A cell-specific glycosylated silk protein from Chironomus thummi salivary glands. Cloning, chromosomal localization, and characterization of cDNA.

Chironomid salivary glands contain 40 cells dedicated to the synthesis of a relatively small ensemble of silk proteins. Glands in some species contain a special lobe composed of 4 cells distinguishable from the others. We have cloned a special lobe-specific cDNA from Chironomus thummi salivary glands. Northern blots of salivary gland RNA demonstrated that the cDNA hybridizes to a 2.5-kilobase transcript present only in the special lobe. In situ hybridization mapped the gene encoding this cDNA to region A2b on polytene chromosome IV, the locus of the special lobe-specific Balbiani ring a. The deduced amino acid sequence encodes a protein with a calculated molecular mass of 77 kDa and numerous potential glycosylation sites; it appears unrelated to other known chironomid silk proteins. Polyclonal antibody, raised against a cDNA-encoded fusion protein, reacted exclusively with a special lobe-specific 160-kDa silk protein. Lectin binding studies indicate that the immunoreactive 160-kDa protein contains both N- and O-linked glycan moieties. We conclude that glycosylation most likely contributes to the difference between calculated and apparent molecular masses and that this cDNA encodes the special lobe-specific silk protein previously described as ssp160 (Kolesnikov, N. N., Karakin, E. I., Sebeleva, T. E., Meyer, L., and Serfling, E. (1981) Chromosoma 83, 661-677).

Chironomid salivary glands contain 40 cells dedicated to the synthesis of a relatively small ensemble of silk proteins. Glands in some species contain a special lobe composed of 4 cells distinguishable from the others. We have cloned a special lobe-specific cDNA from Chironomus thummi salivary glands. Northern blots of salivary gland RNA demonstrated that the cDNA hybridizes to a 2.5-kilobase transcript present only in the special lobe. In situ hybridization mapped the gene encoding this cDNA to region A2b on polytene chromosome IV, the locus of the special lobe-specific Balbiani ring a. The deduced amino acid sequence encodes a protein with a calculated molecular mass of 77 kDa and numerous potential glycosylation sites; it appears unrelated to other known chironomid silk proteins. Polyclonal antibody, raised against a cDNA-encoded fusion protein, reacted exclusively with a special lobe-specific 160-kDa silk protein. Lectin binding studies indicate that the immunoreactive 160-kDa protein contains both N-and O-linked glycan moieties. We conclude that glycosylation most likely contributes to the difference between calculated and apparent molecular masses and that this cDNA encodes the special lobe-specific silk protein previously described as ssp160 (Kolesnikov,  Silks are produced by a wide variety of arthropods including spiders and larvae of hundreds of insect species. Few silks are well-characterized, but emerging evidence suggests that differences in biochemical and biophysical properties are attributable to constituents that vary considerably among species (1). In fact, silk proteins from silkworms (2)(3)(4), spiders (5), and aquatic larva of the midge, Chironomus tentans (6,7), are remarkably different.
The chironomid salivary gland is an exceptional system for the study of silk protein synthesis and assembly from gene to finished product (6,7). The polytene chromosomes are well-characterized and contain Balbiani rings (BRs), 1 sites of intensive transcription of silk protein-encoding genes (6). BR mRNAs that encode the 1000-kDa silk proteins are visible during transcription, packaging, transport, and translation (8 -12).
The nucleotide sequences for 13 major C. tentans silk proteins are known (6,7). These range in size from 12 to Ͼ1000 kDa, and their primary structures are highly unorthodox. Most consist of 50 -150 tandem repeats of unusual amino acid sequences with a Pro-or Cys-containing motif, or both. Some of the silk proteins are glycosylated (13,14) or phosphorylated (15,16), and all are secreted and stored in the lumen of the salivary gland. Proteins isolated from the lumen exist as soluble complexes, capable of disassembly and reassembly in vitro (17). However, in vivo, the lumenal contents are pumped on demand through a long, thin salivary duct and exit the animal's mouth as an insoluble silk fiber. Nothing is known about the mechanism by which this phase transition occurs.
