Conserved Neuron Promoting Activity in Drosophila and Vertebrate Laminin α1

Drosophila S2 cells were transfected with constructs that code for two portions of the Drosophila laminin α chain. Construct recαL coded for domains III, I/II, and G of laminin α. Construct recαS coded for only the COOH-most 12% of the I/II domain and the G domain. The corresponding polypeptides were isolated and characterized from the culture media. The recαL chain partly formed disulfide-linked heterotrimers with the endogenously produced β and γ laminin chains. Like normal Drosophila laminin, a substrate coating of either recαL or recαS supported neuron differentiation and neurite extension of primary Drosophila embryo cell cultures. However, at the same low concentrations, only Drosophila laminin-1, but neither recαL nor recαS supported myogenesis in these cultures. Previously, an overlapping set of dodecapeptides that covered a region of the murine laminin α1 chain similar to recαS had been synthesized and tested for cell culture support properties (Nomizu, M., Kim, W. H., Yamamura, K., Utani, A., Otaka, A., Roller, P. P., Kleinman, H. K., and Yamada, Y. (1995) J. Biol. Chem. 270, 20583-20590). The Drosophila laminin α homologues of the six most active vertebrate dodecapeptides were now synthesized and tested as substrates for differentiation of primary Drosophila embryo cells. Peptides that contained either the Drosophila sequence SIKVGV or the murine homologue, SIKVAV, provided support for neurite extension.

Laminins are glycoproteins found mostly in the layers of extracellular matrix (ECM) 1 called basement membranes. These underlie epithelial cells, envelop muscles and fat cells, and are associated with the peripheral nervous system. In cell cultures, laminins facilitate adhesion, spreading, and migration of cells and form an excellent substrate for outgrowth of neurite processes. Closely related laminins have been found in the basement membranes of vertebrates and invertebrates (2), and the single laminin reported so far in Drosophila melanogaster has been characterized extensively (3)(4)(5)(6)(7)(8)(9)(10)(11). This report focusses on the actions of portions of Drosophila laminin on the in vitro differentiation of primary cell cultures obtained by dissociation of late Drosophila gastrula embryos.
Each heterotrimeric laminin molecule contains one ␣, ␤, and ␥ chain (12,13). The canonical mouse laminin-1, obtained from Engelbreth Holmes Swarm tumor (EHS) (14), consists of ␣1, ␤1, and ␥1 chains (Fig. 1a). The only Drosophila laminin molecule described so far (Fig. 1b) consists of ␣, ␤, and ␥ chains (previously called A, B1, and B2) and now denoted for specificity as D␣65, D␤28, and D␥67. Each number refers to the band of the Drosophila polytene chromosome map in which the corresponding gene occurs. This nomenclature avoids potentially erroneous assignments of Drosophila chains to vertebrate counterparts and allows for discoveries of additional Drosophila laminin chains. The three chains of each heterotrimeric laminin molecule are associated through a triple-chained coiled-coil of amphipathic ␣ helices, which forms the long arm of the cross-shaped molecule visualized by electron microscopy. The three chains are mutually disulfide-linked. The amino portion of each chain forms a separate "short arm" that protrudes from the center of the molecule. There are some significant sequence differences between these regions of the mouse ␣1 and Drosophila ␣ chains (8,10). (This Drosophila D␣65 chain is probably more closely related to the very recently described vertebrate ␣5 chain (15).) From the ends of the long arms, the COOH portions of the two ␣ chains protrude as globular G domains, each consisting of 5 homologous modules, G1-5. Major sites for cell binding of mouse laminin-1 have been ascribed to the central regions of the cruciform structure and to the COOH-terminal portion of the long arm (13, 16 -23). Portions of these sequences were expressed as recombinant proteins, and functional activity of sequences in the ␣1 chain near the peripheral end of the long arm may require reconstitution with ␤ and ␥ chains (18 -20). To examine the potential contributions of the long arm and G domain regions of the Drosophila ␣ chain, we now expressed portions of this ␣ chain in two recombinant forms rec␣L (Fig. 1c) and rec␣S (Fig. 1d). Parts of vertebrate laminin chains have been copied as synthetic peptides, which were then used for interaction studies with cells (1, 17, 21, 24 -28). Recently (1), this approach was extended to cover the whole G domain of the mouse laminin ␣1 chain as a nested set of dodecapeptides, and six peptides of particular biological interest were identified as shown in Fig. 1a. These peptides and their Drosophila laminin ␣ chain homologues (Fig. 1b) were now tested for their action in Drosophila primary cell cultures.
