Stabilization of short collagen-like triple helices by protein engineering

Recombinant expression of collagens and fragments of collagens is often difficult, as their biosynthesis requires specific post-translational enzymes, in particular prolyl 4-hydroxylase. Although the use of hydroxyproline-deficient variants offers one possibility to overcome this difficulty, these proteins usually differ markedly in stability when compared with the hydroxyproline-containing analogs. Here, we report a method to stabilize collagen-like peptides by fusing them to the N terminus of the bacteriophage T4 fibritin foldon domain. The isolated foldon domain and the chimeric protein (GlyProPro)(10)foldon were expressed in a soluble form in Escherichia coli. The recombinant proteins and the synthetic (ProProGly)(10) peptide were characterized by circular dichroism (CD) spectroscopy, differential scanning calorimetry, and analytical ultracentrifugation. We show that the foldon domain, which comprises only 27 amino acid residues, forms an obligatory trimer with a high degree of thermal stability. The CD thermal unfolding profiles recorded from foldon are monophasic and completely reversible upon cooling. Similar Van't Hoff and calorimertic enthalpy values of trimer formation indicated a cooperative all-or-none transition. As reported previously, (ProProGly)(10) peptides form collagen triple helices of only moderate stability. When fused to the foldon domain, however, triple helix formation of (GlyProPro)(10) is concentration independent, and the midpoint temperature of the triple helix unfolding is significantly increased. The stabilizing function of the trimeric foldon domain is explained by the close vicinity of its N termini, which induce a high local concentration in the range of 1 M for the C termini of the collagen-like-peptide. Collagen-foldon fusion proteins should be potentially useful to study receptor-collagen interactions.     

Characterization of the nucleation step and folding of a collagen triple-helix peptide

Peptide T1-892 is a triple-helical peptide designed to include two distinct domains: a C-terminal (Gly-Pro-Hyp)(4) sequence, together with an N-terminal 18-residue sequence from the alpha1(I) chain of type I collagen. Folding experiments of T1-892 using CD spectroscopy were carried out at varying concentrations and temperatures, and fitting of kinetic models to the data was used to obtain information about the folding mechanism and to derive rate constants. Proposed models include a heterogeneous population of monomers with respect to cis-trans isomerization and a third-order folding reaction from competent monomer to the triple helix. Fitting results support a nucleation domain composed of all or most of the (Gly-Pro-Hyp)(4) sequence, which must be in trans form before the monomer is competent to initiate triple-helix formation. The folding of competent monomer to a triple helix is best described by an all-or-none third-order reaction. The temperature dependence of the third-order rate constant indicates a negative activation energy and provides information about the thermodynamics of the trimerization step. These CD studies complement NMR studies carried out on the same peptide at high concentrations, illustrating how the rate-limiting folding step is affected by changes in concentration. This sequence preference of repeating Gly-Pro-Hyp units for the initiation of triple-helix formation in peptide T1-892 may be related to features in the triple-helix folding of collagens.     

Collagen triple helix formation can be nucleated at either end

The directional dependence of folding rates for rod-like macromolecules such as parallel alpha-helical coiled-coils, DNA double-helices, and collagen triple helices is largely unexplored. This is mainly due to technical difficulties in measuring rates in different directions. Folding of collagens is nucleated by trimeric non-collagenous domains. These are usually located at the COOH terminus, suggesting that triple helix folding proceeds from the COOH to the NH(2) terminus. Evidence is presented here that effective nucleation is possible at both ends of the collagen-like peptide (Gly-Pro-Pro)(10), using designed proteins in which this peptide is fused either NH(2)- or COOH-terminal to a nucleation domain, either T4-phage foldon or the disulfide knot of type III collagen. The location of the nucleation domain influences triple-helical stability, which might be explained by differences in the linker sequences and the presence or absence of repulsive charges at the carboxyl-terminal end of the triple helix. Triple helical folding rates are found to be independent of the site of nucleation and consistent with cis-trans isomerization being the rate-limiting step.     

Collagen stabilization at atomic level: crystal structure of designed (GlyProPro)10foldon

In a designed fusion protein the trimeric domain foldon from bacteriophage T4 fibritin was connected to the C terminus of the collagen model peptide (GlyProPro)(10) by a short Gly-Ser linker to facilitate formation of the three-stranded collagen triple helix. Crystal structure analysis at 2.6 A resolution revealed conformational changes within the interface of both domains compared with the structure of the isolated molecules. A striking feature is an angle of 62.5 degrees between the symmetry axis of the foldon trimer and the axis of the triple helix. The melting temperature of (GlyProPro)(10) in the designed fusion protein (GlyProPro)(10)foldon is higher than that of isolated (GlyProPro)(10,) which suggests an entropic stabilization compensating for the destabilization at the interface.     

