Drosophila has several genes for gap junction proteins

Drosophila has several genes for gap junction proteins

Gene 232 (1999) 191–201 www.elsevier.com/locate/gene Drosophila has several genes for gap junction proteins Kathryn D. Curtin *, Zhan Zhang, Robert ...

718KB Sizes 0 Downloads 39 Views

Gene 232 (1999) 191–201


Drosophila has several genes for gap junction proteins Kathryn D. Curtin *, Zhan Zhang, Robert J. Wyman Department of Molecular, Cellular and Developmental Biology, Yale University, 266 Whitney Avenue, New Haven, CT 06511, USA Received 20 January 1999; accepted 22 March 1999; Received by E. Boncinelli

Abstract The Innexin gene family forms gap junctions in invertebrates. Many genes in this family have been identified in Caenorhabditis elegans, but only two in Drosophila. We have used PCR techniques to identify three new members of this family from Drosophila. These are designated pas-related proteins ( prp) 6, 7, and 33. The putative proteins coded by these new genes show 25–35% identity and 39–66% similarity to other Drosophila innexins and share a similar hydrophobicity profile. The genes form two small clusters on the X-chromosome, with three of the genes sitting within 10 kb of each other. The closeness in sequence and location suggests an evolutionary origin of these genes via local duplication. In situ hybridization shows expression in the CNS, gut and epidermis. Each gene has a distinct pattern of expression in different tissues at different developmental times. However, parts of the expression patterns overlap, especially for prp33 and ogre which may be expressed from the same transcriptional enhancers. This suggest that the Prp33 and Ogre proteins may join in forming heteromeric gap junction channels. © 1999 Elsevier Science B.V. All rights reserved. Keywords: pas-related protein; passover; shakingB; Innexin; Gene family

1. Introduction Innexins are a new gene family in invertebrates which form gap-junction connections (Phelan et al., 1998). The first innexin gene to be studied was passover in Drosophila ( Thomas, 1980; synonym: shakB2, Homyk et al., 1980). Early studies on passover mutants suggested defects at specific electrical synapses ( Thomas and Wyman, 1984). Subsequent cloning of pas ( Krishnan et al., 1993) led to the identification of a new protein family that included two other newly cloned genes, ogre (in flies; Watanabe and Kankel, 1992) and unc-7 (in worms; Starich et al., 1993). Barnes (1994) suggested structural similarities between this family and the connexins, the structural molecules for vertebrate gap junctions. Concurrently, work on unc-7 also suggested that it is involved in gap-junction formation. eat-5, another worm family member with defects in gap-junction function, was cloned shortly thereafter (Starich et al., 1996). Recently, these genes have been confirmed as the structural components of gap junctions by their ability to form de novo junctions in a Xenopus oocyte expression system (Phelan et al., 1998). Abbreviations: aa, amino acid; prp, pas related proteins; GF, giant fiber; GFS, giant fiber system. * Corresponding author. Tel.: +1 203-432-8925; fax: +1 203-432-6161. E-mail address: [email protected] ( K.D. Curtin)

The innexin family is now a large gene family with an additional 24 members identified from the Caenorhabditis elegans sequencing project. There is a high degree of homology among family members, even from species as far apart evolutionarily as Drosophila and C. elegans. For example, Unc-7 is 33% identical to Pas with an additional 15% conserved residues. However, no innexin homologs have been found in vertebrates. The family name innexin is short for invertebrate-connexin (Phelan et al., 1998). Passover ( pas) was originally identified in screens for Drosophila mutants which were unable to fly correctly ( Homyk et al., 1980) or failed to exhibit an escape response ( Thomas, 1980; Thomas and Wyman, 1984). The escape response is elicited by a lights-off stimulus and consists of a jump followed by flight. The neural circuit for the escape response involves the cells of the giant fiber system (GFS). Starting in the brain, the giant fiber (GF ) travels to the thorax where it forms electrical junctions (gap junctions) with the motoneuron to the jump muscle and an interneuron which drives the five flight motoneurons. The presence of gap junctions between the GF and its targets has been demonstrated by electrophysiology ( Thomas and Wyman, 1984) and dye filling ( Wyman and Thomas, 1983; Phelan et al., 1996; Sun and Wyman, 1996). In the adult CNS, pasN (for pas-neural, a specific pas transcript) is expressed in the GFS ( Krishnan et al., 1993; Crompton et al., 1995).