The "special lobe" of Chironomus salivary glands is an enigma. This lobe is present in many, but not all, species and contains four secretory cells, lying adjacent to the salivary duct (18). The special lobe is distinct from the remainder of the gland: (i) its polytene chromosomes have one additional BR (18), (ii) it contains one additional, special lobe-specific, silk protein (19,20); (iii) its cells and secretion contain Beermann's secretory granules (18,21); (iv) the lobe histochemically stains differentially for glycoprotein (22). These observations led to the hypothesis that the special lobe-specific BR contains a gene encoding a glycosylated silk protein that is secreted in Beermann's granules and is responsible for the lobe's differential staining. Furthermore, the proximity of this lobe to the salivary duct suggests that these granules somehow contribute to silk protein assembly into fibers.
We report here the first step in testing this hypothesis in Chironomus thummi. A special lobe-specific cDNA was cloned that hybridizes in situ to special lobe-specific BRa. This cDNA encodes a protein with numerous sites for potential glycosylation. Antibody raised against a cDNA-encoded fusion protein reacts with ssp160, the only known special lobe-specific silk protein in C. thummi salivary glands (20), and lectin binding demonstrates that ssp160 contains both N-and O-linked carbohydrate.

EXPERIMENTAL PROCEDURES
Salivary Gland Dissection-C. thummi thummi (Novosibirsk strain) was raised (22) on powdered dolomitic limestone which appeared to increase the amount of special lobe secretion stored in the gland. Salivary glands, dissected from fourth instar larvae and prepupae, were rinsed in 0.8% NaCl and fixed in 70% ethanol. The special lobe was identified by its location and distinctive appearance and cut away from the gland. For RNA extracts, the remainder of the gland was taken as the "main lobe" fraction. 2 Since proteins originating from different parts of the gland might mix within the lumen, only the distal quarter of the gland was used for main lobe protein extracts. All lobes were stored in 70% ethanol at Ϫ20°C until an adequate amount of material was accumulated. To obtain secreted silk proteins, the cells were dissected away from the core of rubbery lumenal protein, and the branch of special lobe secretion was identified by its granular appearance.
RNA Isolation and Blotting-Poly(A) ϩ RNA was isolated by oligo(dT)-cellulose chromatography of RNA extracted (23) from 200 special or main lobes at one time. Yields were about 2 ng/special lobe and 13 ng/main lobe; this represented about 4% of total RNA. RNA was separated by electrophoresis on denaturing (24), 0.7% agarose gels and electroblotted to nylon membranes. The size of hybridized transcripts was determined by comparison to an RNA size ladder.
cDNA Library Construction and Screening-Double-stranded, SalI-NotI-adapted, 32 P-labeled cDNA was synthesized from special and main lobe poly(A) ϩ RNA using the Superscript Lambda System (Life Technologies, Inc.). Autoradiograms of agarose gels revealed a 2.5-kb special lobe-specific cDNA that was extracted and ligated to gt22 NotI-SalI arms and packaged. gt23 (25) and ZAP (26) libraries were also made from whole salivary gland poly(A) ϩ RNA. The resulting libraries were titered and amplified. Plaques were screened and purified by hybridization of duplicate plaque lifts (27) with a special lobe-specific subtracted hybridization probe (28). Single-stranded, 32 P-labeled cDNA was made from 1 g of special lobe poly(A) ϩ RNA, and sequences common to both lobes were removed by hybridization with 10 g of main lobe poly(A) ϩ RNA followed by hydroxylapatite column chromatography. The resulting probe had a specific activity of 3.3 ϫ 10 8 cpm/g. Hybridization was done with 1 ϫ 10 6 cpm/ml of the special lobe-specific subtracted probe in 0.6 M NaCl, 8 mM EDTA, 0.12 M Tris-HCl, pH 8.0, at 65°C. Inserts in phage DNA from four clones from the 2.5-kb cDNA library were compared by digestion with restriction endonucleases; all appeared identical. 160.1 was studied further.
