Genes from two intermoult puffs in Drosophila virilis polytene chromosomes are differentially transcribed during larval development

Initiation and termination of developmental programs require the coordinate control of the expression of sets of genes involved in stage- and tissue-specific functions. This is impressively exemplified in Drosophila larval salivary gland chromosomes, where a number of intermoult puffs regress simultaneously under the influence of the moulting hormone ecdysone prior to puparium formation (reviewed by Ashbumer, 1972). In D. mela-nogaster, at least five of them harbour genes coding for glue proteins, the expression of which is restricted to the second half of the third larval instar. Five of these genes have been cloned and their molecular analysis is under way (for review see Meyerowitz et al. 1987).

In D. virilis, two glue proteins have been identified (Kress, 1982). The gene for glue protein lgp-1 has been localized genetically in the proximal region of the X-chromosome, close to the prominent intermoult puff 16A (Kress, 1986). For lgp-2 no such allocation has been possible, due to the lack of mutants. We were curious, however, whether genes localized in the prominent intermoult puff 55E on chromosome III, are related to glue protein synthesis, because the puff reacted on the injection of glucosamine into the larval hemolymph by regression (Kress, 1973). This aminosugar is a constituent of the larval glue (Kress, 1982).

We describe the microcloning of genes from these two puffs and analyze their tissue- and stage-specific expression. While puff 16A in fact harbours genes for larval glue proteins, the puff 55E genes are transcribed during all three larval stages, suggesting that they are not involved in the developmental program of glue protein synthesis. This will be discussed in connection with the proposed dual role of dipteran salivary glands during larval development.

All experiments were carried out with animals of a D. virilis wild-type stock obtained from the University of Bochum, Germany, collection. The animals were reared at 25°C in milk bottles on standard cornmeal-agar food.

Minilibraries from puff 16A and 55E DNA were obtained by microcloning (Scalenghe et al. 1981). Excised EcoRI-digested genomic DNA was ligated into the EcoRI cloning site of the vector λNM1149 and recombinant clones were selected by growing phage on the P2 lysogen host E. coli pop 13b (Scherer et al. 1981). The inserts of the three clones used in this study were cloned into the polylinker site of pUC8. Subfragments were cloned into pSP6/T7 (BRL). Plasmids were grown in E. coli DH5. Plasmid DNA was prepared according to Maniatis et al. (1982).

For Northern blotting RNA was extracted from whole animals or from isolated tissues according to Warner and Gorenstein (1977) with modifications. Routinely, salivary glands from 50 third instar larvae were collected at 4°C in Locke’s medium (0.9 g NaCl, 0.02 g CaCl₂ -2H2O, 0.04g KC1, 0.02g NaHCO₃ in 99mlH₂O; 1ml 25% glucose added after autoclaving) and homogenized at room temperature in a KONTES glass homogenizer in 100 μl of buffer containing 10 mM-Tris-HCl pH 7.5, 100mM-LiCl, 10mM-LiEDTA, 1% (w/v) LiDS (LETS-buffer) in the presence of two volumes of phenol/chloroform/isoamylalcohol (25:24:1). Where RNA was extracted from whole animals or from tissues other than salivary glands, homogenization was carried out in up to 500 μl of LETS-buffer. After separating the phases by centrifugation, the organic phase was reextracted with buffer containing 10mM-Tris-HCl pH7.5, lOOmM-LiCl, 10mM-LiEDTA and 0.2% (w/v) LiDS. The aqueous phases were pooled and extracted four times with phenol/chloroform/ isoamylalcohol and four times with chloroform. Finally the RNA was precipitated with 0.02 volumes of 5 M-LiCl and 2.5 volumes of ethanol at –70°C. To reduce DNA contaminations, the RNA was reprecipitated twice in the presence of 200mM-LiCl and stored in 70% ethanol at –70°C until use.

³²P-DNA for filter hybridizations was prepared either by nick translation or by random priming of isolated DNA fragments. The reactions were carried out with commercial kits under conditions recommended by the suppliers (BRL, Boehringer). Labeled DNA was separated from unincorporated nucleotides by chromatography on Bio-Gel PE-60 (BioRad) columns.

For screening the minilibraries obtained by microcloning, ³²P-CDNA was prepared. Poly(A)RNA from about 400 third larval instar salivary glands was isolated as described previously (Kress et al. 1985). Labeled cDNA was prepared by a standard AMV reverse transcriptase reaction in 50μ1 in the presence of 1 i.u. μl¹ RNasin, 100i.u. of enzyme, dATP, dGTP and dTTP at 200 μM and dCTP at 12 μM final concentration and 50μCi ³²P-dCTP. The cDNA was purified by chromatography on Bio-Gel P-60 (BioRad), equilibrated in TE-buffer. Fractions containing the first peak of radioactivity were pooled and stored at –20°C.