0378-1119/99/$ – see front matter © 1999 Elsevier Science B.V. All rights reserved. PII: S0 3 7 8 -1 1 1 9 ( 9 9 ) 0 0 12 3 - 7


K.D. Curtin et al. / Gene 232 (1999) 191–201

In pasN mutants, the gap junctions between the GF and its post-synaptic partners are missing, as shown by the lack of fast synaptic transmission ( Wyman and Thomas, 1983; Thomas and Wyman, 1984) and by the lack of dye passage (Sun and Wyman, 1996; Phelan et al., 1996). Dye transfer does occur between other cells, however, suggesting that many gap junctions in the animal are still intact (Sun and Wyman, 1996). Eat-5 mutations (C. elegans) disrupt pharyngeal pumping (Avery, 1993). In normal worms, the 15 cells of the corpus and bulb of the pharynx are connected with gap junctions and contract simultaneously. In eat5 mutants, excitation cannot pass between the corpus and bulb. While the corpus and terminal bulb each contract normally, they are no longer synchronized (Avery, 1993; Starich et al., 1996). In wild-type animals, dye introduced into the bulb passes through junctions between the cells of the bulb and into the cells of the isthmus and corpus. In eat-5 animals, dye introduced into the bulb still passes between the cells of the bulb and isthmus, but does not pass into the corpus (Starich et al., 1996). As with pasN, only specific gap junctions are eliminated. Unc-7 mutants (C. elegans) are defective in forward locomotion; instead of moving in a sinuous pattern, the body forms into irregular kinks. Mosaic analysis indicates a neural focus. Another family member, unc-9 (Barnes and Hekimi, 1997), shares all the physiological phenotypes with unc-7. In addition, in unc-7 mutants, EM reconstruction shows that the AVA interneurons, which normally form gap-junction synapses only with motoneurons for backward locomotion, now make these synapses with motoneurons for forward locomotion (Starich et al., 1993). Other innexin mutants do not show obvious disruption of specific gap junctions, but are lethal or cause severe defects. These include mutants of pasV and ogre in Drosophila. pasV is a second transcript made from the pas locus; pasN and pasV contain alternate 5∞ exons (Crompton et al., 1995; Krishnan et al., 1995) which code for a different amino termini. The two transcripts are expressed from different promoters, as is obvious from their non-overlapping expression patterns (Crompton et al., 1995). Mutations in pasV are lethal, with the animals dying as late embryos. The other described Drosophila innexin gene is ogre (optic ganglia reduced). ogre is expressed widely in several cell types ( Watanabe and Kankel, 1992) with mutations apparently leading to death of post-embryonic neuroblasts. Individuals bearing lethal alleles die primarily as pupae, while a viable hypomorphic allele results in severely reduced optic ganglia. Although the mutant phenotype of pasV has not been implicated in formation of specific gap junctions, the PasV protein forms de novo gap junctions in the paired Xenopus oocyte system (Phelan et al., 1998). This system

has been used extensively to study the vertebrate gapjunction genes called connexins (Paul, 1986; Swenson et al., 1989). PasV gap junctions (Phelan et al., 1998) have similar physiological properties to junctions formed by connexin43 (a vertebrate gap-junction gene; Swenson et al., 1989). Interestingly, the highly related PasN, which disrupts specific gap junctions in the animal, does not form junctions in the Xenopus oocyte (Phelan et al., 1998). PasN and PasV are 88% identical with an additional 3% conserved residues, with all the differences being in the first third of the protein. Attempts to express Eat-5 or Ogre in Xenopus oocytes have also not led to junction formation (Paul, Curtin, Goodenough, unpublished ). To date, several family members have been tested in this system, and some form junctions while many do not. Innexins are an invertebrate specific family of gapjunction genes: innexins share no sequence homology with connexins, the vertebrate gap junctions genes (reviews: Bennett et al., 1991; Goodenough et al., 1996; Kumar and Gilula, 1996). However, the predicted secondary structure is similar, with four putative transmembrane domains and internal N and C termini (Barnes and Hekimi, 1997; Crompton et al., 1995; Starich et al., 1996) leading to two external loops and one internal loop. Both families also have conserved extracellular cysteines, required in connexins for structural integrity ( Foote et al., 1998). And both families have conserved prolines in the second transmembrane domain. This proline is important for gating in connexins (Suchnya et al., 1993). At least 13 different connexin family members are known from rodents, and homologs for many have been identified in other vertebrate species. Despite attempts to find connexins in invertebrates, none have been found in Drosophila or C. elegans, though the genome sequencing of the latter is essentially completed. In addition, no vertebrate innexin homolog has been found in the data base. Thus it seems that this basic cellular structure, the gap junction, is expressed by two different gene families in vertebrates and invertebrates. Though there are 25 new family members identified from the C. elegans genome project, to date only two innexin genes have been reported in Drosophila, pas and ogre ( Krishnan et al. 1993, 1995; Crompton et al., 1995; Watanabe and Kankel, 1990.) Believing that there must be others, we carried out experiments to identify new Drosophila innexins. In this paper, three new Drosophila innexin genes are sequenced, mapped and their expression patterns defined. 2. Materials and methods 2.1. PCR protocol for identifying new innexin genes Degenerate primers were designed that corresponded to the amino-acid sequence NEK as the 5∞ endpoint of