cDNA Amplification and Labeling-PCR (29) was done in 100 l containing 1 ng of 160.1 DNA, 0.6 M each primer, 0.1 mM tetramethylammonium chloride (30), 1.5 mM MgCl 2 , 0.2 mM each deoxyribonucleotide, 10 mM Tris-HCl, pH 8.3, 50 mM KCl, and 5 units of AmpliTaq polymerase (Perkin Elmer). For radiolabeling, 200 Ci of either [␣-32 P]dATP or -dCTP was included, and nonradioactive nucleotides were reduced to 20 M each. Thirty cycles of 60°C annealing and 72°C extension were performed. To generate a 2.5-kb 32 P-labeled probe which encompassed the entire cDNA insert, gt11 forward and reverse primers (New England Biolabs), which hybridized to the vector's ␤-galactosidase sequence flanking the cDNA insert, were used. To provide a control for Northern blotting experiments, we made a 475-bp PCR probe to a 5.5-kb sp220 mRNA 3 which is found in both special and main lobes.
In Situ Hybridization-Standard salivary gland squash preparations were hybridized in situ at 65°C in 0.75 M NaCl, 68 mM sodium citrate, pH 7.0, 0.1% SDS with a 160.1 2.5-kb PCR probe labeled by randomprimed DNA synthesis in the presence of digoxigenin-dUTP. Hybridization and detection of hybrids with fluorescein-labeled anti-digoxigenin antibody were performed as described (31).
Subcloning and Sequencing-The complete cDNA was excised with SalI and NotI from 160.1 and subcloned into the corresponding sites of M13BM20 (Boehringer Mannheim) to yield M13BM20[S3 N]. To make a construct with the cDNA insert oriented in the reverse direction, an EcoRI-NotI fragment of 160.1 was subcloned to yield M13BM20[N4E]. Nested sets of deletions were made with exonuclease III, aligned by T-tracks, and sequenced using Bst polymerase (Bio-Rad) in a dideoxy chain termination protocol. Data for contiguous segments of both cDNA strands were compiled and analyzed using Pustell's software (32).
Fusion Protein Expression and Purification-In order to produce antibodies against the protein encoded by 160.1 cDNA, M13BM20[S3 N] was cut with AluI, and the 288-bp fragment (cDNA nucleotides 1086 -1373) was blunt-end-ligated to the filled-in SalI site in the expression vector pET12b (Novagen). This placed the cDNA in-frame with the ompT leader. Transformants of Escherichia coli strain BL21(DE3)pLysS (33) were identified by colony hybridization to a PCR cDNA probe. The construct was confirmed by DNA sequencing. To assay for production of fusion protein, vector and vector plus insertcontaining cultures were incubated with and without isopropyl-1-thio-␤-D-galactopyranoside, and their lysates were compared on SDS-polyacrylamide gels (34). To produce fusion protein for antibody production, log-phase cells were induced for 8 h, and total cell lysates were prepared (33). The insoluble fusion protein was pelleted by centrifugation at 5000 ϫ g for 10 min, dissolved in 6 M guanidine HCl, reduced with dithiothreitol, carboxymethylated, and acetone-precipitated (16). Pellets were dissolved in 6 M urea, clarified by centrifugation at 14,000 ϫ g for 10 min, combined with SDS-sample buffer (34), and fusion protein was purified by continuous elution of fractions from a preparative gel electrophoresis apparatus. Fractions were examined by electrophoresis and assayed for protein using bicinchoninic acid (35) with bovine serum albumin as a standard. Those containing fusion protein were pooled, dialyzed against water, and lyophilized. Approximately 0.5 mg of fusion protein was recovered from 5 mg of protein loaded on each preparative gel. The identity of the fusion protein was confirmed by amino acid analysis.
Antibody Production and Purification-Rabbits were immunized subcutaneously at multiple sites with a total of 170 g of purified fusion protein emulsified with Freund's adjuvant (36). Booster injections of 100 g and 280 g were given on days 14 and 31, respectively. Immune serum was collected on day 49.