Laminin is one of the earliest ECM components to appear in Drosophila embryos soon after gastrulation and accumulates in increasing amounts during the embryonic differentiation and development of epithelial, neuronal, and muscular tissues (3-5, 8, 9, 11, 29, 30). Successive expression of other ECM components, such as collagen IV and tiggrin (11, 29 -31) leads to complexes with laminin that occur as electron microscopically distinct basement membranes (32,33). Therefore, cells that interact with Drosophila laminin in the embryo become successively exposed to additional matrix components.
Drosophila laminin adsorbed as a layer on microscope coverslips is an excellent substrate for the in vitro differentiation of freshly dissociated Drosophila embryo cells. This methodology was used here to test the effects of recombinant Drosophila laminin ␣ chains and synthetic peptides. As the primary cells differentiate in test culture, they produce additional ECM components (34,35). Thus, both in vivo and in vitro, laminin, or laminin analogues, initiate a chain of progressive substrate-cell interactions.
Laminin has diverse functions in Drosophila morphogenesis (9,11). LamA Ϫ mutants that lack the laminin D␣65 chain die as late embryos with relatively slight abnormalities. This suggests that other Drosophila laminin ␣ chains might exist. The Drosophila integrin PS1 binds Drosophila laminin. The site of binding is unknown and does not involve an RGD sequence (36,37). There must be additional receptor mechanisms for this laminin (11,35,38).

EXPERIMENTAL PROCEDURES
DNA Constructs and Transfection of Drosophila Cells-Two constructs, named rec␣S and rec␣L, were assembled from the partial cDNAs A54, A31, A44, A126, and A127 of the Drosophila laminin ␣ chain (7). Using the nucleotide sequence numbering of Garrison et al. (7), rec␣S encompasses nt 2410 -nt 6130 and rec␣L covers nt 63-nt 6130. The cDNAs coding for both sequences share the DraI site at nt 6130. The 5Ј end of rec␣S is the BspE1 site at nt 63, and of rec␣L is the EcoRI site nt 6130. Both 5Ј ends were blunt-ended, and NcoI linkers were added. The NcoI-DraI fragments of both cDNAs were subcloned between the NcoI and SmaI sites of a modified pBluescribe (Stratagene) vector, whose EcoRI site had been changed into an NcoI site. The constructs were excised between the NcoI site and an SphI site in the vector. These DNAs were then inserted into the expression vector pEV that had been constructed as follows. There were inserted into pBluescribe: the hsp70 promoter (EcoRI-Asp 718 ) taken from the vector pHT4 (39), followed by an AUG start codon and signal sequence taken as a 444-base pair fragment (nt 76 -520) from the Drosophila bride of sevenless (boss) gene (40), and a polyadenylation sequence taken from the pHT4 vector as a HindIII-HindIII fragment (39). The integrity of this pEV vector was checked by restriction site analysis. Then the above DNA, coding for either the rec␣L or the rec␣S sequence, was inserted into vector pEV between an NcoI site in the boss segment and an SphI site of pBluescribe between the boss and polyadenylation segments.
Drosophila S2 cells (41) were transfected by calcium phosphate coprecipitation with either the expression plasmid for rec␣S, or for rec␣L, together with the plasmid pPC4 which encodes the ␣-amanitin-resistant Drosophila RNA polymerase II gene (42). Cotransfection and the selection of ␣-amanitin-resistant cell clones were performed as described (43). After exposure of the cells to ␣-amanitin for 2 weeks, they were maintained in drug-free M3 medium containing 2% fetal calf serum at 25°C.