K.D. Curtin et al. / Gene 232 (1999) 191–201

the antisense primer and to WFW as the 3∞ endpoint of the sense primer. The sequence of these degenerate primers is as follows (R=A or G, W=A or T, and Y= C or T ). The sense primers: TGC GAATTC ACG CTG AAC ATG TWY AAY GAR AA. This describes six different primer sequences, each 31 nucleotides long. The antisense primers: T GAGTCT AGA GCC AGC ATG AAR WAC CAR AAC CA. This describes four different primer sequences, each 33 nucleotides long. These primers contain sites for directional cloning, XbaI for the sense probe, and EcoRI for the antisense probe. PCR was carried out using 500 ng of genomic DNA as template and approx. 1.25 mM as the final primer concentration. The PCR protocol was as follows: 95°C for 5∞; then 5 cycles of (95°C for 30 s, 30°C for 30 s and 72°C for 30 s) followed by 30 cycles where the hybridization temperature was increased to 56°C with no other changes, and finally 72°C for 5 min. The resulting products were run on a 3% agarose gel and products of approx. 80 bp. were purified from the gel. These products were further amplified using nondegenerate probes: the sense primers: TGC GAATTC ACG CTG AAC ATG; the antisense primers: T GAGTCT AGA GCC AGC ATG AA. The resulting products were directionally cloned into the mp18 plasmid and the clones sequenced using standard plasmid sequence primers. The sequence was translated in all three reading frames and products that showed homology to innexins between the primers were used to screen a genomic library. Probes for screening were prepared by doing five to six rounds of annealing and extension using the non-degenerate sense and antisense primers and Taq polymerase. Southern hybridization was done essentially as described in Sambrook et al. (1989) using a hybridization buffer of 6×SSPE, 5×Denhardt’s, 0.5% SDS, and 0.5 mg/ml salmon sperm DNA. The temperature for hybridization was reduced from 65°C to 52°C, and washing was done at that temperature for 30 min. Genomic clones were used to screen a Drosophila head cDNA library. The resulting clones were cloned into Bluescript and sequenced. Subsequently, the genomic sequence was obtained.

2.2. Chromosomal in situs Chromosomes squashes and hybridization were done as in Zucker et al. (1985). Biotinylated probe was synthesized using Biotin High Prime (BMB) according to the manufacturer’s instructions. Signal was amplified using Vector Lab’s ABC kit according to the manufacturer’s instructions.


2.3. Embryo and adult CNS in situs Embryo in situs were done as in Doe et al. (1991). In situ of adult CNS was done as in Krishnan et al. (1993).

3. Results 3.1. Cloning of new innexin genes Because several gene segments are highly conserved among the various innexin genes (Fig. 2), a PCR strategy was designed for cloning new family members. Genomic DNA was used as a PCR template rather than reverse transcribed message so that all genes would be represented regardless of abundance, time or place of expression. However, Pas, ogre, and unc-7 ( Krishnan et al., 1993, 1995; Watanabe and Kankel, 1990; Starich et al., 1996) all contain large introns (though these are mapped only in Pas: Krishnan et al., 1993, 1995). Genomic sequences containing large introns are difficult to amplify by PCR. Primers were thus selected to correspond to highly conserved residues that were close enough to reduce the possibility that a large intron would interrupt the gene between the primers. The most highly conserved sequences which are close together are NEK and WFW. Degenerate primers were made with codons for these residues as endpoints. There are two disadvantages to choosing these residues as PCR targets. First, the target for the PCR primers is extremely short. Making longer primers adds little specificity because the homology just outside of NEK and WFW is quite low. For this reason, the initial cycles of PCR were done at quite low temperatures to allow these short sequences to bind. Second, there are only five amino acids between the endpoints of the primers (between NEK and WFW ) residues in the innexin genes ( Fig. 2). Although these are fairly well conserved, the small number of residues makes recognition of new innexin gene candidates among the sequenced PCR products somewhat difficult. The advantages to choosing these residues as targets are also twofold. First, they are almost universally conserved within the innexin family. Second the degeneracy for both primers was very low (see Methods). The first round of PCR was carried out using primers which contained restriction sites for subsequent cloning at their 5∞ ends. The 5∞ primer contained an EcoRI site followed by codons for the residues TLNM. L and N are conserved in all innexin family members, while T and M appear in ogre only. This sequence was followed by degenerate sequences corresponding to F/YNE and the first two bases for K (the primer did not end with a degenerate position). The 3∞ primer contained an XbaI site followed by codons for MFF, the last F being conserved in family members. This primer ends with


K.D. Curtin et al. / Gene 232 (1999) 191–201

degenerate sequences corresponding to WFW. The stretch to which the primers were designed is marked in Figs. 1 and 2 and the exact primer sequences are specified in the Methods section. Using these primers, products were generated by amplifying for a few cycles at low temperature followed by several cycles at higher temperatures (see Methods). Because the primer sequences providing specificity were very short and some amplification steps were carried out at low temperature, we were uncertain as to whether most of the products would be from bona fide members of the innexin family. To increase the chances of selecting innexin genes for sequencing, we selected products from the first PCR round that were of an appropriate size for further amplification. The spacing between NEK and WFW is strictly conserved at 5 aa in the known innexin family sequences. Thus products of 81 bp would be expected from members of this gene family using our primers. PCR products of approx. 80 bp were gel isolated from a 3% agarose gel for further amplification using primers that were identical to the first primer set, except that the degenerate sequences coding for F/YNEK and WFW were deleted. These non-degenerate primers would amplify all products of the first round. The resulting products were directionally cloned into a sequencing vector and sequenced. 48 products were sequenced. Several of these had the same nucleotide sequence as ogre; none had the same sequence as Pas. The products were translated in all three reading frames. The amino-acid sequences of three of the products were closely related to, but not identical with, Pas. These short products, about 80 bp long, were used as probes to screen a genomic library ultimately leading to the cloning of three new family members. Screening was done first using a genomic library, because the genes were almost certain to be represented in a genomic library. Given the difficulties in screening discussed below, screening one genomic library, rather than several cDNA libraries, was particularly desirable. Because of the short size of the PCR products, we used a non-standard strategy for labeling the probes and screening the library. First, the short size makes it difficult to make adequate length probes by standard random-primed labeling. To insure full-length probes, we used the same non-degenerate primers described above as primers for a labeling reaction. 32P dATP was used as label because the products have a lot of A/T pairs. In addition, several sequential rounds of labeling were done using Taq polymerase to make the probes as hot as possible. Second, of the 80 bp probe, at most 33 bp were sure to be found in the gene itself; the remaining residues are specified by the non-degenerate portion of the primers and may not (and generally did not) appear in the actual gene. For this reason, library screening was carried out at much lower than normal