Affinity-purified antibodies were obtained by chromatography (37) using 500 g of gel-purified fusion protein coupled to Affi-Gel 10 (Bio-Rad). To remove antibodies against the ompT leader and co-purified E. coli proteins, affinity-purified antibody (14 g of protein in 1 ml of 1% non-fat dry milk, 20 mM Tris-HCl, pH 7.5, 0.5 M NaCl, 0.02% NaN 3 ) was adsorbed for 4 h at 20°C with 275 g of protein from a control lysate of isopropyl-1-thio-␤-D-galactopyranoside-induced BL21(DE3)pLysS cells containing pET12b. The antibody was then diluted and used for Western blotting (38). For competition studies, lysate from BL21 (DE3)pLysS-containing pET12b with the cDNA insert (i.e. with fusion protein) was used for adsorption.
Western Blotting-For analysis of salivary gland proteins, samples of either special and distal-main lobes (cells plus secretion) or secretion alone were solubilized with guanidine HCl, reduced and carboxymethylated, precipitated with acetone, and separated on 3-20% concaveexponential gradient SDS-polyacrylamide gels (16). Following electrophoresis, proteins were electroblotted to nitrocellulose (38) and visualized with Ponceau S to verify transfer (39). Some blots were stained with AuroDye (Amersham). Others were blocked, incubated overnight with rabbit anti-fusion protein antibody diluted 1:500, and rinsed (40). Bound antibody was visualized using goat anti-rabbit secondary antibody conjugated with alkaline phosphatase.
Lectin Binding-Western blots were incubated with digoxigeninlabeled lectins that were subsequently detected with an anti-digoxigenin antibody coupled to alkaline phosphatase. GNA, PNA, and ConA binding were done according to the manufacturer's protocols (Boehringer Mannheim), except that the PNA and ConA were diluted to 5 g/ml and 3% bovine serum albumin (Fraction V) was used as a blocking agent for ConA.
To examine protein binding to lectins, special lobe proteins were extracted as described above (16) and dialyzed against TBS (50 mM Tris-HCl, pH 7.5, 150 mM NaCl) keeping the protein concentration less than 25 g/ml to avoid precipitation. Dialysate was equilibrated with an equal volume of either GNA-Sepharose in TBS or PNA-Sepharose (EY Laboratories) in TBS plus 1 mM each MgCl 2 , CaCl 2 , and MnCl 2 , for 2 h at room temperature with constant mixing. Supernatants containing unbound protein were removed, and the resin was rinsed thoroughly. To release bound protein, resins were then incubated for 1 h in 2 volumes of binding buffer containing 1 M mannose (GNA-Sepharose) (45) or 0.2 M lactose (PNA-Sepharose) (46). The supernatant was removed, dialyzed against water, and proteins were precipitated with acetone. These fractions were subjected to SDS-polyacrylamide gel electrophoresis and immunoblotted.

RESULTS
Cloning a Special Lobe-specific cDNA-Our search for a cDNA clone encoding a special lobe-specific protein began by comparing double-stranded, 32 P-labeled cDNA made from spe-cial and main lobe poly(A) ϩ RNA on an agarose gel (Fig. 1). Several discrete (1.5-, 1.8-, and 4.5-kb) cDNAs were present in both samples and 5.5-and ϳ20-kb cDNAs were enriched in the main lobe. However, the most striking result was an abundant 2.5-kb cDNA that was present only in the special lobe sample. Even with extensive exposure, no corresponding band was seen in main lobe material. This cDNA was used to make a sizefraction library in gt22.
A radiolabeled special lobe-specific cDNA probe was made by subtractive hybridization against an excess of main lobe poly(A) ϩ RNA. To check probe specificity, we first hybridized the probe to a Northern blot containing special and main lobe RNA. Hybridization was seen only to a 2.5-kb special lobe RNA ( Fig. 2A). This confirmed the lobe specificity of the probe and demonstrated that only one size class of special lobe-specific transcripts was detected.
To determine the proportion of special lobe-specific transcripts in the entire gland, we screened two C. thummi whole gland cDNA libraries with the special lobe-specific subtracted probe. 1.7% of 1.2 ϫ 10 4 clones in a primary gt23 library and 1.8% of 1.8 ϫ 10 3 clones in an amplified ZAP library were hybridized. However, when this subtracted probe was used to screen the 2.5-kb size fraction gt22 library, 92% of the insertcontaining clones hybridized. The first four clones selected had identical restriction enzyme cleavage patterns, suggesting that they are representative of cDNA molecules taken from the agarose gel. One, 160.1, was chosen for further characterization.