Production, Isolation, and Analysis of Recombinant Proteins-Transfected cells were heat-shocked at 37°C for 30 min in the presence of fetal calf serum, resuspended in serum-free M3 medium, and, after 6 h culture at 25°C, the conditioned medium was collected. Proteins that precipitated from the medium at 45% saturation with (NH 4 ) 2 SO 4 were dialyzed into PBS, pH 7.2, and adsorbed onto a Heparin-Sepharose CL-4B column (Pharmacia Biotech Inc.) that had been equilibrated with PBS. After washing with the same buffer, proteins were eluted with PBS containing 1 M NaCl and were then applied to a Superose 6 column (Pharmacia Biotech Inc.) in 0.3 M sucrose, 0.15 M NaCl, 0.1% Triton X-100, 30 mM Tris-HCl, pH 7.6. The molecular sieve column was developed with the same buffer to separate the different laminin constituents. For further purification, each set of pooled fractions was applied to a Mono Q ion exchange column (Pharmacia Biotech Inc.) in 0.3 M sucrose, 0.1% Triton X-100, 30 mM Tris-HCl, pH 7.25, and the bound proteins were eluted with a 0 -0.5 M NaCl gradient in this buffer. Prior to their use as cell culture substrates, the purified recombinant proteins were again adsorbed onto Heparin-Sepharose CL-4B and, after removing Triton X-100 detergent by extensive washing with PBS, the proteins were eluted with 1 M NaCl in PBS and then dialyzed into PBS. The normal laminin that was used as a cell culture substrate was isolated from Drosophila Kc cell culture medium as described (44). Purification steps were monitored by SDS-polyacrylamide gel electrophoresis, silver staining, and Western blots using standard procedures (45). The primary antibodies for Western blots were rabbit polyclonal antibodies that recognize all three subunits of Drosophila laminin (44), and the secondary antibodies were goat anti-rabbit IgG conjugated to alkaline phosphatase (Promega). Deglycosylation with peptide:N-glycosidase F (PNGase F) (New England Biolabs) was carried out according to the manufacturer's protocol. For electron microscopy, the isolated proteins were dissolved in 0.2 M NH 4 HCO 3 , mixed with glycerol, sprayed onto mica, and rotary-shadowed (46). Prints made at ϫ 100,000 magnification were scanned at 600 dpi with a Relisys 9624 scanner. Image thread lengths were determined with a Power MacIntosh computer and the software NIH Image version 1.58.
Test Cell Cultures-Primary cell cultures were established from D. melanogaster Oregon R post-gastrula embryos (3-4.5 h after egg laying) that had been sterilized with 1:10 diluted saponated cresol solution (Japanese Pharmacopoeia). Cells were isolated from the mechanically disrupted embryos as described and suspended in D-22 medium at 3 ϫ 10 6 cells/ml (34,35). Each autoclaved glass coverslip (Corning, 18 mm) was incubated overnight at 4°C with the specified laminin or recombinant protein coating solution, rinsed with PBS and D-22 medium, and placed in a 35-mm diameter dish (Falcon) to which 1.0 ml of cell suspension was then added.
For studies on synthetic peptide substrates, Drosophila homologues of cell biologically active mouse laminin peptides (1) were chosen from alignments of the Drosophila and mouse laminin ␣1 sequences made with the Clustal V program (public domain software of Dr. D. Higgins, European Molecular Biology Laboratory). The dodecapeptides were manually synthesized by the 9-fluorenylmethoxycarbonyl strategy and prepared in the COOH-terminal amide form as described (1). The identities of the peptides were confirmed by amino acid analysis and fast atom bombardment mass spectral analysis. 100 l of peptide solution, containing 50 g of peptide and 25 g of BSA dissolved in Milli-Q water, was layered onto a glass coverslip and allowed to dry overnight. A cell suspension was then added as above. Control experiments showed that addition of BSA suppressed nonspecific neurite extension that otherwise occurred to some extent with all highly concentrated peptides. Attachment assays of human fibrosarcoma HT-1080 cells to coatings of dodecapeptides and to mouse laminin-1, prepared from EHS tumor, were made as described previously (1). In this method, the coatings in 96-well plates are blocked with BSA and the subsequently attached cells are stained with crystal violet, washed, lysed with SDS, and the optical density at 560 nm is measured.
Cultures were incubated for the specified times or overnight at 25°C, washed, and prepared for immunofluorescence analysis as described (34). Primary antibodies were as follows. Drosophila PS integrins were stained with the monoclonal antibody CF.6G11 (47). Muscle myosin was detected with muscle-specific, rabbit polyclonal antibodies kindly provided by Dr. D. Kiehart (48). The neural cell-specific monoclonal antibody against ELAV protein was a gift of Dr. L. Zipursky. Neural cells were also detected with fluorescein-conjugated goat IgG against horseradish peroxidase (HRP) purchased from Cappel (49). The secondary antibodies were fluorescein-labeled anti-rabbit or rhodamine-labeled anti-mouse goat IgGs (Kirkegaard & Perry Laboratories). Cultures were examined with a Zeiss Axiophot microscope equipped for epifluorescence, using 100ϫ and 40ϫ oil objectives.