Biotinylated DNA was prepared according to the protocol provided by ENZO, using a commercial nick translation kit (BRL).

DNA fragments were separated on agarose gels and isolated by electroelution into dialysis bags according to Maniatis et al. (1982). RNA was separated in formaldehyde gels as described by Davis et al. (1986) in the presence of ethidium bromide (0.5 μg ml^-1). A RNA-ladder (0.24–9.5 kb; BRL) and Haelll digested ϕX—DNA were coelectrophoresed as length standards. After separation, RNA was transferred onto nylon membranes in 10×SSC under capillary blotting conditions without prior destaining and u.v.-fixed. Prior to separation, aliquots of the samples were electrophoresed under nondenaturing conditions in ethidium-bromide-stained agarose gels (1.5%) in 1×TBE-buffer for testing the quality and quantity of the RNA.

Squash preparations and surface spreads of polytene chromo-somes to be used for in situ hybridizations were obtained as described previously (Kress et al. 1985).

Plaque hybridizations with ³²P-cDNA were carried out on nitrocellulose filters according to Maniatis et al. (1982). Prior to prehybridization, filters were treated with Proteinase K (Hertner et al. 1986). Northern blot hybridizations were carried out at 42°C in 50% formamide, lxDenhardt’s, 1% SDS, 5×SSC in sealed cylinders under constant rotation overnight. For repeated use of filters, labeled probes were removed in boiling 0.1 ×SSPE (15mM-NaCl, 1mM-NaH₂PO4, 0.1 mM-EDTA), 1 % SDS for 5–10 min. In situ hybridizations with Biotin-DNA were carried out as described in detail earlier (Kress et al. 1985). Bound DNA was visualized by streptavidin-horseradish-peroxidase (Detec-I-hrp, ENZO) according to a protocol provided by U. Walldorf, Biocenter Basel.

All chemicals were obtained from commercial sources. Isotopes were purchased from NEN (³²P).

DNA from one (55E) or two (16A) chromosomes was excised for microcloning. Clones containing genomic DNA, which is transcribed during the late third larval instar in salivary glands, were identified by plaque hybridization with ³²P-labeled cDNA, transcribed from poly(A)RNA, which had been isolated from salivary glands of that stage. From the puff 16A library, one positive clone (λ DvA31) was isolated; it contains a 2.9 kb insert (=fA31). Among the puff 55E clones, five were positive. Three of them showed the same insert length of 4.8 kb and cross hybridized with each other. The two remaining clones had a 6.1 kb insert and cross hybridized also. There was no cross hybridization between the 4.8 and 6.1 kb clones, indicating that they contain different sequences. From each of the two classes, the members of which are identical in either case (tested by restriction mapping; data not shown), one clone was chosen for further analysis. The inserts of clones λ DvA31, λ DvK21 (4.8kb) and λ DvK30 (6.1 kb) were cloned into pUC8 (=pDvA31, pDvK21 and pDvK30) and characterized by restriction mapping and sequencing. Simplified versions of the restriction maps are shown in Fig. 1. By cloning and sequencing of an overlapping genomic DNA fragment (E.Thiiroff, unpublished data), we could verify that the two puff 55E genomic fragments fK21 and fK30 are contiguous, the order being 5′-fK21-fK30-3′.

The analysis of transcripts by Northern blotting was carried out with total RNA from whole animals or from isolated tissues. Where the complete 2.9 kb fA31 fragment was used as hybridization probe, three bands were labeled on blots of third larval instar salivary gland RNA. The strongest signal came from a band of approximately 1100nt (Fig. 2A, transcript A). Slightly above it, at about 1300 nt, a weaker signal was present (Fig. 2A, transcript B). We estimate that its intensity was less than 5 % of the transcript A signal intensity. In the >4000 nt range, a third labeled band of low and varying intensity could be identified (Fig. 2A, transcript X).

Since both A and X transcripts were labeled exclusively by a central 500 bp SstI-RsaI and the adjacent 400 bp RsaI-EcoRV subfragment (compare Fig. 1, fA31, fragment A and Fig. 2B), we conclude that these transcripts share common sequences. For transcript X, we must assume that more than 3000 nt originate from sequences that are not present on the cloned fA31 genomic fragment. A similar conclusion is drawn in the case of transcript B, which was exclusively labeled by a 380 bp XbaI-HincII subfragment (Fig. 1, fA31, fragment B), although its electrophoretic mobility suggests a length of 1300 nt (Fig. 2C).