Site-specific NMR monitoring of cis-trans isomerization in the folding of the proline-rich collagen triple helix

Understanding the folding of the proline-rich collagen triple helix requires consideration of the effects of proline cis-trans isomerization and may shed light on the misfolding of collagen in connective tissue diseases. Folding was monitored in real time by heteronuclear 2D NMR spectroscopy for the (15)N labeled positions in the triple-helical peptide T1-892 [GPAGPAGPVGPAGARGPAGPOGPOGPOGPOGV]. In the equilibrium unfolded monomer form, each labeled residue showed multiple peaks with interconversion rates consistent with cis-trans isomerization of Gly-Pro and Pro-Hyp bonds. Real-time NMR studies on the folding of T1-892 showed slow decay of monomer peaks and a concomitant increase in trimer peaks. Gly25 in the C-terminal rich (Gly-Pro-Hyp)(4) domain folds first, consistent with its being a nucleation domain. Analysis of the kinetics indicates that the folding of Gly25 is biphasic and the slower step represents cis-trans isomerization of imino acids. This illustrates that nucleation is limited by cis-trans isomerization. Monitoring Gly6, Gly10, Ala12, and Gly13 monomer and trimer peaks captures the C- to N-terminal propagation of the triple helix, which is also limited by Gly-Pro cis-trans isomerization events. The zipper-like nature of the propagation process is confirmed by the slower rate of folding of Ala6 compared to Gly13, reflecting the larger number of isomerization events encountered by the more N-terminal Ala6. The cis-trans isomerization events at multiple proline residues is a complex statistical process which can be visualized by these NMR studies.     

The NC2 Domain of Collagen IX Provides Chain Selection and Heterotrimerization

The mechanism of chain selection and trimerization of fibril-associated collagens with interrupted triple helices (FACITs) differs from that of fibrillar collagens that have special C-propeptides. We recently showed that the second carboxyl-terminal non-collagenous domain (NC2) of homotrimeric collagen XIX forms a stable trimer and substantially stabilizes a collagen triple helix attached to either end. We then hypothesized a general trimerizing role for the NC2 domain in other FACITs. Here we analyzed the NC2 domain of human heterotrimeric collagen IX, the only member of FACITs with all three chains encoded by distinct genes. Upon oxidative folding of equimolar amounts of the α1, α2, and α3 chains of NC2, a stable heterotrimer with a disulfide bridge between α1 and α3 chains is formed. Our experiments show that this heterotrimerization domain can stabilize a short triple helix attached at the carboxyl-terminal end and allows for the proper oxidation of the cystine knot of type III collagen after the short triple helix.     

Folding delay and structural perturbations caused by type IV collagen natural interruptions and nearby Gly missense mutations

The standard collagen triple helix requires Gly as every third residue in the amino acid sequence, yet all nonfibrillar collagens contain sites where this repeating pattern is interrupted. To explore the effects of such natural interruptions on the triple helix, a 4- or 15-residue sequence from human basement membrane type IV collagen was introduced between (Gly-Xaa-Yaa)(n) domains within a recombinant bacterial collagen. The interruptions had little effect on melting temperature, consistent with the high thermal stability reported for nonfibrillar collagens. Although the 4-residue interruption cannot be accommodated within a standard triple helix, trypsin and thermolysin resistance indicated a tightly packed structure. Central residues of the 15-residue interruption were protease-susceptible, whereas residues near the (Gly-Xaa-Yaa)(n) boundary were resistant, supporting a transition from an alternate conformation to a well packed triple helix. Both interruptions led to a delay in triple-helix folding, with the 15-residue interruption causing slower folding than the 4-residue interruption. These results suggest that propagation through interruptions represents a slow folding step. To clarify the relation between natural interruptions and pathological mutations, a Gly to Ser missense mutation was placed three triplets away from the 4-residue interruption. As a result of this mutation, the 4-residue interruption and nearby triple helix became susceptible to protease digestion, and an additional folding delay was observed. Because Gly missense mutations that cause disease are often located near natural interruptions, structural and folding perturbations arising from such proximity could be a factor in collagen genetic diseases.     