Production and Properties of Recombinant Laminin ␣
Chains-Two variants of the Drosophila laminin ␣ chain were expressed in Drosophila S2 cells. The smaller one, rec␣S (amino acid residues 2625-3712 of the ␣ chain) comprises the COOH terminus of the laminin ␣ chain and consists of the total G domain plus an adjacent 12% of the I/II domain that is part of the distal end of the trimeric molecule's long arm and con-tains the sequence IKVGV (Fig. 1d). A peptide corresponding to the homologous mouse ␣1 laminin sequence, IKVAV, competes with mouse laminin in cell spreading, migration, and neurite outgrowth tests (24). The larger construct, rec␣L, contains all of rec␣S and continues farther toward the NH 2 end (residues 1784 -3712) to encompass all of domain I/II and 97% of domain III (Fig. 1c). DNA coding for the Drosophila hsp 70 promoter and for a signal sequence was placed 5Ј to each construct. Drosophila S2 cells were permanently transfected with the corresponding plasmids and heat-shocked. The secreted laminin proteins were isolated from the culture medium by heparin-Sepharose affinity chromatography. Gel sieve chromatography on Superose 6 partially separated the minor amounts of endogenously produced, disulfide-linked Drosophila laminin (␣,␤,␥) from a mixture of disulfide-linked heterotrimer (rec␣L,␤,␥) and rec␣L molecules and these from smaller molecular weight contaminating proteins. Ion exchange chromatography on a Mono Q column removed the remaining contaminants and further separated the various forms of laminin (Fig. 2). Untransfected S2 host cells secrete small amounts of the normal, trimeric Drosophila laminin molecules, each consisting of disulfidelinked ␣, ␤, and ␥ chains (Fig. 2h). The successive order of elution from the Mono Q column by a 0 -0.5 M NaCl gradient was: rec␣L (Fig. 2i), then the disulfide-linked heterotrimer (rec␣L,␤,␥) (shown after reduction in Fig. 2j) followed by a mixture of this material with the endogenous Drosophila laminin heterotrimer (␣,␤,␥) (Fig. 2k) and finally the endogenous trimer alone (Fig. 2h). The rec␣S protein was isolated by the same preparative steps from correspondingly transfected cells and is demonstrated in Fig. 2g and, after reduction, in Fig. 2e. All these materials reacted with antiserum raised against endogenous laminin (3), as shown in the Western blot (Fig. 2, h-k). The rec␣L polypeptide was recovered from the conditioned medium largely as individual chains, but a small portion had also formed the disulfide-linked trimer (rec␣L,␤,␥), as demonstrated in Fig. 2j and in the silver-stained Fig. 2, a and c. Although heat shock induction had caused production of a large excess of the rec␣L chain over the resident ␣ laminin chain, the two heterotrimers (rec␣L,␤,␥) and (␣,␤,␥) were present in comparable amounts. The electrophoretic mobility of the rec␣L polypeptide fraction decreased slightly on reduction, indicating intrachain disulfide bridges and lack of interchain disulfide links (not shown). The reduced rec␣L chains migrated slightly slower than the Drosophila ␤ chain (Fig. 2, i and h). Digestion of the recombinant proteins with endoglycosidase PNGaseF caused a slight increase in SDS-polyacrylamide gel electrophoresis electrophoretic mobility, as shown for rec␣S in Fig. 2, e and f, indicating attachment of N-linked oligosaccharides to the recombinant laminin chains. All the laminin forms were bound by heparin-Sepharose.