temperatures (52°C compared to a standard 65°C for the buffer used) to allow the probes to bind. In this way, genomic clones in lambda phage were obtained for all three PCR products. Southern blots of the phage DNA, using the 80 bp PCR products as probe, allowed identification of restriction sites that produced bands that were of a size large enough that they might contain the whole coding sequence of the gene. Genomic clones were subcloned using these enzymes by shotgun cloning into Bluescript and screening colonies using the same PCR probes. These subclones were then used to screen an adult head cDNA library (Pas is expressed in the adult head, Krishnan et al., 1993; Crompton et al., 1995). Full-length cDNAs were found for two of the genes; these were named pas related proteins ( prp) 7 and 33. These cDNAs were sequenced. The cDNA and protein sequences are shown in Fig. 1. No cDNAs were found for the third sequence, prp6, though several libraries from different developmental stages were screened. The available coding sequence was obtained from sequencing a genomic clone. The sequence of this clone indicates that it starts with a large stretch of non-coding intron sequences. The clone picks up the gene in the middle of its coding sequence and continues to the end of the gene without further intronic interruption. Sequencing genomic clones for prp7 and prp33, revealed no introns in prp33 and two small introns in prp7 amounting to 297 bp and 71 bp ( location specified in Fig. 1). The paucity of intronic sequences is in contrast to pasN, which contains seven introns with a total of 27 kb of intronic sequence and pasV with five introns totaling approx. 13 kb. Although the intronic structure of ogre is not known, it must also have large introns since its genomic sequence spans approx. 10 kb, while its message accounts for only 2.9 kb ( Watanabe and Kankel, 1990). 3.2. Degree of sequence similarity Fig. 2 shows a pileup comparing Prp7, Prp33, and Prp6 to several innexin family members in Drosophila and C. elegans. Fig. 5A shows a pair-wise comparison of the sequence similarity of the Drosophila innexins. Prp33, PasN and Ogre are all highly related. Prp33 is 65% similar to PasN and 66% similar to Ogre, while PasN and Ogre show 60% similarity. Prp7 is noticeably less related to these three than they are to each other, being 48% similar to Prp33, 47% similar to PasN and 44% similar to Ogre. Though only half the protein sequence of Prp6 is available, a reasonable comparison to the other family members was possible (see Fig. 5A, legend ). Like Prp7, Prp6 is less related to Prp33, PasN and Ogre that they are to each other (44% similar to Prp33, 48% similar to PasN, and 51% similar to Ogre). Finally, Prp6 and Prp7 are even less related, with only

K.D. Curtin et al. / Gene 232 (1999) 191–201




Fig. 1. (A) cDNA and protein sequence of prp7. The locations of introns are indicated by the diamonds at residues 423 and 1132. The first introns are 297 bp and 71 bp long, respectively. (B) cDNA and protein sequence of prp33. The SalI site used to determine the transcription orientation is indicated with a diamond and is in italics. In both (A) and (B), the underlined sequences represent the target regions for the PCR primers used for cloning and the double-underline represents the target of the degenerate portion of the primers.


K.D. Curtin et al. / Gene 232 (1999) 191–201

Fig. 2. Pileup of innexin family members, including all the known Drosophila genes and a C. elegans gene (eat-5) for comparison. The pileup alignment was done using the Clustal method in the Lasergene program. Identical residues are blackened. Dark lines labeled M1–M4 are found above the putative transmembrane spanning domains. Dark arrows are below the residues that were targets for the primers during PCR.

