Phenotypic spectrum of epidermolysis bullosa associated with α6ß4 integrin mutations.

BACKGROUND: Integrin α6ß4 is a transmembrane receptor and a key component of the hemidesmosome anchoring complex. It is involved in cell-matrix adhesion and signalling in various tissues. Mutations in the ITGA6 and ITGB4 genes coding for α6ß4 integrin compromise dermal-epidermal adhesion and are associated with skin blistering and pyloric atresia (PA), a disorder known as epidermolysis bullosa with PA (EB-PA). OBJECTIVES: To elucidate the molecular pathology of skin fragility in eight cases, disclose the underlying ITGA6 and ITGB4 mutations and study genotype-phenotype correlations. METHODS: DNA was isolated from ethylenediaminetetraacetic acid-blood samples, and the coding exons and exon-intron boundaries of ITGA6 and ITGB4 were amplified by polymerase chain reaction (PCR), and directly sequenced. Skin samples were submitted to immunofluorescence mapping with antibodies to adhesion proteins of the dermal-epidermal junction. Primary keratinocytes were isolated, and used for RNA and protein extraction, reverse transcription PCR and immunoblotting. Ultrastructural analysis of the skin was performed in one patient. RESULTS: We disclose 10 novel mutations, one in ITGA6 and nine in ITGB4. Skin cleavage was either intraepidermal or junctional. Lethal outcome and PA correlated with loss-of-function mutations in two cases. Solely mild skin involvement was associated with deletion of the C-terminus of ß4 integrin. Combinations of missense, nonsense or frameshift mutations caused severe urinary tract involvement in addition to skin fragility in five cases. CONCLUSIONS: The present study reveals novel ITGA6 and ITGB4 gene mutations and supports previous reports showing that the phenotype may lack PA and be limited to skin and nail involvement. In four out of six cases of EB-PA, life expectancy was not impaired. A high frequency of urinary tract involvement was found in this study, and represented the main cause of morbidity. Low levels of ß4 integrin expression were compatible with hemidesmosomal integrity and a mild skin phenotype.

Collagen II containing a Cys substitution for Arg-alpha1-519: abnormal interactions of the mutated molecules with collagen IX.

Single amino acid substitutions in collagen II cause heterogeneous cartilage disorders including some chondrodysplasias and certain forms of heritable osteoarthritis. In this study, we examined molecular interactions between normal collagen II and collagen IX, and the effect of a Cys substitution for Arg-alpha1-519 in collagen II on these interactions. Binding assays showed that the association equilibrium constant of collagen IX-collagen II interaction is 15 x 10(6) M(-1). Specificity of the interaction was analyzed by the binding of collagen IX to recombinant collagen II variants lacking fragments of 234 amino acids corresponding to particular D-periods. The results indicated that the C-terminal half of collagen II, which includes the D3 and D4 periods, has a high affinity for collagen IX, and that the nontriple helical telopeptides of collagen II are not essential for the specific binding of collagen IX. Computer analysis of the surface of the mutated collagen II and binding assays showed that a Cys substitution for Arg-alpha1-519 changes electrostatic properties around the mutation site, increases the affinity of mutant collagen II for collagen IX, and possibly alters the specificity of the interaction. Thus, the results indicate that interactions between collagen II and collagen IX are site specific and that single amino acid substitutions in collagen II may change the molecular interactions with collagen IX that could destabilize the cartilaginous matrix.

Papilin in development; a pericellular protein with a homology to the ADAMTS metalloproteinases.

Papilin is an extracellular matrix glycoprotein that we have found to be involved in, (1) thin matrix layers during gastrulation, (2) matrix associated with wandering, phagocytic hemocytes, (3) basement membranes and (4) space-filling matrix during Drosophila development. Determination of its cDNA sequence led to the identification of Caenorhabditis and mammalian papilins. A distinctly conserved 'papilin cassette' of domains at the amino-end of papilins is also the carboxyl-end of the ADAMTS subgroup of secreted, matrix-associated metalloproteinases; this cassette contains one thrombospondin type 1 (TSR) domain, a specific cysteine-rich domain and several partial TSR domains. In vitro, papilin non-competitively inhibits procollagen N-proteinase, an ADAMTS metalloproteinase. Inhibiting papilin synthesis in Drosophila or Caenorhabditis causes defective cell arrangements and embryonic death. Ectopic expression of papilin in Drosophila causes lethal abnormalities in muscle, Malpighian tubule and trachea formation. We suggest that papilin influences cell rearrangements and may modulate metalloproteinases during organogenesis.

Structure and function of procollagen C-proteinase (mTolloid) domains determined by protease digestion, circular dichroism, binding to procollagen type I, and computer modeling.

