divided into seven exons, five of which encode the protein. Three distinct mature transcripts of this gene were identified in mosquito tissues. Two of them code isoforms of the CHC polypeptide differing in their NH 2 -terminal sequences, and are specifically expressed in female germ-line cells. The third transcript has a 3-untranslated region about 1 kilobase longer than the other variants, and is found only in the somatic cells. Tissue-specific 5-exon splicing and alternative polyadenylation of the primary transcript combine to give rise to these mRNAs. We identified two alternative promoters, distal and proximal, separated by approximately 10 kilobases involved in tissuespecific regulation of mosquito chc gene expression. Our data provide the first molecular evidence for complex structure and regulation of a chc gene, in this case occurring at both the transcriptional and post-transcriptional levels.
Insect oocytes are highly specialized for receptor-mediated endocytosis of yolk protein precursors, and are characterized by a concentration of endocytotic organelles that is unparalleled in any other cell. The principal organelles of this endocytotic machinery are clathrin-coated pits and clathrin-coated vesicles. Three clathrin heavy chains (CHC) 1 combine with three clathrin light chains (CLC) to create hexameric triskelions (1)(2)(3), which associate with adaptor proteins or adaptins (4). Adaptins provide a link between the triskelion and the cyto-plasmic domain of target membrane receptors destined for internalization (3,5,6). During endocytosis, certain receptors for extracellular ligands are selectively concentrated in clathrin-coated pits. With the aid of the GTPase dynamin (7), a fully-invaginated coated pit is pinched off into the cytoplasm as a coated vesicle that contains the receptors and their ligands. In the cytoplasm, the clathrin coat disassembles rapidly and the uncoated vesicle fuses with its specific target compartment. Clathrin triskelions released from coated vesicles presumably recycle to the plasma membrane to undergo additional cycles of assembly and disassembly.
Clathrin is ubiquitously expressed by eukaryotic cells of all organisms analyzed so far, from Protozoa to Metazoa (8). Although unicellular organisms such as yeast (9,10) and Dictyostelium (11) can grow slowly and survive in the absence of clathrin, even the relatively simple cellular differentiation of these organisms require its function. Recent genetic studies in Drosophila have demonstrated that in multicellular organisms clathrin function is essential, and moreover, that there is a specialized role for clathrin in certain cell differentiation processes, like spermatogenesis (12).
Although coated vesicles were first identified in insect cells, viz. mosquito oocytes (13), detailed biochemical characterizations of clathrin (14) and molecular analyses of both the heavy and light chains have been undertaken mostly in mammalian (15)(16)(17) and yeast (9,10,18,19) systems. Sequence analyses have demonstrated a relatively high degree of identity (about 50%) among CHC polypeptides from such disparate species as rat and yeast, but only about 18% identity among CLCs. Mammals (rat, cattle, and human) express two types of CLC, which are coded by distinct single copy genes (16,17,20), whereas yeast and Amplysia cells express only one type of CLC (9,18,21). In addition, mammalian CLC genes undergo tissue-specific mRNA splicing (22). In contrast, only one single-copy gene for CHC has been identified in mammals (15,23,24), yeast (19), Dictyostelium (25), Caenorhabditis elegans (26), and Drosophila (12), and no splicing events of CHC mRNA have been described for any of these species. However, very recently, a second gene for CHC with differential expression in human sceletal muscle was identified (27,28), suggesting that the molecular heterogeneity of clathrin triskelions in at least some tissues could arise from a combination of CLC and CHC variants.