K.D. Curtin et al. / Gene 232 (1999) 191–201


Fig. 3. Mapping of prp7 and prp33. (A) In situ hybridization to polytene chromosomes using a genomic probe that contains part of both prp7 and prp33. The distal tip of the X is to the left and chromosomal subdivision 6E is indicated by the brackets. The bands (1–6) within 6E are marked. The probe hybridizes at band 6E 4,5. (B) Schematic diagram of 6E with the putative locations of ogre, prp7 and prp33 marked with lines. The deletions, Df(1)Sxlbt and Df(1)HA32, used to refine prp7 and prp33 mapping, are also indicated. (C ) Two of the lambda clones from the ogre walk ( Watanabe and Kankel, 1990), 2K and 1P, are indicated in relationship to Df(1)HA32. Distal is to the left and proximal to the right. prp7 hybridized to 2K and prp33 to both 2K and 1P. A restriction map of this area is shown below the lambda clones. The exact map locations are shown for ogre, prp7 and prp33. The direction of transcription is shown for ogre and prp33 and is unknown for prp7. By Southern analysis, prp7 maps to the XhoI fragment of clone 2K which is indicated by the arrowhead below the XhoI map. The SalI site that interrupts prp33 is indicated by an arrowhead below the SaII map. This SalI site is also indicated in Fig. 1B.

39% similarity. Prp6 and Prp7 are clearly more divergent than the other Drosophila innexins. Hydrophobicity analysis of Prp7 and Prp33 shows these proteins to be similar to other family members with regard to the location of putative membrane spanning sequences (Fig. 2). The proposed folding for this family places both the N and C termini inside the cell. Prp7 has about 20 more amino acids in the putative internal loop than the other Drosophila innexin genes. In this regard, Prp7 is more like several C. elegans

proteins, including Eat-5, Unc-7 and Unc-9 which have about 28 extra amino acids (Barnes and Hekimi, 1997). 3.3. Mapping: innexin genes cluster on the chromosome Prp7 and prp33 map very close to each other and to ogre. Using the original PCR probes corresponding to prp7 and prp33, we identified several genomic clones for each gene. Southern analysis of these clones with the PCR probes showed that one contained sequences for


K.D. Curtin et al. / Gene 232 (1999) 191–201

both genes. Subsequent Southern blots using prp7 and prp33 genomic subclones as probes confirmed this. This clone was used in chromosomal in situs and found to hybridize near ogre (Fig. 3A). The genomic clone for prp6 was used in chromosomal in situs and hybridized in 19A 1,2. This location is just distal to pas at 19E3. In situs with the prp7/prp33 genomic probe were then done on chromosomes that were deleted near ogre: Df(1)HA32 and Df(1)Sxlbt. The distal breakpoints of these deletions bracket ogre; ogre lies between these two deletions in 6E2/3 (see Fig. 3B). The distal breakpoint of Df(1)Sxlbt is in 6E1,2 and it deletes the entire ogre locus. The clone containing part of both prp7 and prp33 did not hybridize at all to the Df(1)Sxlbt chromosome. This placed it proximal to 6E1,2. The distal breakpoint of Df(1) HA32 is in 6E4/5 and it does not eliminate ogre. The prp7/prp33 clone showed a faint signal (compared to wild-type chromosomes) when hybridized to the Df(1)HA32 chromosome. This weak hybridization suggested that the clone overlaps the Df(1)HA32 breakpoint with a small amount of the clone lying distal to the breakpoint. This information places prp7 and prp33 approximately in 6E4,5. In order to more finely map these genes we obtained a walk around ogre ( Watanabe and Kankel, 1990). prp7 hybridized to clone 2K from this walk (see Fig. 3C ). prp33 hybridized to both 1P and 2K. Southern analysis of clone 2K showed prp7 mapping to the indicated XhoI fragment, approx. 8–10 kb proximal to ogre. prp33 lies just beyond this, approx. 8 kb proximal to prp7. In addition, by using probes to both ends of the prp33 gene, we could demonstrate that the orientation of prp33 is as indicated in Fig. 3C, opposite to the orientation of ogre. Specifically, probes were made to two parts of prp33. The first probe was from the start of the cDNA clone through the indicated SalI site ( Fig. 1). The second probe ran from that SalI site to the 3∞ end of the cDNA. These were hybridized to a Southern blot of clones 2K and 1P cut with several enzymes, including EcoRI alone, SalI alone, and both enzymes together. The two probes lit up the same EcoRI fragment, but different SalI fragments. The probe to the 5∞ end hybridized to an approx. 4.5 kb band unique to an EcoRI, SalI double digestion of clone 2K. The 3∞ end clone hybridized to a 3 kb EcoRI, SalI fragment of clone 2K and 1P. This orients the clone as shown (Fig. 3C ). 3.4. Expression patterns Probes to prp6, prp7, prp33, and were used for in situ hybridization. Probes were hybridized to embryos at all stages of embryonic development and to the adult central nervous system. Probes for prp6 revealed no staining during embryogenesis. prp7 showed distinct hybridization only during embryonic stages 15–17, when expression was seen around the lobes of the gut, with

more concentrated staining where the lobes join each other ( Fig. 4A). There is no obvious staining in the adult central nervous system, although it was found as

Fig. 4. In situ hybridization of prp7 and prp33 probes to embryos and adult CNS. (A) Prp7 hybridizes around and especially between the lobes of the gut in stage 15–17 embryos. Staining between the gut lobes of a stage 15–16 embryo is indicated by the arrows. Dorsal view, anterior is to the right. (B) prp33 stains the cephalic furrow (arrow) of stage 8 embryos, as well as a repeating pattern in the germ band that may be epidermal. Anterior is to the right, dorsal is up. (C ) prp33 staining of a stage 16–17 embryo. The right arrow indicates the proventriculus. Staining of other parts of the foregut can be seen further anterior (to the right). The left arrow indicates staining of the hindgut. The arrowheads indicate the repeated epidermal staining. Anterior is to the right, dorsal is up. (D) prp33 hybridizes to cells of the optic lamina in the adult brain (arrows). No staining is seen anywhere else in the brain or thoracic ganglia.