Collagen XXIV,a vertebrate fibrillar collagen with structural features of invertebrate collagens: selective expression in developing cornea and bone

Tissue-specific assembly of fibers composed of the major collagen types I and II depends in part on the formation of heterotypic fibrils, using the quantitatively minor collagens V and XI. Here we report the identification of a new fibrillar-like collagen chain that is related to the fibrillar alpha1(V), alpha1(XI), and alpha2(XI) collagen polypeptides and which is coexpressed with type I collagen in the developing bone and eye. The new collagen was designated the alpha1(XXIV) chain and consists of a long triple helical domain flanked by typical propeptide-like sequences. The carboxyl propeptide is classic, with 8 conserved cysteine residues. The amino-terminal peptide contains a thrombospodin-N-terminal-like (TSP) motif and a highly charged segment interspersed with several tyrosine residues, like the fibril diameter-regulating collagen chains alpha1(V) and alpha1(XI). However, a short imperfection in the triple helix makes alpha1(XXIV) unique from other chains of the vertebrate fibrillar collagen family. The triple helical interruption and additional select features in both terminal peptides are common to the fibrillar chains of invertebrate organisms. Based on these data, we propose that collagen XXIV is an ancient molecule that may contribute to the regulation of type I collagen fibrillogenesis at specific anatomical locations during fetal development.     

A peptide study of the relationship between the collagen triple-helix and amyloid

Type XXV collagen, or collagen‐like amyloidogenic component, is a component of amyloid plaques, and recent studies suggest this collagen affects amyloid fibril elongation and has a genetic association with Alzheimer's disease. The relationship between the collagen triple helix and amyloid fibrils was investigated by studying peptide models, including a very stable triple helical peptide (Pro‐Hyp‐Gly)₁₀, an amyloidogenic peptide GNNQQNY, and a hybrid peptide where the GNNQQNY sequence was incorporated between (GPO)_n domains. Circular dichroism and nuclear magnetic resonance (NMR) spectroscopy showed the GNNQQNY peptide formed a random coil structure, whereas the hybrid peptide contained a central disordered GNNQQNY region transitioning to triple‐helical ends. Light scattering confirmed the GNNQQNY peptide had a high propensity to form amyloid fibrils, whereas amyloidogenesis was delayed in the hybrid peptide. NMR data suggested the triple‐helix constraints on the GNNQQNY sequence within the hybrid peptide may disfavor the conformational change necessary for aggregation. Independent addition of a triple‐helical peptide to the GNNQQNY peptide under aggregating conditions delayed nucleation and amyloid fibril growth. The inhibition of amyloid nucleation depended on the Gly‐Xaa‐Yaa sequence and required the triple‐helix conformation. The inhibitory effect of the collagen triple‐helix on an amyloidogenic sequence, when in the same molecule or when added separately, suggests Type XXV collagen, and possibly other collagens, may play a role in regulating amyloid fibril formation. © 2012 Wiley Periodicals, Inc. Biopolymers 97: 795–806, 2012.     

The crucial role of trimerization domains in collagen folding

Collagens contain large numbers of Gly-Xaa-Yaa peptide repeats that form the characteristic triple helix, where the individual chains fold into a polyproline II helix and three of these helices form a right-handed triple helix. For the proper folding of the triple helix collagens contain trimerization domains. These domains ensure a single starting point for triple helix formation and are also responsible for the chain selection in heterotrimeric collagens. Trimerization domains are non-collagenous domains of very different structures. The size of trimerization domains varies from 35 residues in type IX collagen to around 250 residues for the fibrillar collagens. These domains are not only crucial for biological functions, but they are also attractive tools for generating recombinant collagen fragments of interest as well as for general use in protein engineering and biomaterial design. Here we review the current knowledge of the structure and function of these trimerization domains.     

Trimerization and triple helix stabilization of the collagen XIX NC2 domain

The mechanisms of chain selection and assembly of fibril-associated collagens with interrupted triple helices (FACITs) must differ from that of fibrillar collagens, since they lack the characteristic C-propeptide. We analyzed two carboxyl-terminal noncollagenous domains, NC2 and NC1, of collagen XIX as potential trimerization units and found that NC2 forms a stable trimer and substantially stabilizes a collagen triple helix attached to either end. In contrast, the NC1 domain requires formation of an adjacent collagen triple helix to form interchain disulfide bridges. The NC2 domain of collagen XIX and probably of other FACITs is responsible for chain selection and trimerization.     