The intensive process of receptor-mediated endocytosis of yolk protein precursors in the developing oocytes of the yellow fever mosquito, Aedes aegypti, provides an excellent model system for analyzing the biogenesis of clathrin-associated transport machinery (29). Systematic characterization of the major components of this machinery at the molecular level is not only of general biological interest but also of practical significance, because the processes of egg maturation and dis-ease transmission are intimately tied in anautogenous mosquitoes through the requirement of a blood meal. Recently, one of the major components of this vesicular transport system, the mosquito vitellogenin receptor, was characterized (30,31) and its cDNA isolated and sequenced (32). This paper reports the molecular characterization of the gene coding another major component of coated vesicles, the CHC. Our results demonstrate that the A. aegypti chc gene is regulated at both transcriptional and post-transcriptional levels. Combinations of alternative 5Ј-exon splicing and alternative polyadenylation generate three distinct mature CHC transcripts, which exhibit tissue-, stage-, and sex-specific expression. Our data also suggest that two alternative promoter regions are involved in A. aegypti chc gene regulation resulting in two CHC isoforms in ovarian germ-line cells. In addition, whole mount in situ hybridization experiments with mosquito ovaries have revealed that the A. aegypti chc gene is expressed very early during previtellogenic development of the oocyte and that the ovarian CHC transcripts are found only in germ-line cells.

MATERIALS AND METHODS
Insects-Mosquitoes, A. aegypti, were maintained in laboratory culture as described elsewhere (33). Vitellogenesis was initiated 3-5 days after eclosion with a blood meal on rats.
Cloning and Characterization of Mosquito CHC cDNA and Genomic Sequences-A cDNA fragment of the A. aegypti CHC was first amplified from 20 g of ovarian total RNA by reverse transcriptase-polymerase chain reaction ( (12,15). To facilitate cloning of PCR products, an anchor sequence (italics) containing an XbaI restriction site (underlined) was added to the 5Ј-end of the primers. A 650-bp cDNA fragment was generated and subcloned into the pUC119 vector, and its identity verified by double-strand sequencing (34).
Seven independent cDNA clones and four partially overlapping genomic DNA clones, comprising the entire coding region of the A. aegypti chc gene, were subsequently isolated by hybridization screening of a ZAP II cDNA library generated from previtellogenic female mosquitoes, and of a Fix II genomic library prepared from adult mosquito whole bodies (35). In addition, two cDNA clones, representing the 5Ј-most end of CHC transcripts, were generated by the rapid amplification of cDNA ends (RACE) technique (36). For this, 1 g of poly(A) ϩ RNA isolated from vitellogenic ovaries 6 h post-blood meal (PBM) was reverse transcribed using random hexamer primers. The RNA/DNA hybrid was digested with RNase H, and the single stranded cDNA was tailed at the 3Ј-end with dGTP using terminal deoxynucleotidyl transferase. First strand cDNA was then amplified using a 5Ј-anchor primer (5Ј-AGAGAATTCCCCCCCCCCCCCC-3Ј) and a gene-specific reverse primer RP3 (5Ј-TGCGCATCGTTCATATCGATAATG-3Ј) designed from the genomic sequence of clone CHC N1. Two different PCR-generated fragments of 0.34 and 0.68 kb were obtained, subcloned, and sequenced.
The nucleotide and deduced amino acid sequences of the putative A. aegypti CHC clones were analyzed using the FASTA and GAP programs (University of Wisconsin, Genetics Computer Group software).
Primer Extension and S 1 Nuclease Mapping-Primer extension experiments were performed following the method described by Sambrook et al. (34). A 30-base oligonucleotide (5Ј-TTTTGACGTTTTGCTCTCAT-GAATGTCGAC-3Ј) was end-labeled with [␥-32 P]ATP (3000 Ci/mmol, DuPont NEN) by T4 polynucleotide kinase (Boehringer Mannheim) and used in primer extension reactions with 10 g of total ovarian RNA isolated from previtellogenic females and 10 g of tRNA (as a control). The products were fractionated on a DNA sequencing gel. Bacteriophage M13mp18 DNA was sequenced as a marker using M13 primer from a Sequenase version 2.0 kit (U.S. Biochemical Corp.). S 1 nuclease mapping was performed with a 522-bp genomic SacI-SalGI fragment derived from phage clone CHC4D. The 5Ј-end of the SalGI site of this fragment was dephosphorylated with calf intestine alkaline phosphatase (Boehringer Mannheim) and 32 P-labeled with T4 polynucleotide kinase. After hydridization of the probe with 10 g of total ovarian RNA and 10 g of tRNA (as a control), and subsequent digestion with S 1 nuclease as described by Sambrook et al. (34), the reaction products were processed in the same manner as in primer extension analyses.