K.D. Curtin et al. / Gene 232 (1999) 191–201

a rare species in the adult cDNA library that we searched. Prp33 stains the cephalic furrow in stage 8 embryos, as well as a repeating pattern in the germ band ( Fig. 4B). This pattern may be epidermal, since a prominent repeating epidermal pattern is obvious at later stages. prp33 expression could be seen in the cortex at the cellular blastoderm stage (not shown). In stage 15–17 embryos prp33 is expressed in the foregut, proventriculus, and hindgut ( Fig. 4C ). prp33 is also expressed in a repeating epidermal pattern in stage 15–17 embryos (Fig. 4C ). In the adult CNS, prp33 is expressed in the optic lamina (Fig. 4D). In situs with a probe to ogre show nearly identical staining to prp33 at all stages. This is unlikely to be due to cross hybridization, since probes from the untranslated 5∞ sequence of prp33 give the same staining pattern as probes from the whole gene and prp33 and ogre share no homology in untranslated sequences. The fact that prp33 shows almost complete overlap in expression with ogre can easily be explained if these two genes share transcriptional enhancers. This possibility is supported by the observation that prp33 and ogre map physically close to one another and are transcribed in opposite directions.


1988). One copy of a repeated sequence can mispair with another copy, or, in a region with tandem repeats, the mispairing can occur between longer stretches of repetitive sequences. This mispairing leads to unequal crossing-over: the DNA between the mispaired sequences gets duplicated in one chromatid and deleted in the other. There are some hints in our data about the evolutionary sequence of duplication and divergence in this gene family. prp7 and prp6 are more dissimilar in sequence to prp33, ogre and pasN than any of these are from each other (see Results). In addition, prp7 and prp6 are not highly related (Fig. 5A). This could be due to two factors. First, prp7 and prp6 may have resulted from an early duplication, and thus had more evolutionary time to diverge. Second, prp33, ogre and pas are all expressed in partially overlapping patterns, with prp33 and ogre being highly overlapping. This could mean that these proteins physically interact in some tissues. If that is the case there could be co-evolutionary constraints on these family members that restricted their divergence. In addition to its more divergent sequence, other data suggest an early divergence of prp7 from pas, at least.

4. Discussion We have identified three new innexin genes in Drosophila, designated pas-related proteins ( prp) 6, 7, and 33. The mapping of the new innexin genes indicates that these genes are found in clusters along the chromosome. prp6 maps distal to pas in the same chromosomal division at the base of the X. Passover is at 19E3 and prp6 is at 19A 1,2. Both prp7 and prp33 map even closer to ogre, with prp7 being 8–10 kb distal to ogre and prp33 being another approx. 8 kb distal to prp7. This clustering suggests that the different genes originated by repeated gene duplication. In fact, reiteration of loci seems to occur frequently at the base of the X chromosome, where pas and prp6 lie. There are several loci in the most proximal chromosomal divisions of the X (19 and 20) that seem to be repeated. This is judged from the similarity of mutant phenotypes; genes exhibiting this phenomenon include uncoordinated (unc) and uncoordinated-like (uncl ), melanized (mel ) and melanized-like (mell ), little-fly (lf ) and little-fly-like (lfl ) (see Schalet and Lefevre, 1976). To this list we can now add pas and pas-related protein-6 ( prp-6). The reiteration of loci may be common at the base of the X because of the presence of repetitive DNA. In this region, the euchromatic DNA grades into the heterochromatic DNA. Heterochromatic DNA is extraordinarily repetitive; the transition region (chromosome divisions 19 and 20) also contains a high degree of repetitive DNA ( Yamamoto et al., 1990; Healy et al.,

Fig. 5. (A) Sequence similarity of Drosophila Innexins. For each pair of innexin genes the percent similarity is shown with the percent identical residues in parentheses. For pas, the comparison with pasN is shown; pasV does not give significantly different results. Sequence alignment and identity percentages are from the ’Align’ computer program. However, stricter criteria were used for calculating amino acid similarities. The following were considered conservative substitutions: A/G; D/E; K/R; N/Q; S/T; V/L/I/M; W/F/W. Only the carboxyl half of the prp6 sequence is available. Ogre and pas are slightly less similar over this region (40% identity) than over the whole protein (43% identity). The prp6 values were thus adjusted by multiplying the values by 43/40 to approximate the relatedness of prp6 to other innexins over the whole protein. (B) Schematic diagram of possible sequence of evolutionary divergence of Drosophila innexins.