Folding of collagen IV

Collagen IV dimers of two collagen IV molecules connected by their C-terminal globular NC1 domains were isolated by limited digestion with bacterial collagenase from mouse Engelbreth-Holm-Swarm (EHS) sarcoma tissue. The collagenous domains were only 300 nm long as compared to 400 nm of intact collagen IV but the disulfide bonds in the N-terminal region of the major triple helix were retained. Unfolding of the collagenous domains as monitored by circular dichroism occurred in a temperature range of 30 to 44 degrees C with a midpoint at 37 degrees C. The transition is significantly broader than that of the continuous triple helices in collagens I, II and III, a feature which can be explained by the frequent non-collagenous interruptions in the triple-helical domain of collagen IV. Refolding at 25 degrees C following complete unfolding at 50 degrees C was monitored by circular dichroism, selective proteolytic digestion of non-refolded segments and by a newly developed method in which the recovered triple-helical segments were visualized by electron microscopy. Triple-helix formation was found to proceed in a zipper-like fashion from the C-terminal NC1 domains towards the N-terminus, indicating that this domain is essential for nucleations. For collagen IV dimers with intact NC1 domains the rate of triple-helix growth was of comparable magnitude to that of collagen III, demonstrating that the non-collagenous interruptions do not slow down the refolding process where the rate-limiting step is the cis-trans isomerization of proline peptide bonds. Refolding was near to 100% and the refolding products were similar to the starting material as judged by thermal stability and electron microscopic appearance. Removal of the NC1 domains by pepsin or dissociation of their hexametric structures by acetic acid led to a loss of the refolding ability. Instead products with randomly dispersed short triple-helical segments were formed in a slow reaction. In no case, even when the disulfide bonds in the N-terminal region of the triple-helical domain were intact, was refolding from the N- towards the C-terminus observed. Taken together with results in other collagens, this suggests that C to N directionality might be an intrinsic property of triple-helix folding.     

An Activating Mutation Reveals a Second Binding Mode of the Integrin α2 I Domain to the GFOGER Motif in Collagens

The GFOGER motif in collagens (O denotes hydroxyproline) represents a high-affinity binding site for all collagen-binding integrins. Other GxOGER motifs require integrin activation for maximal binding. The E318W mutant of the integrin α2β1 I domain displays a relaxed collagen specificity, typical of an active state. E318W binds more strongly than the wild-type α2 I domain to GMOGER, and forms a 2:1 complex with a homotrimeric, collagen-like, GFOGER peptide. Crystal structure analysis of this complex reveals two E318W I domains, A and B, bound to a single triple helix. The E318W I domains are virtually identical to the collagen-bound wild-type I domain, suggesting that the E318W mutation activates the I domain by destabilising the unligated conformation. E318W I domain A interacts with two collagen chains similarly to wild-type I domain (high-affinity mode). E318W I domain B makes favourable interactions with only one collagen chain (low-affinity mode). This observation suggests that single GxOGER motifs in the heterotrimeric collagens V and IX may support binding of activated integrins.     

Mapping of SPARC/BM-40/osteonectin-binding sites on fibrillar collagens

The 33-kDa matrix protein SPARC (BM-40, osteonectin) binds several collagen types with moderate affinity. The collagen-binding site resides in helix alphaA of the extracellular calcium-binding domain of SPARC and is partially masked by helix alphaC. Previously, we found that the removal of helix alphaC caused a 10-fold increase in the affinity of SPARC for collagen, and we identified amino acids crucial for binding by site-directed mutagenesis. In this study, we used rotary shadowing, CNBr peptides, and synthetic peptides to map binding sites of SPARC onto collagens I, II, and III. Rotary shadowing and electron microscopy of SPARC-collagen complexes identified a major binding site approximately 180 nm from the C terminus of collagen. SPARC binding was also detected with lower frequency near the matrix metalloproteinase cleavage site. These data fit well with our analysis of SPARC binding to CNBr peptides, denaturation of which abolished binding, indicating triple-helical conformation of collagen to be essential. SPARC binding was substantially decreased in two of seven alpha2(I) mutant procollagen I samples and after N-acetylation of Lys/Hyl side chains in wild-type collagen. Synthetic peptides of collagen III were used to locate the binding sites, and we found SPARC binding activity in a synthetic triple-helical peptide containing the sequence GPOGPSGPRGQOGVMGFOGPKGNDGAO (where O indicates 4-hydroxyproline), with affinity for SPARC comparable with that of procollagen III. This sequence is conserved among alpha chains of collagens I, II, III, and V. In vitro collagen fibrillogenesis was delayed in the presence of SPARC, suggesting that SPARC might modulate collagen fibril assembly in vivo.     