Northern Hybridization and RT-PCR Analyses-Total RNA was extracted and purified from mosquito tissues by the guanidine isothyocyanate method as described previously (37). BioMag Oligo(dT) 20 beads (PerSeptive Diagnostics, Inc.) were used for poly(A) ϩ RNA isolation from total ovarian RNA according to the supplier's instructions. For Northern blot analyses, total RNA was fractionated in 1% agarose/ formaldehyde gels, blotted to a nitrocellulose membrane, and hybridized to 32 P-labeled DNA probes under high stringency conditions (34).
RT-PCR analyses were used to determine developmental expression profiles of two CHC isoforms in mosquito ovaries, as well as their tissue and sex specificity. For each time point tested, 10 g of total RNA was reverse-transcribed using random hexamer primers. The PCR reaction was performed under conditions previously described (38) using isoform-specific forward primers, FP1a (5Ј-AGCATGGCATCGCTTCTG AAGCTGACC-3Ј) and FP2b (5Ј-CATGTCGCAGGCTCTCCCGATTC GTTTCC-3Ј), and a common reverse primer RP3, the same as for the 5ЈRACE-PCR reaction. PCR products were separated by agarose gel electrophoresis, blotted to a nitrocellulose membrane, and hybridized to probes specific to the individual transcript being analyzed.
Whole Mount in Situ Hybridization-Whole mount in situ hybridization was performed according to the method of Tautz and Pfeifle (39) as described previously (32,40).

RESULTS
Structure of the A. aegypti chc Gene-To characterize the mosquito CHC, putative cDNA and genomic clones were isolated and analyzed. Using degenerate primers based on Drosophila and rat sequences (12,15), a 650-bp fragment of CHC cDNA was amplified from ovarian total RNA by RT-PCR, and subsequently subcloned and sequenced. The amino acid sequence deduced from this cDNA fragment exhibits identities of 85% to Drosophila, 75% to rat, 55% to Dictyostelium, and 46% to yeast CHCs, confirming that the isolated clone was from a mosquito CHC. Using this DNA fragment as a probe, several independent clones were isolated from a cDNA library of previtellogenic females, characterized by restriction analysis, and partially sequenced. Sequencing of both strands of the largest cDNA (3.9 kb) revealed that this clone contained a 2.5-kb protein-coding region and a 1.4-kb 3Ј-untranslated region (UTR). All isolated clones showed extensive congruence of DNA sequences in the protein coding region, but they fell into two classes differing in their 3Ј-UTR: one group exhibited a long 3Ј-UTR of 1.4 kb, whereas the other had a short 3Ј-UTR of 0.35 kb. Northern analyses revealed two types CHC transcripts in mosquito tissues (Fig. 1A): an "ovarian" 6.5-kb mRNA found only in the ovaries of previtellogenic and vitellogenic females, and a "somatic" 7.5-kb mRNA present in the somatic tissues of females (females minus ovaries, or carcasses) and whole body preparations of males. Significantly, when the unique portion of the long 3Ј-UTR (1-kb fragment) was used as a probe, hybridization was observed to the somatic mRNA only (Fig. 1B), indicating that the major difference between the two mRNAs was the length of the 3Ј-UTR.