K.D. Curtin et al. / Gene 232 (1999) 191–201

prp7 has only one small intron compared to several in pas, but the position of that intron is not the same as the positions for any of the pas introns ( Krishnan et al., 1995). The ogre coding region covers about 10 kb ( Watanabe and Kankel, 1990), suggesting the presence of introns but these have never been mapped, making comparison to ogre impossible. The close proximity of ogre and prp33, as well as their overlapping expression, suggest that these duplicated more recently from a common ancestor. The fact that ogre, prp33 and prp7 are clustered proximally on the chromosome, while prp6 and pas are near each other at the opposite end of the chromosome, suggests a close evolutionary relationship between these two even though they are not highly related in sequence. A tentative time line for duplication is shown in Fig. 5B. The innexin family is complex, with 24 members in C. elegans alone. What is the evolutionary value of having a large family of innexin genes? Once a gene has duplicated, the copies may come under the influence of different promoters, allowing distinct regulation of their expression. The proteins themselves may also evolve so that their physiological activities are tailored to the function of the particular cell types in which they are expressed. While the above factors are true for any protein, the fact that innexins multimerize and that gapjunction proteins must be present on both pre- and postsynaptic cells suggests more special considerations in their evolution. The fact that prp33, ogre and even pas show overlap in expression patterns indicates that more than one innexin protein is produced in some cell types. This suggests the possibility that some innexin channels are heteromeric, i.e., single hemi-channels are composed of more than one innexin protein. Such heteromeric junctions could extend even further the range of physiological properties of junctions. Mixed junctions may have different pore-size, gating properties or possibilities for biochemical modification than homomeric junctions. There is evidence that connexins form heteromeric junctions in vivo (Jiang and Goodenough, 1996), though the biological significance of this is unknown. Evolving several gap-junction genes may also lead to a mechanism for controlling which tissues or individual cells form junctions with each other. The hypothesis is that cells will form gap junctions with each other only if the innexin channels expressed in adjacent cells can recognize each other. Most (but not all ) vertebrate connexins can form homotypic channels (both hemichannels are composed of the same connexin), but only about half of all tested pairs of connexins can mate to form heterotypic channels in the Xenopus oocyte expression system ( White and Bruzzone, 1996). In vivo, however, there is little evidence as to whether heterotypic channels specify what cell types will communicate with each other. Some Innexins may act only in a heterotypic

fashion; at least many family members do not seem to form junctions in the paired Xenopus oocyte system. Current data on invertebrate gap junctions are consistent with the hypothesis that different Innexins are used in different synapses of the same cells. In Drosophila, the antennal afferents connect with many other neurons via gap junctions. In Passover mutants, these afferents lose connection to the GF while retaining connection to many other descending neurons. Similarly, in C. elegans, gap junctions form between all the cells of the worm pharynx. Mutation of eat-5 leads to loss of junctions between individual lobes of the pharynx while the cells within each part are still connected to each other (Starich et al., 1996). Weir and Lo (1982, but also see Weir and Lo, 1985) injected Lucifer Yellow into the Drosophila wing imaginal disc. The die passed through gap junctions into cells at the boundary between the anterior and posterior developmental compartments from either direction. However, the dye did not move between cells that were in contact across the compartment boundary. It is possible that different innexins are expressed in the two compartments and they do not form junctions with each other.

Acknowledgements We thank Professor Larry Salkoff ( Washington University) for helping us design and begin the PCR. We thank Professor Doug Kankel ( Yale) for providing the ogre walk and for contributing graphics for Fig. 3.

References Avery, L., 1993. The genetics of feeding in Caenorhabditis elegans. Genetics 133, 897–917. Barnes, T.M., 1994. OPUS: a growing family of gap junction proteins? Trends Genet. 10, 303–305. Barnes, T.M., Hekimi, S., 1997. The Caenorhabditis elegans avermectin resistance and anesthetic response gene unc-9 encodes a member of a protein family implicated in electrical coupling of excitable cells. J. Neurochem. 69 (6), 2251–2260. Bennett, M.V.L., Barrio, L.C., Bargiello, T.A., Spray, D.C., Hertzberg, E., Saez, J.C., 1991. Gap junctions: new tools, new answers, new questions. Neuron 6, 721–732. Crompton, D., Todman, M., Wilkin, M., Ji, S., Davies, J., 1995. Essential and neural transcripts from the Drosophila shaking-B locus are differentially expressed in the embryonic mesoderm and pupal nervous system. Dev. Biol. 170, 142–158. Doe, C.Q., Chu-LaGraff, Q., Wright, D.M., Scott, M.P., 1991. The prospero gene specifies cell fates in the Drosophila central nervous system. Cell 65, 451–464. Foote, C.I., Zhou, L., Zhu, X., Nicholson, B.J., 1998. The pattern of disulfide linkages in the extracellular loop regions of connexin-32 suggests a model for the docking interface of gap junctions. J. Cell Biol. 140 (5), 1187–1197. Goodenough, D.A., Goliger, J.A., Paul, D.L., 1996. Connexins, connexons and intercellular communication. Annu. Rev. Biochem. 65, 475–502.