Mutation in cyclophilin B that causes hyperelastosis cutis in American Quarter Horse does not affect peptidylprolyl cis-trans isomerase activity but shows altered cyclophilin B-protein interactions and affects collagen folding

The rate-limiting step of folding of the collagen triple helix is catalyzed by cyclophilin B (CypB). The G6R mutation in cyclophilin B found in the American Quarter Horse leads to autosomal recessive hyperelastosis cutis, also known as hereditary equine regional dermal asthenia. The mutant protein shows small structural changes in the region of the mutation at the side opposite the catalytic domain of CypB. The peptidylprolyl cis-trans isomerase activity of the mutant CypB is normal when analyzed in vitro. However, the biosynthesis of type I collagen in affected horse fibroblasts shows a delay in folding and secretion and a decrease in hydroxylysine and glucosyl-galactosyl hydroxylysine. This leads to changes in the structure of collagen fibrils in tendon, similar to those observed in P3H1 null mice. In contrast to cyclophilin B null mice, where little 3-hydroxylation was found in type I collagen, 3-hydroxylation of type I collagen in affected horses is normal. The mutation disrupts the interaction of cyclophilin B with the P-domain of calreticulin, with lysyl hydroxylase 1, and probably other proteins, such as the formation of the P3H1·CypB·cartilage-associated protein complex, resulting in less effective catalysis of the rate-limiting step in collagen folding in the rough endoplasmic reticulum.     

Designed coiled coils promote folding of a recombinant bacterial collagen

Collagen triple helices fold slowly and inefficiently, often requiring adjacent globular domains to assist this process. In the Streptococcus pyogenes collagen-like protein Scl2, a V domain predicted to be largely α-helical, occurs N-terminal to the collagen triple helix (CL). Here, we replace this natural trimerization domain with a de novo designed, hyperstable, parallel, three-stranded, α-helical coiled coil (CC), either at the N terminus (CC-CL) or the C terminus (CL-CC) of the collagen domain. CD spectra of the constructs are consistent with additivity of independently and fully folded CC and CL domains, and the proteins retain their distinctive thermal stabilities, CL at ～37 °C and CC at >90 °C. Heating the hybrid proteins to 50 °C unfolds CL, leaving CC intact, and upon cooling, the rate of CL refolding is somewhat faster for CL-CC than for CC-CL. A construct with coiled coils on both ends, CC-CL-CC, retains the ～37 °C thermal stability for CL but shows less triple helix at low temperature and less denaturation at 50 °C. Most strikingly however, in CC-CL-CC, the CL refolds slower than in either CC-CL or CL-CC by almost two orders of magnitude. We propose that a single CC promotes folding of the CL domain via nucleation and in-register growth from one end, whereas initiation and growth from both ends in CC-CL-CC results in mismatched registers that frustrate folding. Bioinformatics analysis of natural collagens lends support to this because, where present, there is generally only one coiled-coil domain close to the triple helix, and it is nearly always N-terminal to the collagen repeat.     

Changes in the synthesis of minor cartilage collagens after growth of chick chondrocytes in 5-bromo-2'-deoxyuridine or to senescence

Analyses were made of the minor collagens synthesized by cultures of chondrocytes derived from 14-day chick embryo sterna. Comparisons were made between control cultures, cultures grown for 9 days in 5-bromo-2'-deoxyuridine (BrdU) and clones of chondrocytes grown to senescence. Separation of minor collagens from interstitial collagens was achieved by differential salt precipitation in the presence of carrier collagens in acid conditions. The precipitate at 0.9 M NaCl 0.5 M acetic acid from control cultures was shown by CNBr peptide analysis to contain only the alpha 1(II) chain of type II collagen, whereas after BrdU treatment or growth to senescence synthesis of only alpha 1(I) and alpha 2(I) chains occurred. The synthesis of type III collagen was not detected. Analysis of the precipitate at 2.0 M NaCl, 0.5 M HAc from control cultures demonstrated the synthesis of 1 alpha, 2 alpha and 3 alpha chains together with the synthesis of short chain (SC) collagen of Mr 43000 after pepsin digestion. After BrdU treatment or growth to senescence alpha chains were isolated which possessed the migration positions on polyacrylamide gel electrophoresis (PAGE), or the elution positions on CM-cellulose chromatography, of the alpha 1(V) and alpha 2(V) chains of type V collagen. In addition, for BrdU-treated but not for control cultures, intracellular immunofluorescent staining was observed with a monoclonal antibody which specifically recognizes an epitope present in the triple helix of type V collagen. Synthesis of short chain (SC) collagen was not detected after BrdU treatment or growth to senescence. These results suggest that chick chondrocytes grown in conditions known to cause switching of collagen synthesis from type II to type I collagen also undergo a switch from the synthesis of 1 alpha, 2 alpha and 3 alpha chains to the synthesis of the alpha 1(V) and alpha 2(V) chains of type V collagen. It appears that there are several cartilage-specific collagens which together undergo a regulatory control to the synthesis of collagens typical of other connective tissues.     