Because the largest cDNA clone contained only the 3Ј-half of the CHC protein-coding region, a fragment from its 5Ј-end was used to screen cDNA libraries (constructed from previtellogenic females and vitellogenic ovaries) and a genomic library in an attempt to isolate the full CHC coding sequence. A 14-kb clone containing the beginning of the coding region of the CHC was isolated from the genomic library. This genomic clone, CHC1N, was characterized by restriction mapping, hybridization to cDNAs, and extensive sequence analysis (Fig. 2). Based on comparisons with other clathrin sequences, we estimated that it contained 99% of the CHC protein-coding region, including four exons (3, 4, 5, and 6; Fig. 2); the last of these also coded the first 148 bp of the 3Ј-UTR. However, sequences coding for the 5Ј-UTR, the very beginning of the translated region (ϳ14 aa), and the majority of the 3Ј-UTR were unrepresented in this clone.
It was not known how distant the remaining short proteincoding region at the 5Ј-end was from the beginning of exon 3, and sequencing over 1 kb into the intron failed to reveal it. The strategy we adopted was to determine the sequence of this region from 5Ј-RACE PCR-generated cDNA, which would allow us to design probes from its sequence to isolate genomic clones containing it. Suprisingly, sequencing of the RACE-PCR products revealed two classes of cDNA (0.34-and 0.68-kb) coding CHC isoforms, A. aegypti CHCa and A. aegypti CHCb, differing in their 5Ј-UTRs and NH 2 -terminal portions of the protein (Fig.  3). Probes designed from both isoform-specific sequences were used to screen the genomic library, and two positive phage clones, CHC4B and CHC4D, were isolated (Fig. 2). Restriction mapping and sequencing of restriction fragments hybridizing to probes from A. aegypti CHCa and A. aegypti CHCb demonstrated that the two clones overlap and that the two isoforms are coded by alternatively spliced exons: A. aegypti CHCa is produced by splicing exon 1a to exon 3, and A. aegypti CHCb is generated by splicing exons 1b, 2b, and 3 ( Fig. 2; Table  I). Exon 1a is 182 bp long, is located ϳ15 kb upstream of exon 3, and codes the 5Ј-UTR and the beginning of the NH 2 -terminal sequence of A. ageytpi CHCa. Exons 1b and 2b are separated by a small intron of 53 bp and are located ϳ5 kb upstream of exon 3. Exon 1b codes part of the 5Ј-UTR and exon 2b codes the remainder of the 5Ј-UTR and the NH 2 -terminal sequence of A. aegypti CHCb.
To isolate a clone containing the 3Ј-end of the gene, the unique 1-kb fragment of the long 3Ј-UTR cDNA was used as a probe. This screening successfully yielded the remainder of the 3Ј-UTR: the isolated genomic clone, CHC5-1, coded the last exon (exon 7; Fig. 2). Because the clones CHC5-1 and CHC1N did not overlap, the size of the intron separating Arrow indicates the position of the intron which separates tissue-specific alternative exons 1a (codes for A. aegypti CHCa) and 2b (codes for AaCHCb) from common exon 3. Amino acid residues that are conserved in all seven of the aligned sequences, or at least six of seven are in bold. exons 6 and 7 was gauged by Southern blot analysis of mosquito genomic DNA. The minimum size of this intron was estimated to be ϳ15 kb (data not shown). The structure of exon 7 was determined by sequencing a 7.1-kb genomic EcoRI-EcoRI fragment derived from CHC5-1 (Fig. 2). This genomic sequence was compared to the cDNA sequences of the long 1.4-kb and short 0.35-kb 3Ј-UTRs (Fig. 4). The genomic sequence from phage CHC5-1 was identical to the long cDNA sequences, and no splice variants were observed in this region.
The sequence of the 3Ј-UTR of the A. aaegypti chc gene (Fig.  4) is coded by the last 148 nucleotides of exon 6 and exon 7, about 1.3-kb. Several putative consensus sites, ATTTA, which are thought to increase instability of mRNA (41,42), are located in this region. Two of these sites were identified in the nucleotide sequence of the ovarian transcript (short 3Ј-UTR), and seven sites were found in the somatic transcript sequence (long 3Ј-UTR) (Fig. 4). A nonamer consensus sequence, TTATT-TA(T/A)(T/A), containing repeat number 5 of the somatic transcript sequence and the tandemly arranged repeats numbers 6 and 7 are two motifs recently shown to significantly increase mRNA instability in many cases (43).