To determine the size of the RNA from which this cDNA was derived, primers complementary to flanking vector sequences were used to amplify and radiolabel the cloned 2.5-kb cDNA by PCR. Northern blots of special and main lobe-derived RNA showed specific hybridization only to a 2.5-kb RNA from special lobes (Fig. 2B). Subsequent rehybridization of this blot with a probe for 5.5-kb sp220 mRNA verified that lack of hybridization to main lobe transcripts was not due to RNA degradation (Fig. 2C). We conclude that the cloned cDNA is a near fulllength copy of a 2.5-kb special lobe-specific RNA.
Localizing the Gene-The chromosomal location of the gene encoding this cDNA was determined by in situ hybridization to salivary gland polytene chromosomes. The hybridization signal of the PCR-amplified and digoxigenin-labeled cDNA probe was localized to region A2b of chromosome IV (Fig. 3). Samples from two subspecies, C. thummi thummi and C. thummi piger, both gave the same result. Region A2b on polytene chromosomes in main lobe cells contains a thick, condensed band that forms BRa in special lobe cells (41).
Determining the Primary Sequence-The cloned cDNA has a length of 2387 nucleotides encoding one long open reading frame spanning nucleotides 17-2242 (Fig. 4). The calculated molecular mass for the deduced 742-amino acid sequence is 77 kDa. A central, highly basic region confers a predicted pI of 9.4 to the protein. The amino acid composition reveals a high percentage of serines (15%) and threonines (12%), many of which are found in four distinct regions. The first three (amino acids 105-140, 274 -307, and 452-501), each contain six or seven potential sites for N-linked glycosylation (NX(S/T) where X P) (42). The fourth region lies near the carboxyl end of the protein (amino acids 663-687) and consists of a stretch of 21 threonines in 25 residues, potential sites for O-linked glycosylation (43).
Analysis of the regions containing the NX(S/T) motifs reveals  (2) RNA were separated in parallel to RNA markers (numbers on left) by electrophoresis on denaturing 0.7% agarose gels, electroblotted to nylon membranes, and hybridized with the following radioactive probes. A, main lobe-subtracted, special lobe-specific cDNA; B, a 2.5-kb PCR probe from 160.1 cDNA; C, a 475-bp PCR probe for sp220 mRNA. A and B are different blots; however, C was made by rehybridization of B after radioactivity in the first probe was allowed to decay. cDNA for ssp160 that region I is comprised of six almost-perfect tandem copies of the hexameric repeat TSSNST (Fig. 5). Region II has very similar hexamers; variations, including a 2-residue deletion, appear limited to the first three residues. Region III is separated from the others by the central basic core and is most divergent; however, seven 6 -11-residue segments can be aligned by a conserved NXT motif. These patterns suggest that internal sequence duplication has occurred during evolution, preserving the NXT motifs.
While there is no definitive consensus for sites involved in O-linked glycosylation, residues that are so modified tend to be located ϩ1 or Ϫ3 residues from a proline (e.g. P(S/T) or (S/ T)XXP) or located near other serines, threonines, or alanines (43). Based on these criteria, Thr 60 , Ser 185 , Ser 251 , Thr 335 , Ser 581 , Thr 683 , and Ser 727 are possible candidates for O-glycosylation along with the numerous serines and threonines which neighbor each other.