Electron micrographs of the sprayed and rotary-shadowed laminin products are shown in Fig. 3. The appearance of the endogenous heterotrimeric laminin (Fig. 3a) agrees with that previously reported for laminin isolated from Drosophila Kc cells (3,50). One of the short arms is distinctly longer than those of the other two, and the peripheral end of the long arm carries two globules, corresponding to modules (G1-G2-G2) and (G3-G5), separated by a spacer (3,7,50). In contrast, the recombinant heterotrimer (rec␣L,␤,␥) (Fig. 3b) has one short arm which is shorter than the other two short arms, but also has the two globular G regions at the end of the long arm. The mean lengths of the long arms of these two types of heterotrimeric molecules were identical (within Ϯ6% of the thread length from the cross to the start of the nearest G globule). Images of rec␣L molecules, not combined with either ␤ or ␥ chains, were also obtained. A recurrent appearance was a long, thread-like portion, either with or without a globule of varying size at one end (Fig. 3c). The ratio of this thread length to that of the thread portion of the heterotrimeric long arms (crossintersection to proximal edge of G globules) was 0.9 Ϯ 0.1 (means of 22 images of each). It is likely that these images of rec␣L represent either homotrimers ((rec␣L) 3 ) or homodimers ((rec␣L) 2 ). The unpaired ␣-helical domain I/II of a single laminin ␣ chain would not be expected to survive as a rod-like thread through the preparation of spraying and rotary shadowing. Among the mostly globular rec␣S material, a significant number of paired dot images were seen, corresponding to the (G1-G2-G2) and (G3-G5) separated by their spacer (Fig. 3d). Primary Cell Culture on Recombinant Laminin Forms-Primary Drosophila cells, obtained by dissociation of post-gastrula embryos, differentiate into several cell types during short term culture on a laminin substrate, in the absence of added serum (34,35). We tested the ability of purified rec␣L and rec␣S to substitute for laminin in such cultures. Our available quantities of purified, heterotrimeric (rec␣L,␤,␥) were insufficient for this assay. The differentiation of primary cells into epithelia and muscle was monitored by morphological appearance and by staining with antibodies against the ␤ PS subunit of PS integrins and against muscle myosin (34). Coverslips were coated with laminin (100 g/ml), rec␣L, and rec␣S solutions. The laminin and rec␣S solutions were equimolar. On laminin substrates, clusters of epithelial cells appeared (Fig. 4a) and multinucleate myotubes formed by fusion of myoblasts (Fig.  4c). Laminin coatings made from more dilute solutions, equimolar with the rec␣L solution used for the results of Fig. 4, also had myoblast promoting activity. In contrast, on rec␣L only slight, rudimentary epithelial-like structures formed (Fig.  4b), and most of the myoblasts either remained unfused or failed to show significant elongation, although the above antigens were expressed (Fig. 4d). The results with rec␣S are not shown as they were very similar to those demonstrated for rec␣L in Fig. 4. Control coatings of bovine serum albumin (BSA) gave poor or no differentiation (not shown). Thus, rec␣L and rec␣S were unable to replace laminin as a support for the differentiation of epithelia and muscle under these conditions. In principle, a better comparison standard for these recombinant proteins would have been isolated Drosophila laminin ␣ chains, but these could not be obtained without disrupting their intra-chain disulfide bridges and their native folding.
Vertebrate neurite outgrowth is promoted by the COOH portion of mammalian laminin (16,22). Differentiation of Drosophila neurites, with extensive arborization of neurite processes, occurs in embryo primary cells cultured on a Drosophila laminin substrate (34,35). Primary embryo cell cultures were grown on substrates of rec␣L, rec␣S, and BSA and then doublestained for two neuron-specific epitopes. An early nuclear marker of neurons is the ELAV protein, which is expressed from the embryonic lethal abnormal visual system (elav) gene when neuronal precursor cells differentiate into neurons (51). A second marker was a neuron-specific, cell surface carbohydrate epitope that binds anti-horseradish peroxidase (HRP) antibodies (49).
During the initial few hours of culture, similar changes were observed in primary cultures grown on substrates of either laminin or BSA. The ELAV-positive cells increased on both. After a 3-h culture on BSA, some cells were neuron-positive as indicated by the nuclear ELAV marker (Fig. 5a), but only few of these also expressed the neuron-specific cell surface marker (Fig. 5b). The clusters of ELAV-positive cells increased in size and probably in cell number, but it was difficult to count the cells reliably within bunched clusters. After a 9-h culture on BSA, there were more ELAV-positive nuclei, (Fig. 5e), but only a subset of the rounded cells showed the neuron cell surface marker (Fig. 5f). In contrast, after a 9-h culture on laminin, ample clusters of cells expressed the nuclear ELAV marker (Fig. 5c) and strongly the anti-HRP neuron surface marker, which also highlighted initial neurite extensions (Fig. 5d). Further elongation of neurite extensions occurred in parallel with expansion of neuronal clusters, and, after 20 h of culture, long, extensive, and branching neurite projections were observed by anti-HRP antibody staining of cultures grown on laminin (Fig.  5g), rec␣L (Fig. 5h), and rec␣S (Fig. 5i). In contrast, after a 20-h culture on BSA, clusters of cells stained by anti-HRP remained rounded, without neurite processes (Fig. 5j), and had changed little from their initial, earlier appearance (Fig. 5f). At longer times, some projections eventually appeared, which we ascribe to laminin or other ECM proteins that are elaborated by the mixed primary cells themselves on prolonged culture (34). A sporadic appearance of neurite extensions on uncoated glass surfaces in these cultures was blocked by a 1-h exposure of coverslips to 5% BSA solution just before applying the primary cells. The same BSA treatment did not abolish the neurite promoting activities of the natural and recombinant Drosophila laminin proteins. We conclude that rec␣L and rec␣S support neurite differentiation and outgrowth.