Analysis of exon-intron organization of the A. aegypti chc gene (Table I) revealed that all splice sites conform to the GT/AG consensus (44): the introns start with a GT-dinucleotide and end with an AG-dinucleotide. Examination of intron phasing indicates that introns 1a, 1b, and 2 are positioned between codons, while all other introns which interrupt the proteincoding region (3, 4, and 5) are situated between the first and second nucleotide within the codon.
Transcription Start Site Mapping-Primer extension and S 1 nuclease protection experiments were conducted to determine the transcription start site of the ovarian mRNA-coded isoform, A. aegypti CHCb. A 30-base oligonucleotide primer, complementary to the sequence of putative exon 1b (position 91-121 of the A. aegypti CHCb cDNA generated by 5Ј-RACE PCR) (Gen-Bank accession number GSDB:S:1076487), was hybridized to total previtellogenic ovarian RNA and extended using Superscript II reverse transcriptase. This resulted in a 552-bp extension product (Fig. 5A), placing the transcription start site 432 bp upstream of the available cDNA.
We performed S 1 nuclease protection experiments to confirm that this position is the genuine start site of chc gene transcription, and that the extension product was not prematurely terminated due to mRNA secondary structure. When a 522nucleotide DNA probe, corresponding to a genomic fragment containing most of putative exon 1b and extending up-stream, was hybridized to total ovarian RNA, one strong S 1resistant product of 502 bp was detected (Fig. 5B). Thus, the size of exon 1b was calculated to be 692 bp, and the inferred start site position corresponds well with that predicted from the primer extension experiments.
Tissue, Sex, and Stage Specificity of Alternative Exon Splicing-Tissue specificity of alternative 5Ј-end exon splicing, and the possible correspondence of the A. aegypti CHCa and A. aegypti CHCb isoforms to the ovarian (6.5-kb) and somatic (7.5-kb) transcripts were examined by RT-PCR of RNA fractions from ovaries of previtellogenic and vitellogenic females,  female carcasses, males, and early embryos. Three PCR primers were used: the reverse primer RP3 (containing a sequence from exon 3 common to both the A. aegypti CHCa and A. aegypti CHCb isoforms), and two forward primers, FP1a (specific to the A. aegypti CHCa isoform, derived from the exon 1a sequence) and FP2b (specific to the A. aegypti CHCb isoform, derived from the exon 2b sequence). PCR amplification using exon 1a-and 3-specific primers produced an intense band of the expected size (203 bp) from ovaries of previtellogenic females (1 and 4 days after eclosion) and early vitellogenic females (6 h PBM), and a less intense band from ovaries of later vitellogenic females (18 h PBM) (Fig. 6). No bands were observed using RNA from ovaries of late vitellogenic females (24 and 48 h PBM), female carcasses, males, or early embryos (Fig. 6). In contrast, PCR amplification using primers specific to exons 2b and 3 yielded intense bands of the expected size (177 bp) for the ovaries of previtellogenic and vitellogenic females, as well as for female carcasses, males, and early embryos (Fig. 6).
Localization of CHC Transcripts in the Ovary-We examined the spatial distribution of the AaCHC transcripts in the ovary by whole mount in situ hybridization. The hybridization signal was observed in the oocytes of developing primary follicles during early previtellogenic stages and later in their nurse cells (Fig. 7). There was no apparent accumulation in the somatic follicle cells. Interestingly, A. aegypti CHC RNA could be detected even in oocytes of the undifferentiated ovary, 0 -1 day after eclosion (Fig. 7A). It dramatically increased in abundance by 3-4 days after eclosion (Fig. 7B).