Identifying the Encoded Protein-A 288-bp AluI fragment of cDNA encoding the central, basic region was subcloned into pET12b to provide inducible expression of an ompT/cDNA-encoded fusion protein. Anti-fusion protein antibody was reacted with Western blots made from total salivary gland proteins separated by electrophoresis on SDS-polyacrylamide gels. Immunoreactivity was limited to one special lobe protein with an apparent molecular mass of about 160 kDa (Fig. 6A). To demonstrate that immunoreactivity was due to epitopes encoded by the cDNA, a parallel blot was reacted with antibody that had been preadsorbed with fusion protein (Fig.  6B). No immunoreactive band was seen, even when the color reaction was deliberately overdeveloped. To determine whether the immunoreactive protein was secreted into the glandular lumen, we examined samples of purified secretion. An immu-noreactive 160-kDa protein was present in special lobe secretion only (Fig. 6C). Based on lobe specificity, electrophoretic mobility, and secretion into the lumen, the immunoreactive protein fits the criteria defining ssp160 (20).
Examining Carbohydrate Moieties-Since the cDNA-encoded protein has numerous glycosylation sites which, if used, could contribute to differences in calculated and apparent molecular masses, Western blotted proteins were reacted with lectins for detection of carbohydrate. Nearly all stainable proteins are common to special and main lobes, except the 160-kDa protein present only in the special lobe lane (Fig. 7A). N-linked oligosaccharides react with ConA, which binds to internal and nonreducing terminal ␣-mannosyl residues (44). Over a dozen ConA-binding proteins were common to both special and main lobes (Fig. 7B); however, one additional ConA-binding protein, migrating at 160 kDa (arrow) was present only in the special lobe. GNA binds only those glycans having a terminal mannose linked to mannose (45). Consequently, while all GNA-binding proteins react with ConA, only a subset of the N-glycosylated proteins reacting with ConA are detected with GNA. At least two GNA-binding proteins were present in both special and main lobes (Fig. 7C); a 160-kDa GNA-binding protein (arrow) was present only in the special lobe. PNA specifically recognizes the disaccharide, Gal␤1-3GalNAc (46), which is often present as the core of O-linked glycan moieties in non-yeast eukaryotes (47). PNA bound to three proteins in both special and main lobes (Fig. 7D). PNA also bound a 160-kDa protein (arrow) present only in the special lobe. This suggested that the 160-kDa special lobe-specific immunoreactive and lectin-binding proteins are one and the same: ssp160. Conclusive evidence for this was obtained by reacting extracts of special lobe proteins with lectins bound to Sepharose. Western blots demonstrated that both GNA-and PNA-Sepharoses were capable of depleting immunoreactive ssp160 from extracts of special lobe proteins, and that the immunoreactive protein was specifically released from the lectins by incubation with cognate sugars (Fig. 8). DISCUSSION Chironomid silk proteins have been studied most extensively in C. tentans. The molecular biological data base for the major proteins is complete (6), and studies are underway to reveal how they fold and assemble into fibers (17,48). However, C. tentans evidently lacks a characteristic special lobe-specific BR and protein (18,49). The silk protein data base for species that have a special lobe (C. pallidivittatus (50) and C. thummi (51,52)) is small but growing 3 (40), but comparable data for special lobe proteins is hard to come by. Salivary glands in these species are smaller (ϳ1 mm in length), and yields of special lobe poly(A) ϩ RNA (2 ng/lobe) and silk proteins (Ͻ100 ng/lobe) are limited. This project required the manual dissection of over 4000 salivary glands. Nonetheless, a special lobe-specific cDNA has been acquired, and the encoded protein has been identified.