K.D. Curtin et al. / Gene 232 (1999) 191–201 Healy, M.J., Russell, R.J., Miklos, G.L., 1988. Molecular studies on interspersed repetitive and unique sequences in the region of the complementation group uncoordinated on the X chromosome of Drosophila melanogaster. Mol. Gen. Genet. 213, 63–71. Homyk, T., Szidonya, J., Suzuki, D.T., 1980. Behavioral mutants of Drosophila melanogaster. III. Isolation and mapping of mutations by direct visual observation of behavioral phenotypes. Mol. Gen. Genet. 177, 553–567. Jiang, J.X., Goodenough, D.A., 1996. Heteromeric connexons in lens gap junction channels. Proc. Natl. Acad. Sci. USA 93, 1287–1291. Krishnan, S.N., Frei, E.F., Schalet, A.P., Wyman, R.J., 1995. Molecular basis of intracistronic complementation in the Passover locus of Drosophila. Proc. Natl. Acad. Sci. USA 92, 2021–2025. Krishnan, S.N., Frei, E.F., Swain, G.P., Wyman, R.J., 1993. Passover: a gene required for synaptic connectivity in the giant fiber system of Drosophila. Cell 73, 967–977. Kumar, N.M., Gilula, N.B., 1996. The gap junction communication channel. Cell 84, 381–388. Paul, D.L., 1986. Molecular cloning of cDNA for rat liver gap junction protein. J. Cell. Biol. 103, 123–134. Phelan, P., Nakagawa, M., Wilkin, M.B., Moffat, K.G., O’Kane, C.J., Davies, J.A., Bacon, J.P., 1996. Mutations in shaking-B prevent electrical synapse formation in the Drosophila giant fiber system. J. Neurosci. 16 (3), 1101–1113. Phelan, P., Stebbings, L.A., Baines, R.A., Bacon, J.P., Davies, J.A., Ford, C., 1998. Drosophila Shaking-B protein forms gap junctions in a paired Xenopus oocyte experiment. Nature 391 (6663), 181–184. Sambrook, J., Fritsch, E.F., Maniatis, T., 1989. Molecular Cloning, A Laboratory Manual. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY. Schalet, A., Lefevre Jr., G., 1976. The proximal region of the X chromosome. In: Ashburner, M., Novitski, E. ( Eds.), In: Genetics and Biology of Drosophila vol. 1b. Academic Press, London, pp. 848–896. Starich, T.A., Herman, R.K., Shaw, J.E., 1993. Molecular and genetic analysis of unc-7, a C. elegans gene required for coordinated locomotion. Genetics 133, 527–541. Starich, T.A., Lee, R.Y.N., Panzarella, C., Avery, L., Shaw, J.E., 1996.


Eat-5 and unc-7 represent a multigene family in Caenorhabditis elegans involved in cell-cell coupling. J. Cell Biol. 134, 537–548. Suchnya, T.M., Xu, L.X., Gao, F., Fourtner, C.R., Nicholson, B.J., 1993. Identification of a proline residue as a transduction element involved in voltage gating of gap junctions. Nature 365, 847–849. Sun, Y.A., Wyman, R.J., 1996. Passover eliminates gap junctional communication between the neurons of the Giant Fiber System in Drosophila. J. Neurobiol. 30, 340–348. Swenson, K.I., Jordan, J.R., Beyer, E.C., Paul, D.L., 1989. Formation of gap junctions by expression of connexins in Xenopus oocyte pairs. Cell 57, 145–155. Thomas, J.B., 1980. Mutations affecting the giant fiber system of Drosophila. Soc. Neurosci. Abstr. 6, 742 Thomas, J.B., Wyman, R.J., 1984. Mutations altering synaptic connectivity between identified neurons in Drosophila. J. Neurosci. 4, 530–538. Watanabe, T., Kankel, D.R., 1990. Molecular cloning and analysis of l(1)ogre, a locus of Drosophila melanogaster with prominent effects on the post embryonic development of the central nervous system. Genetics 126, 1033–1044. Watanabe, T., Kankel, D.R., 1992. The l(1)ogre gene of Drosophila is expressed in postembryonic neuroblasts. Dev. Biol. 152, 172–183. Weir, M.P., Lo, C.W., 1982. Gap junction communication compartments in the Drosophila wing disc. Proc. Natl. Acad. Sci. USA 79, 3232–3235. Weir, M.P., Lo, C.W., 1985. An anterior/posterior communication compartment border in engrailed wing discs: possible implications for Drosophila pattern formation. Dev. Biol. 110, 84–90. White, T.W., Bruzzone, R., 1996. Multiple connexin proteins in single intercellular channels: connexin compatibility and functional consequences. J. Bioenerget. Biomembranes 28, 339–350. Wyman, R.J., Thomas, J.B., 1983. What genes are necessary to make an identified synapse? Cold Spring Harbor Symp. Quant. Biol. 48, 641–652. Yamamoto, M.T., Mitchelson, A., Tudor, M., O’Hare, K., Davies, J.A., Miklos, G.L., 1990. Molecular and cytogenetic analysis of the heterochromatin–euchromatin junction region of the Drosophila melanogaster X chromosome using cloned DNA sequences. Genetics 125, 821–832. Zucker, C.S., Cowan, A.F., Reuben, G.F., 1985. Isolation and structure of a rhodopsin gene from D. melanogaster. Cell 40, 851–858.