Among the subfragments of fK21 the contiguous HpaI/EcoRV and EcoRN/SstI fragments (see Fig. 1, fK21), 1100bp in total, hybridized with two transcripts in the 850 – 950nt range (Fig. 2D), suggesting the existence of a single gene from which these transcripts are released. There was always a higher amount of the larger of the two transcripts present. In the case of fK30, a 580 bp XbaI/Nsil fragment (Fig. 1, fK30) hybridized at equal intensity with two transcripts in the 500 – 600 nt range (Fig. 2E). We conclude that the gene residing on this fragment also gives rise to two transcripts.

Relevant information comes from the sequence data, which will be presented in detail in subsequent papers. In the present context it is sufficient to know that two of the puff 16A transcripts represent the mRNAs for larval glue proteins. The A transcript codes for glue protein, lgp-1, the B transcript for another related glue protein. With great probability the corresponding gene is localized 5′ of the cloned fA31 fragment in the puff 16A DNA, while the small subfragment hybridizing with the B transcripts may represent a pseudogene (Swida et al. 1990).

The amino acid sequences deduced from the open reading frames of the transcribed K21 and K30 DNA sequences (E.Thiiroff, unpublished results) reveal no relation to any of the known Drosophila glue protein sequences.

Biotinylated DNA probes were used for the cytological localization of the cloned sequences. For puff 16A sequences, the clone pDvA31 was used. In the case of puff 55E, a plasmid (pDvETl) was constructed, the 1.7 kb insert of which contained only the subfragments hybridizing with its transcripts (Fig. 1, fK21 and fK30). In either case, the corresponding puff was the single site of hybridization on conventionally squashed chromo-somes (data not shown). In situ hybridization to surface spread polytene chromosomes (Kalisch and Hägele, 1981) suggests that the transcribed sequences in these two puffs are localized in chromatin, which exhibits different degrees of decondensation. Hybridization of clone pDvA31 DNA, harbouring 2.9 kb genomic sequences, resulted in a broad band of labeling in the puffed region. On the contrary, the puff 55E probe showed a single, sharp band of intense hybridization at the chromomere 55E1 (Fig. 3), although there are about 2 kb of DNA between the two transcribed subfragments comprising 1.7 kb (see Fig. 1).

During development, the genes residing in puffs 16A and 55E are transcribed in different ways. The two transcripts either from the K21 or from the K30 gene in puff 55E are formed during all three larval instars (Fig. 4A and B). The puff 16A transcripts could only be identified in RNA from third larval instar glands (Fig. 4C; transcript X signal not shown).

Tissue specificity of expression was examined in third instar larvae. Total RNA from brain (including ventral ganglia), fat body, proventricle and anterior portion of the midgut (stomach, including gastric caecae), salivary glands and carcasses was isolated separately. K21 and K30 transcripts were exclusively found in RNA from salivary glands, indicating that the corresponding genes are only active in these organs. Typical hybridization patterns are shown in Fig, 5A and B, where RNAsfrom salivary glands, fat body and stomach are compared. This restriction was less pronounced in case of the puff 16A transcripts. Although they are preferentially produced in salivary glands (Fig. 5C and D), long time exposure of blots revealed weak signals from stomach RNA also (Fig. 5E; data for transcript B is similar, but not shown). During development these transcripts were found in the stomach only during the second half of the third larval instar (data not shown), i.e. at the time when the corresponding genes are abundantly transcribed in the salivary glands. Relatively strong signals (in relation to the signals from stomach) were observed in RNA from fat body. In this case it is interesting, however, that transcription was confined to the first few hours of the wandering phase of larvae, which commences about 11 h prior to puparium formation (Fig. 6).

In the present study, we initiate the molecular analysis of genes in ecdysone-responsive puffs of D. virilis salivary gland polytene chromosomes. The genes we selected were microcloned from DNA of the two intermoult puffs 16A and 55E. The two gene sites exhibit different chromatin morphology (see Fig. 3). It is possible that the supposed differences in decondensation are correlated with differences in the transcription rates. We estimate that, on average, the ratio between the signal intensities of the Lgp-1 transcripts of puff 16A and the puff 55E transcripts is more than 10:1 (compare legends to Figs 4 and 5).