We conclude that 160.1 cDNA encodes the major, if not only, FIG. 6. Western blots of salivary gland proteins reacted with affinity-purified anti-fusion protein antibody. Protein from special (1) and main (2) lobes was denatured, reduced, carboxymethylated, and separated by electrophoresis on SDS-polyacrylamide gels and transferred to nitrocellulose. Blots were incubated with affinity-purified, anti-fusion protein antibody and binding was detected with an alkaline phosphatase-linked secondary antibody. A, total lobe (cells and secretion); color developed for 15 min. B, identical with A except that the affinity-purified antibody was preabsorbed with fusion protein; color developed for 40 min. C, purified secretion; color developed for 15 min. special lobe-specific silk protein in C. thummi salivary glands. Its mRNA is lobe-specific (Fig. 2) and cDNA most abundant (Fig. 1). Its gene resides in lobe-specific BRa (Fig. 3). The encoded immunoreactive protein is absent from main lobes and found in special lobes and their secretion (Fig. 6). These characteristics fit criteria defining ssp160 (20); however, there is a discrepancy between the calculated (77 kDa) and apparent (160 kDa) molecular mass of these proteins (Figs. 4 and 6). The presence of 77-kDa dimers is possible but unlikely since, for Western blotting, protein extraction included denaturation in 6 M guanidine HCl, reduction of disulfide bonds, and covalent modification of cysteinyl sulfhydryl groups. An alternative explanation is that decreased electrophoretic mobility is due to protein glycosylation. For example, N-and O-linked sugars comprise 50 -70% of the mass of some glycoproteins (53). ssp160 can incorporate label from [ 3 H]glucosamine (20), and the amino acid sequence inferred from the cDNA (Fig. 4) contains numerous sites for potential glycosylation. Lectin binding coupled with Western blotting experiments showed that the 160-kDa special lobe-specific immunoreactive protein contains both O-and N-linked sugars (Figs. 7 and 8). These results indicate that the cDNA-encoded and immunoreactive proteins are the same. To estimate how much glycosylation contributes to the mobility of this protein, we attempted enzymatic removal of sugars followed by immunoblotting with anti-fusion protein antibody to detect the resulting protein core. Despite successful removal of sugars from purified control proteins and lack of detectable proteolysis, repeated attempts to deglycosylate silk proteins were inconclusive; lectin binding was abolished, but we failed to detect an immunoreactive band. Since the bacterially expressed fusion protein could not have contained carbohydrate epitopes, the evident loss of immunoreactivity could not be due to loss of sugars per se. A more plausible explanation is that detection was hampered by the heterogeneous distribution of partially deglycosylated products. Thus, we conclude that 160.1 cDNA does, in fact, encode ssp160.
The primary structure of ssp160 is novel. A survey (54) of data bases revealed no overall similarities but some related regions. For example, ssp160's putative hydrophobic leader peptide (MNIKVILVCALVAIFFA) resembles that of Drosophila cuticle protein precursors (e.g. MFKILLCALVALVAA) (55). Sequences similar to ssp160's threonine-rich region are found in Drosophila laminin A chain (56), glutactin (57), and human ankyrin G (58). Whether or not the threonines in these proteins are glycosylated is unknown. ssp160 is unusual even among chironomid silk proteins. These proteins are composed nearly entirely of blocks of tandemly repeated sequences (6). ssp160's repeats are few, comprising only a small portion of the protein.
Since ssp160 lacks the Pro-and Cys-containing motifs characteristic of most other silk proteins, its gene would likely be of independent evolutionary descent, including limited internal sequence duplication of regions preserving the NXT motifs (Fig. 5). sp240/420 of C. tentans also lacks the Pro-and Cyscontaining motifs, is rich in serine and threonine, and is Nglycosylated (14); however, its numerous tandem repeats lack any resemblance to ssp160.
Chironomid salivary glands are dedicated to producing large amounts of a small ensemble of silk proteins. This is reflected in the cDNA banding pattern seen for both special and main lobes; several distinct cDNAs are common to both lobes, and their sizes coincide with mRNAs for small and midsize silk proteins. The amount of ssp160 cDNA (the 2.5-kb special lobespecific band in Fig. 1) is remarkable, prompting expectations of proportionally high levels of ssp160; however, Coomassie-stained gels 4 and AuroDye-stained blots (Fig. 7) indicate that neither ssp160, nor any other special lobe-specific protein, accumulates to such levels. This contrasts with other silk proteins whose steady-state level of mRNA and protein coincide (59 -61). These observations suggest that either translational control of ssp160 synthesis or its half-life in the gland differs dramatically from that of other silk proteins.
The role of the special lobe and ssp160 is uncertain. There are no reported differences in either tube-building behavior, structure of silken feeding/pupation tubes, or properties of silk in species that do and do not have special lobes. The molecular probes and antibodies acquired in this study will enable us to examine Beermann's granules for the presence of ssp160 and begin a phylogenetic investigation of the evolution and expression of ssp160-encoding genes that may lead to elucidation of the function of the special lobe.