Primary Cell Culture on Synthetic Peptides-Nomizu et al. (1) synthesized a set of overlapping dodecapeptides that covers that region of the mouse laminin ␣1 sequence which corresponds, approximately, to rec␣S. Table I and Fig. 1a show the six most interesting mouse peptides of this set (LAM-L, AG-10, AG-22, AG-32, AG-56, and AG-73) as judged by their effects on several vertebrate cell lines in cell attachment, spreading, inhibition of these processes by antibodies against integrin chains ␣2, ␣3, ␣6, and ␤1 and attachment to a mouse laminin substrate (1,24,28,52). A corresponding set of six Drosophila dodecapeptides was synthesized and named DOR-12, DG-10, DG-22, DG-32, DG-56, and DG-73 (Table I). Each of these peptides is the Drosophila homologue of the correspondingly paired mouse peptide. An additional dodecapeptide was made, LAM-RM. This corresponds to a mouse peptide that was found to react well with the Drosophila cells, LAM-L, except that the original order of five centrally placed amino acid residues, IKVAV, was scrambled to VVIAK (Tables I and II). Freshly prepared Drosophila embryo primary cells were seeded onto coverslips coated with a mixture of one dodecapeptide and BSA.   BSA (a, b, e, f, j). Cells were stained with antibodies against the neuron-specific nuclear protein ELAV (a, c, e), and with aHRP antibodies that specifically react with a Drosophila neuron cell surface epitope (b, d, f, g, h, i, j). Image pairs (ab, cd, ef) were double-immunostained with rhodamine and fluorescein second antibodies. Bar equals 18 m.
After 15 h of culture, the neurite outgrowths were visualized by staining with anti-HRP antibodies. Some of the results are illustrated in Fig. 6.
Under the conditions of this experiment, the best neurite outgrowths were shown on the Drosophila peptides DOR-12 (Fig. 6a) and DG-22 (Fig. 6d). For the corresponding mouse peptide homologues, LAM-L (Fig. 6b) had activity, but the AG-22 peptide had practically none (Fig. 6e). The scrambled mouse peptide, LAM-RM, showed very little activity (Fig. 6c) and had been found to be inactive in mammalian cell attachment tests (28). While the peptides that supported neurite outgrowth clearly differed from those that failed to do so, it was not possible to devise meaningful finer, quantitative comparisons between their actions. Only a portion of these mixed, primary embryo cells were neurons, the number of neurons in brightly fluorescent clusters could not be determined precisely and a variety of neurite processes projected from cell groups. We conclude that the cell biological activity originally described for the mouse sequence IKVAV, that occurs in the peptide LAM-L, is also active in promoting neurite outgrowth in Drosophila primary embryo cell cultures, as is the Drosophila homologous sequence IKVGV of the peptide DOR-12. Correspondingly, this Drosophila IKVGV sequence had some cell binding action on mammalian cells, as is shown in Fig. 7. To some extent this mammalian HT1080 cell line attached to the two Drosophila peptides DOR-12 and DG-22, but this effect was distinctly weaker than binding to the homologous mouse peptides, respectively, LAM-L and AG-22. Thus, the sequence similarity of the mouse IKVAV and the Drosophila IKVGV sequences is mirrored by functional homology. These sequences occur embedded in dodecapeptides that have 50% amino acid identity (Tables I and II), suggesting evolutionary conservation.