In puff 55E, two genes (K21 and K30) were identified on a 11 kb genomic DNA fragment, which presumably represents most of the DNA in the small chromomere 55E1. Their transcripts are present exclusively in the salivary glands during all three larval stages in increasing amounts (compare Fig. 4A and B). From the relation between the strength of the signals and the numbers of animals used for RNA extraction, we estimate that the amounts of transcripts differ by a factor of about 40-fold between glands of the first and third larval instar. Since this corresponds well to the increase of the gland volume prior to the onset of glue protein synthesis (about 50- to 60-fold), the abundance of these transcripts in relation to total RNA seems to stay fairly constant. This is an indication at the molecular level that the glands possess a continuous function throughout all larval stages. One possibility is their connection to the proposed early function of dipteran larval salivary glands as digestive organs (Phillips and Swift, 1965). This function should be required throughout the feeding stages and the wandering phase of late third instar larvae until the gut is emptied prior to puparium formation. Consistently, during the last 10h of larval life, the size of puff 55E regresses continuously (Kress, 1973). In D. hydei and in D. melanogaster, secretory granules have been identified in mid third larval instar salivary glands, which gradually disappear during the course of glue protein synthesis (Berendes, 1965; von Gaudecker, 1972). They were interpreted to contain digestive enzymes. During the same developmental period, the synthesis of several glycosylated proteins ceases in D. virilis (Kress, 1982).

The cloning of the 2.9 kb fA31 DNA fragment from puff 16A led to the identification and characterization of the gene that codes for larval glue protein lgp-1 (Swida et al. 1990). The Lgp-1 transcripts (=A transcripts) are about 1100 nt in length and accumulate at remarkably high levels in the salivary glands exclusively during the third larval instar. The same developmental pattern was found for the 1300 nt B transcripts, which represent the mRNA for another related glue protein, the gene of which also resides in puff 16A (Swida et al. unpublished data). It turns out, therefore, that this puff region harbours a cluster of at least two genes, which code for two related glue proteins. A comparable organization has been reported for the cluster of three related glue protein genes in puff 68C of D. melanogaster, which reside on a 5 kb genomic DNA fragment (Meyerowitz and Hogness, 1982; Garfinkel et al. 1983).

It was surprising to find puff 16A transcripts also in RNA preparations from midgut and from fat body. While the level of these transcripts in midgut is rather low (Fig. 5) and persists throughout the second half of the third larval instar, we observed a short but intense pulse of transcript synthesis for only a few hours in the fat body (Fig. 6). In the latter case, the transcripts seem to be unstable in contrast to the salivary gland Lgp-1 transcripts, which may be translated for at least 5 h at a similar rate in the absence of RNA synthesis (Kress, 1979). It is improbable, therefore, that in the fat body significant amounts of the corresponding protein are synthesized and the biological role of this transient activation remains obscure. In D. melanogaster midgut and fat body, bona fide transcripts of glue protein genes have not been identified. However, in ADH-null strains transformed with fusion genes containing promoter sequences of the glue protein gene Sgs-4 and of ADH gene sequences, ADH expression was found in larval midgut and fat body. It was concluded that trans-acting factors that are required for the activity of the Sgs-4 enhancer are present in these organs (Shermoen et al. 1987). It is possible, therefore, that in D. melanogaster midgut glue protein gene transcripts are present, but below the level of detection. Since in this species, in contrast to D. virilis, the staging of larvae is difficult, a short pulse of transcript formation in the fat body may have escaped its detection.

Two different physiological functions have been proposed for the dipteran larval salivary gland: a long-term role as a digestive organ working throughout the larval stages and a short-term role during the final larval stage, when it synthesizes glue proteins (Phillips and Swift, 1965). The cloning of genes from the D. virilis 55E puff, the products of which may be involved in the long-term function of the salivary glands, deserves special attention. First, from the analysis of these genes and of their products we could learn more about the early function of larval salivary glands in Drosophila. Second, since these genes are transcribed in the same tissue as the glue protein genes, but are activated at an earlier stage, it should be possible to discriminate those cw-regulatory elements, which are responsible for the stage-specific induction of puff 55E and puff 16A genes, respectively, or, on the other hand, to identify similar ones, which are involved in their concomitant repression prior to puparium formation. It is the aim of our ongoing work to identify these elements.

We would like to thank Dr H. Jaeckle for his contribution to the microcloning experiments. We are grateful to Ms U. Cinque and Ms S. Maletz for excellent technical assistance. We are indebted to the Carl Zeiss Company, Berlin for giving us the opportunity to use the ACE system. This work was supported by the Deutsche Forschungsgemeinschaft (Kr 509/ 7) and by funds of the FU Berlin (FGS Regulationsstrukturen von NukleinsSuren und Proteinen).