During the vitellogenic period, A. aegypti CHC mRNA was present in both oocytes and nurse cells from 3 to 12 h PBM (Fig.  7C) but virtually disappeared from the primary follicles by 24 h PBM. It is noteworthy that, at midvitellogenesis, the accumulation of A. aegypti CHC transcripts could also be clearly detected in the oocytes of secondary follicles (not shown). Significantly, no hybridization was observed to a probe derived from the long 3Ј-UTR, which specifically recognizes the 7.5-kb somatic transcript (not shown), suggesting that only the 6.5-kb ovarian transcript is synthesized in germ-line derived cells during pre-and vitellogenic stages. Finally, control (sense) DNA probes never produced hybridization signals (not shown), confirming the specificity of the in situ hybridization technique used in this study. DISCUSSION The A. aegypti chc gene spans ϳ45 kb and is composed of 7 exons, five of which encode the protein (Fig. 2). The proteincoding sequence of the A. aegypti chc gene is highly conserved evolutionarily, at both the nucleotide and amino acid levels. Although the protein-coding sequences of CHCs are well characterized in several organisms, until now no detailed molecular characterization of a chc gene has been reported. Hybridization analyses demonstrated that the Drosophila CHC transcription unit is not less than 10-kb; and P element-mediated transformation has shown that a 13-kb genomic DNA fragment, including the transcription unit, is sufficient to rescue lethal CHC mutations (12). Thus, the length of the Drosophila CHC locus should be approximately three times less than that of the mosquito, consistent with the 5-fold larger size of the A. aegypti genome (45).
Two CHC isoforms, A. aegypti CHCa and A. aegypti CHCb, were identified when 5Ј-RACE PCR products were sequenced (Fig. 3), and analyses of overlapping genomic clones revealed that these isoforms are coded by distinct transcripts generated as a result of alternative splicing of exons at the 5Ј-end of the gene (Fig. 2). The transcription start site was identified for the A. aegypti CHCb mRNA (Fig. 5), but has not yet been determined for the A. aegypti CHCa transcript. Preliminary experiments suggest that there may be another exon upstream of exon 1a containing an untranslated sequence that has yet to be isolated. Nevertheless, it is clear that transcription of the two A. aegypti CHC mRNA variants is regulated by two alternative promoters, which we designate distal (for A. aegypti CHCa) and proximal (for A. aegypti CHCb).
Employment of isoform-specific primers in RT-PCR showed that A. aegypti CHCa transcripts are present in ovarian tissue only, but that A. aegypti CHCb transcripts can also be detected in somatic tissues of females and males (Fig. 6). These results strongly suggest that there is tissue-and sex-specific usage of exon 1a, but not of exons 1b and 2b.
Differential stage-specific expression is characteristic of the splice variants as well. Both the A. aegypti CHCa transcript, regulated by a distal promoter, and the A. aegypti CHCb transcript, regulated by a proximal promoter, exhibit similar kinetics, peaking in the late previtellogenic and early vitellogenic stage (Fig. 6). However, the distal ovarian transcript (A. aegypti CHCa) is expressed only before 24 h PBM (peak of vitellogenesis), and is not detectable later. In contrast, the proximal ovarian transcript (A. aegypti CHCb) is present throughout the entire vitellogenic period, persisting until the later stages of egg development and even in newly-laid eggs (Fig. 6). We suggest that the distal ovarian transcript is required for the receptor-mediated machinery of developing oocytes; it is uniquely germ-line specific and is expressed prior to or during yolk protein endocytosis, disappearing at the time of uptake termination. The proximal ovarian transcript with its persistent presence in fully-developed eggs indicates that this mRNA may be produced to meet the future needs of the mosquito embryo.