Folding defects in fibrillar collagens

Fibrillar collagens have a long triple helix in which glycine is in every third position for more than 1000 amino acids. The three chains of these molecules are assembled with specificity into several different molecules that have tissue-specific distribution. Mutations that alter folding of either the carboxy-terminal globular peptides that direct chain association, or of the regions of the triple helix that are important for nucleation, or of the bulk of the triple helix, all result in identifiable genetic disorders in which the phenotype reflects the region of expression of the genes and their tissue-specific distribution. Mutations that result in changed amino-acid sequences in any of these regions have different effects on folding and may have different phenotypic outcomes. Substitution for glycine residues in the triple helical domains are among the most common effects of mutations, and the nature of the substituting residue and its location in the chain contribute to the effect on folding and also on the phenotype. More complex mutations, such as deletions or insertions of triple helix, also affect folding, probably because of alterations in helical pitch along the triple helix. These mutations all interfere with the ability of these molecules to form the characteristic fibrillar array in the extracellular matrix and many result in intracellular retention of abnormal molecules.     

Complete primary structure of human collagen alpha 1 (V) chain

Several cDNA clones, encoding prepropeptide of human collagen alpha 1(V) chain, have been isolated. The prepropeptide (1838 amino acids length) of the alpha 1(V) chain was composed of a putative signal peptide, a large NH2-terminal noncollagenous region, a main collagenous region, and a COOH-terminal noncollagenous region. The signal peptide contained many leucine residues. The NH2-terminal noncollagenous region was much larger than those of the other collagens and had a region homologous to the COOH-terminal domain of laminin A chain, but it did not contain a cysteine-rich region that was maintained in the region of the other collagens. This region also contained probable tyrosine sulfation sites, and short collagenous sequences that were interrupted by three noncollagenous segments. The main collagenous region of the alpha 1(V) chain consisted of 338 repeats of Gly-X-Y-triplet. This region had a high degree (82%) of homology with the amino acids of the collagen alpha 1(XI) chain. The COOH-terminal noncollagenous region resembled that of the alpha 1(XI) chain, too, and 8 residues of cysteine that were important for the formation of the triple helix structure of collagens were observed. These results suggest that the alpha 1(V) chain belongs to the fibrillar collagen relative to the alpha 1(XI) chain, but codon usage of the alpha 1(V) cDNA was clearly different from those of the other fibrillar collagens including the alpha 1(XI), while it was similar to type IV collagen. This result supposes a different evolution of the alpha 1(V) gene from those of the other fibrillar collagens.     

A short sequence in the N-terminal region is required for the trimerization of type XIII collagen and is conserved in other collagenous transmembrane proteins

The recombinant transmembrane protein type XIII collagen is shown to reside on the plasma membrane of insect cells in a 'type II' orientation. Expressions of deletion constructs showed that sequences important for the association of three alpha1(XIII) chains reside in their N- rather than C-terminal portion. In particular, a deletion of residues 63-83 immediately adjacent to the transmembrane domain abolished the formation of disulfide-bonded trimers. The results imply that nucleation of the type XIII collagen triple helix occurs at the N-terminal region and that triple helix formation proceeds from the N- to the C-terminus, in opposite orientation to that of the fibrillar collagens. Interestingly, a sequence homologous to the deleted residues was found at the same plasma membrane-adjacent location in other collagenous transmembrane proteins, suggesting that it may be a conserved association domain. The type XIII collagen was secreted into insect cell medium in low amounts, but this secretion was markedly enhanced when the cytosolic portion was lacking. The cleavage occurred in the non-collagenous NC1 domain after four arginines and was inhibited by a furin protease inhibitor.