In previous tests of the mouse dodecapeptides on mammalian cells, the peptide AG-73 showed high activity (1). This  Table I  peptide also partly promoted Drosophila neurite outgrowth in the presence of BSA. Yet its Drosophila positional homologue, the dodecapeptide DG-73, was inactive. Although these two dodecapeptides have 33% amino acid identity, the core sequence LQVQLSIR of AG-73, that had been established to be essential for its action on mammalian cells (1), only shares one amino acid residue with the Drosophila peptide DG-73. Thus, while the Drosophila cells can respond to the mammalian amino acid sequence to some extent, this portion of the Drosophila laminin ␣ chain presumably evolved differently from its mouse homologue. All additional Drosophila dodecapeptides listed in Table I failed to promote Drosophila neurite outgrowth in the presence of BSA. DISCUSSION The portions of the Drosophila laminin ␣ chain that were expressed as rec␣L and rec␣S in Drosophila cells are likely to have been normally glycosylated, folded and disulfide-linked, and secreted. This is suggested particularly by the electron microscopic appearance of the heterotrimer (rec␣L,␤,␥): a long arm of the same length as normal Drosophila laminin, two globular G domains, two short arms, and one very short arm. Similarly to normal Drosophila laminin, all the laminin constructs were retained by heparin-Sepharose. The rec␣L chain was recognized by the ␤ and ␥ chains, became interchain disulfide-linked with them, and was normally secreted. However, although the rec␣L chain was produced in considerable excess over the resident laminin ␣ chain, comparable amounts of the trimers (rec␣L,␤,␥) and (␣,␤,␥) formed. Perhaps the ␣ chain is better suited for trimer formation than the rec␣L chain, but, more likely, a substantial portion of the rec␣L chains formed homodimers or trimers and thereby became unavailable for heterotrimer formation.
Partial association of rec␣L chains into homomeric, coiledcoil amphipathic helices is suggested by the electron microscopic finding of thread-like molecules with a globular end, as a thread consisting of a single ␣-helically coiled polypeptide is rarely visualized (53). As the rec␣L threads are only 0.9 Ϯ 0.1 times the length of the normal long arm of laminin and have variable terminal knobs, there is probably some collapse of structure at the ends of the putative homomeric associations of the rec␣L chain, as compared with (rec␣L,␤,␥) heterotrimers. Molecular sieve chromatography of rec␣L also suggests association in solution. The heterotrimer (rec␣L,␤,␥) co-eluted with rec␣L, and this mixture eluted as somewhat smaller material than the native, heterotrimeric laminin (␣,␤,␥) of M r ϭ 784,000. One would not expect a single rec␣L chain (M r ϭ 210,000) to have a molecular sieving behavior that closely resembles its heterotrimer with the paired ␤ and ␥ chains (M r ϭ 585,000). On a calibrated Sepharose 6 column, the heterotrimer (rec␣L,␤,␥) eluted ahead of the globular-plus-arms 510-kDa protein peroxidasin (44). The rod-shaped, coiled-coil stabilized forms of rec␣L are calculated to have M r ϭ 420,000 for the homodimer or M r ϭ 660,000 for the homotrimer. While coiled-coil interactions of vertebrate laminin ␣1 chains are opposed by repulsive electrical charges (54), the different charge distribution along the I/II domain of Drosophila laminin ␣65 would probably not hinder homomeric association. 2 The G-domain regions of rec␣S (M r ϭ 110,000) folded into a disulfide-linked structure with the appropriate biglobular electron microscopic appearance (Fig. 3) (3, 50) and eluted from Sepharose 6 near the mostly globular protein glutactin (M r ϭ 116,000). A coating of rec␣S protein supported neuron differentiation in primary embryo cell culture, as judged by the appearance of two neuron-specific epitope markers and the extension of neurites. An additional coating of BSA did not abolish this action of our recombinant protein, but did block spurious neurite extension on uncoated glass surfaces. Correspondingly, the larger recombinant protein rec␣L, which contains the sequence of rec␣S, also supported this cell differentiation. This contrasts with reports of the support of cell attachment and differentiation of the proteolytic fragment E8 of mouse laminin-1 (16,20) and of the action of a recombinant form of the vertebrate G-domain (19). These investigators concluded that only laminin-1 fragments that also contained portions of the laminin ␤1 and ␥1 chains retained such cell biological functions of the complete laminin-1 molecule. Our peptide studies indicate that the neurite promoting action of the rec␣S protein will be associated, at least in part, with the SIKVGV sequence that is contained within the 12% of the Drosophila laminin ␣ domain I/II at the NH 2 end of rec␣S. This NH 2 terminus is unlikely to be stabilized by coiled-coil ␣-helical interactions as molecular sieve criteria indicate rec␣S to be monomeric.