In addition to the two types of transcripts generated through alternative splicing, two tissue-specific size classes of A. aegypti CHC transcripts (6.5-kb ovarian and 7.5-kb somatic) were detected by Northern hybridization experiments (Fig. 1). Sequencing of cDNA clones revealed that the ϳ1-kb difference in size is accounted for by a corresponding difference in the lengths of the 3Ј-UTRs. Comparison of the genomic sequence of this region with those of the cDNA clones (Fig. 4) indicated that the size variants do not arise through alternative splicing of exons in the 3Ј-region. Instead, alternative polyadenylation is the probable mechanism generating the observed size variants of A. aegypti CHC mRNAs (Fig. 1). Two putative alternative polyadenylation signals were identified at positions 5357 and 6426 (Fig. 4). It is thus likely that alternative usage of these polyadenylation signals results in expression of two different transcripts in somatic and female reproductive tissues.
The deduced alternative splicing and transcription termination events occurring in the transcripts derived from the A. aegypti chc gene are presented in Fig. 8. The dual mechanisms of alternative splicing in the 5Ј-region and alternative polyadenylation in the 3Ј-region combine to produce at least three types of mature CHC transcripts in mosquito tissues, which we designate A. aegypti CHCa-O (ovarian), A. aegypti CHCb-O (ovarian), and A. aegypti CHCb-S (somatic) (Fig. 8). Thus, the A. aegypti chc gene is apparently regulated at both the tran- scriptional and post-transcriptional levels in a tissue-, sex-, and stage-specific manner.
In contrast to A. aegypti CHCa-O and A. aegypti CHCb-O, the A. aegypti CHCb-S transcript is found only in female somatic tissues (Fig. 1). Both A. aegypti CHCb-O and A. aegypti CHCb-S are regulated by the proximal promoter ( Figs. 2 and 8), but alternative polyadenylation results in unique sequences in the 3Ј-UTR. Eukaryotic gene expression is partly controlled by rate of mRNA degradation, which is generally a function of regulatory sequences in the 3Ј-UTR (46). The longer 3Ј-UTR of the A. aegypti CHCb-S transcript contains more and more potent consensus sequences (42,43) that promote mRNA decay than does the shorter 3Ј-UTR of the A. aegypti CHCa-O and AaCHCb-O transcripts (Fig. 4). These data suggest that the ovarian A. aegypti CHC transcripts are more stable than the somatic one. Such stability would facilitate accumulation of A. aegypti CHC mRNA in developing oocytes, permitting largescale production of clathrin protein during vitellogenesis when endocytosis of massive amounts of vitellogenin is occurring. Ovarian AaCHC transcripts with short 3Ј-UTR begin to accumulate very early in oocyte differentiation (Fig. 7A) and resistance to decay would be of obvious advantage.
Clathrin is widely distributed in both germ-line and somatic cells of eukaryotes, playing an extremely important role in the universally critical processes of receptor-mediated endocytosis and secretion. It is likely that clathrin exhibits specialized functions in various tissues, implying a need for specific regulation of chc gene expression during the development of multicellular organisms. Genetic studies in yeast and Drosophila demonstrated that different functions of clathrin are genetically distinguishable (12,19). Yeast CHC with a carboxylterminal deletion can rescue viability defects of null mutations but fails to complement defects in a-factor processing (19). Three of four lethal mutations of the CHC locus in Drosophila block development late in embryogenesis probably affecting the neuromotor function (12). Individuals with the fourth allele occasionally survive to adulthood but are characterized by invariable male sterility, pointing to a specialized role for clathrin in spermatogenesis.
The results presented in this paper provide the direct molecular evidence for complex regulation of a chc gene in higher organisms. Two promoter regions have been identified in the A. aegypti chc gene. Alternative usage of these promoters clearly has tissue-, sex-, and stage-specific characteristics. One would expect that mutations associated with these sequences will effect different functions and lead to various phenotypes. In addition, alternative polyadenylation has been observed at the 3Ј-end of the A. aegypti CHC primary transcript, which may differentially effect the stability, and perhaps translation efficiency, of A. aegypti CHC mRNA in germ-line and somatic cells. The combination of all these complex genetic processes provides a putative basis from which a diversity of CHC transcripts is derived, a diversity which may be essential for specialized functions of clathrin in eukaryotic cells.