Procollagen C-proteinase-2 (pCP-2, mTld) is derived from the longest splicing variant of the gene encoding bone morphogenetic protein 1 (BMP-1). The variants have identical amino terminal signal peptides, prodomains and astacin-like protease domains. However, they differ in the length of their carboxy terminal part, which in pCP-2 has the composition CUB1, CUB2, EGF-like1, CUB3, EGF-like2, CUB4, CUB5, and C-tail. In the shorter form, pCP-1 (i.e., BMP-1), the sequence ends after the CUB3-domain. Using a combination of mutagenesis and structural approaches, we have investigated the structure and function of subfragments of pCP-2. The full-length latent recombinant enzyme and its N-terminally truncated form lacking the prodomain were tested for their enzymic activity. The intact protein showed only partial processing of procollagen type I, whereas the truncated form expressed enzymic activity indistinguishable from its native counterpart purified from chick embryo tendons. These results clearly demonstrated that the prodomain is required for the latency of the enzyme but not for its correct folding. Limited proteolysis of the recombinant protein with alpha-chymotrypsin produced four discrete fragments revealing the location of cleavage sites between the repetitive CUB/EGF domains. The results provide evidence that the CUB sequences form independently folded modules that are stabilized by two pairs of internal disulfide bridges. The modules are linked to each other by more flexible, hinge-like peptides. Solid-phase binding assays with isolated CUB domains and immobilized procollagen type I demonstrated that the first three but not the last two CUB domains specifically bound to the substrate. To define putative sites for CUB-CUB or CUB-substrate interactions, we generated molecular models for pCP-2 CUB domains. The models were obtained using as a template the structure of CUB domain in zona pellucida adhesion protein PSP-I/PSP-II from porcine sperm. The predicted conformations for homology models were, subsequently, confirmed by circular dichroism spectroscopy of polypeptide domains isolated following limited proteolysis with alpha-chymotrypsin.

Human Ehlers-Danlos syndrome type VII C and bovine dermatosparaxis are caused by mutations in the procollagen I N-proteinase gene.

Ehlers-Danlos syndrome (EDS) type VIIC is a recessively inherited connective-tissue disorder, characterized by extreme skin fragility, characteristic facies, joint laxity, droopy skin, umbilical hernia, and blue sclera. Like the animal model dermatosparaxis, EDS type VIIC results from the absence of activity of procollagen I N-proteinase (pNPI), the enzyme that excises the N-propeptide of type I and type II procollagens. The pNPI enzyme is a metalloproteinase containing properdin repeats and a cysteine-rich domain with similarities to the disintegrin domain of reprolysins. We used bovine cDNA to isolate human pNPI. The human enzyme exists in two forms: a long version similar to the bovine enzyme and a short version that contains the Zn++-binding catalytic site but lacks the entire C-terminal domain in which the properdin repeats are located. We have identified the mutations that cause EDS type VIIC in the six known affected human individuals and also in one strain of dermatosparactic calf. Five of the individuals with EDS type VIIC were homozygous for a C-->T transition that results in a premature termination codon, Q225X. Four of these five patients were homozygous at three downstream polymorphic sites. The sixth patient was homozygous for a different transition that results in a premature termination codon, W795X. In the dermatosparactic calf, the mutation is a 17-bp deletion that changes the reading frame of the message. These data provide direct evidence that EDS type VIIC and dermatosparaxis result from mutations in the pNPI gene.

Recombinant procollagen II: Deletion of D period segments identifies sequences that are required for helix stabilization and generates a temperature-sensitive N-proteinase cleavage site.

A cDNA cassette system was used to synthesize recombinant versions of procollagen II in which one of the four blocks of 234 amino acids that define a repeating D periods of the collagen triple helix were deleted. All the proteins were triple helical and all underwent a helix-to-coil transition between 25 and 42 degreesC as assayed by circular dichroism. However, the details of the melting curves varied. The procollagen lacking the D1 period unfolded 3 degreesC lower than a full-length molecule. With the procollagen lacking the D4 period, the first 25% of unfolding occurred at a lower temperature than the full-length molecule, but the rest of the structure unfolded at the same temperature. With the procollagen lacking the terminal D0.4 period, the protein unfolded 3 degreesC lower than the full-length molecule and a smaller fraction of the protein was secreted by stably transfected clones than with the other recombinant procollagens. The results confirmed previous suggestions that the collagen triple helix contains regions of varying stability and they demonstrated that the two D periods at the end of the molecule contain sequences that serve as clamps for folding and for stabilizing the triple helix. Reaction of the recombinant procollagens with procollagen N-proteinase indicated that in the procollagen lacking the sequences, the D1 period assumed an unusual temperature-sensitive conformation at 35 degreesC that allowed cleavage at an otherwise resistant Gly-Ala bond between residues 394 and 395 of the alpha1(II) chain.

[Disturbance of the normal cell cycle by T antigens of SV40 viruses and JC cause some tumors]. / Zaburzenia prawidlowego przebiegu cyklu komórkowego przez antygeny T wirusów SV40 i JC przyczyna niektórych nowotworów.

The mechanism that may lead to tumor formation in SV40 or JCV infected tissues based on previously reported interactions between T antigens and two factors (p53 and pRb) controlling cell cycle has been discussed. p53 is a known tumor suppressor protein that delays S phase entry causing cell arrest in G1 phase or apoptosis when the DNA damage occurs. Phosphorylation of pRB releases E2F family proteins that activate genes encoding S phase promoting factors. Binding of T antigens to pRB induces tumor formation, whereas tumor proliferation requires knockout of p53 activity.