Neither myogenesis nor the differentiation of epithelia from primary cells were supported by either the rec␣S or the rec␣L substrates at the concentrations that were available from these purified recombinant materials. The molarity of the most concentrated rec␣S solution that could be tested for coating was 5-fold greater than the most dilute Drosophila laminin solution that supported differentiation. We do not know whether the different materials adsorbed comparably onto the glass surfaces. Although the coatings of rec␣L functioned as supports for neurite extension, their concentrations may have been inadequate for myogenesis. While the nominal molarity of the rec␣L solution that was used for coating equalled that of a laminin solution that provided an adequate coating, the effective molarity of the rec␣L solution would have been decreased by chain association into homodimers or homotrimers. Our limited negative evidence suggests that the region of Drosophila laminin represented by rec␣S may by itself be insufficient for the support of myogenesis.
There exists a striking parallelism between results with homologous peptides of the form IKV(G/A)V derived from the Drosophila and mouse laminin ␣1 chains. In the tests of neuronal differentiation and neurite extension, the two Drosophila dodecapeptides which had the best, equally good, effects on Drosophila primary cells were AnSIKVgVnfkp (DOR-12) and peptide DG-22. The murine dodecapeptides that had the most pronounced effect on vertebrate cells were AaSIKVaVsadr (LAM-L) (28, 52) and peptide AG-73 (1). While peptides DG-22 and AG-73 are unrelated, the dodecapeptides AnSIKVgVnfkp (DOR-12) and AaSIKVaVsadr (LAM-L) have 50% amino acid 2 K. Beck, personal communication. identity and occupy homologous positions in the Drosophila and murine laminin ␣1 chains (Table II). It is noteworthy that these sequences are part of a larger region that is strongly conserved in the vertebrate laminin ␣1 and ␣2 chains and retains homology in ␣3, ␣4, and ␣5 (Table II). Furthermore, over this larger region there is 100% amino acid conservation for the respective murine and human laminin ␣1 and ␣2 chains. The dodecapeptide AaSIKVaVsadr (LAM-L) was also the only murine peptide that had some action on the Drosophila cells. Rearrangement of only the central, IKVaV portion of this sequence to VVIaK in dodecapeptide LAM-RM (Table II) abolished this activity (Fig. 6, b and c). In the reciprocal test, the Drosophila dodecapeptides ANSIKVGVNFKP (DOR-12) and DG-22 had weak binding actions on vertebrate cells (Fig.  7). The murine dodecapeptide AaSIKVaVsadr (LAM-L) promoted neurite extension of mammalian PC12 cells, as did the Drosophila dodecapeptides ANSIKVGVNFKP (DOR-12) (52). The failure of action on Drosophila cells of the murine peptides AG-22 and AG-73 is directly correlated with inadequate amino acid identity between these murine and Drosophila amino acid sequences in regions known to be important for action on murine cells (1). The progressive differentiation of various cell types in Drosophila primary embryo cell cultures mirrors in vivo processes, as instanced by successive changes in gene expression and morphology (29,34,35). Our tests of these cultures on substrates prepared from dilute solutions of Drosophila rec␣L and rec␣S proteins are extensions of these studies of normal processes. In contrast, the deposits of synthetic peptides at high concentrations provide pharmacological tests of cell responses. Yet the positive evidence of both types of results suggests that the evolutionary retention of the A(N/A)SIKV(G/A)V sequence in the mammalian and Drosophila laminin ␣ chains is related to a shared function of interaction with cells, as shown particularly in the processes of neurite extension. The rec␣L and rec␣S proteins promote neurite outgrowth, and both contain the sequence ARQLANSIKVGV. In addition, peptides containing the Drosophila and mouse versions of the sequence were unique in their ability to promote adhesion of both Drosophila and mammalian cells. While we must interpret these results with caution, we find the correlation between results obtained with these peptides and with laminin fragments striking. Other sequences will contribute to the adhesion of cells to laminin ␣ chains, yet it would appear that an evolutionarily conserved role of laminin in development and differentiation is mediated, at least in part, by these conserved sequences.