Procollagen N-proteinase and procollagen C-proteinase. Two unusual metalloproteinases that are essential for procollagen processing probably have important roles in development and cell signaling.

As soon as procollagen precursors of fibrillar collagens were discovered in the early 1970s, it became apparent that connective tissues must contain proteolytic activities that cleave the N-propeptides and the C-propeptides from procollagens. Isolation and characterization of the enzymic activities, however, proved to be unexpectedly difficult. Both proteinases are large and are synthesized in several different forms with polypeptide chains ranging in size from 70 kDa to about 130 kDa. The N-proteinase has the unusual property of cleaving the N-propeptides from type I and type II procollagens if the proteins are in a native conformation, but not if the proteins are partially unfolded so that the N-telopeptides are no longer in a hair-pin configuration. The C-proteinase specifically cleaves native and denatured types I, II and III procollagens. It also specifically cleaves a precursor of lysyl oxidase and laminin 5. Both enzymes and their variants have structures that place them in a large and expanding super-family of over 200 zinc-binding metalloproteinases. The larger of two forms of the N-proteinase contains an RGD sequence for binding through integrins and properdin repeats similar to those found in thrombospondin. The shorter 70 kDa form of the C-proteinase is identical to the protein that was previously identified as bone morphogenic protein-1. Both the 70 kDa C-proteinase and two larger forms are homologous to proteins that are expressed early in development in a variety of organisms, including Drosophila, sea urchin, and fish. Therefore, the data suggest that both the N- and C-proteinases have important biological functions in addition to the roles in the processing of procollagens.

Synthesis and conformational properties of a recombinant C-propeptide of human type III procollagen.

A cDNA was prepared that coded for the signal peptide of type III procollagen linked to the complete C-propeptide of the protein. The cDNA was then used to express the protein in a baculovirus recombinant system. Recombinant protein was recovered as a trimer from the medium of transfected cells in a yield of 1 to 2.5 mg per liter. Mapping of peptide fragments with and without reduction indicated that the protein contained the expected interchain disulfide bonds. Analysis by circular dichroism suggested that the conformation of the protein corresponded to the native conformation. Therefore, the protein should be appropriate for further tests of its biological function and analysis of structure by X-ray diffraction.

A cDNA cassette system for the synthesis of recombinant procollagens. Variants of procollagen II lacking a D-period are secreted as triple-helical monomers.

Currently there is a lack of experimental systems for defining the functional domains of the fibrillar collagens. Here we describe an experimental strategy that employs the polymerase chain reaction (PCR) to create a series of cDNA cassettes coding for seven separate domains of procollagen II. The system was used to prepare novel recombinant procollagens II from which one of the four repetitive D-periods of the triple helix was deleted. Four constructs, each lacking a different D-period, were expressed in stably transfected mammalian cells (HT-1080). Truncated procollagens of the predicted size were recovered from the medium. All were triple-helical as assayed by circular dichroism. Therefore, deletion of a complete D-period containing 234 amino acids does not destabilize the triple helix of homotrimeric collagen II as much as some naturally occurring mutations in the heterotrimeric monomer of collagen I that delete shorter sequences or that convert obligate glycine residues to residues with bulkier side chains. Moreover, the results suggest that the strategy developed here can be used to map in detail the binding sites on fibrillar collagens for other components of the extracellular matrix and for the binding, spreading and signaling of cells.

cDNA cloning and expression of bovine procollagen I N-proteinase: a new member of the superfamily of zinc-metalloproteinases with binding sites for cells and other matrix components.

Procollagen N-proteinase (EC 3.4.24.14) cleaves the amino-propeptides in the processing of type I and type II procollagens to collagens. Deficiencies of the enzyme cause dermatosparaxis in cattle and sheep, and they cause type VIIC Ehlers-Danlos syndrome in humans, heritable disorders characterized by accumulation of pNcollagen and severe skin fragility. Amino acid sequences for the N-proteinase were used to obtain cDNAs from bovine skin. Three overlapping cDNAs had an ORF coding for a protein of 1205 residues. Mammalian cells stably transfected with a complete cDNA secreted an active recombinant enzyme that specifically cleaved type I procollagen. The protein contained zinc-binding sequences of the clan MB of metallopeptidases that includes procollagen C-proteinase/BMP-1. The protein also contained four repeats that are homologous to domains found in thrombospondins and in properdin and that can participate in complex intermolecular interactions such as activation of latent forms of transforming growth factor beta or the binding to sulfatides. Therefore, the enzyme may play a role in development that is independent of its role in collagen biosynthesis. This hypothesis was supported by the observation that in some tissues the levels of mRNA for the enzyme are disproportionately high relative to the apparent rate of collagen biosynthesis.

A recombinant homotrimer of type I procollagen that lacks the central two D-periods. The thermal stability of the triple helix is decreased by 2 to 4 degrees C.

A D-period cassette system was developed that can be used to synthesize a variety of recombinant homotrimers of type I procollagen. A construct lacking the central two D-periods of pro alpha 1(I) chains was assembled and expressed as a recombinant protein in the mammalian cell line. The recombinant protein was purified to homogeneity and the thermal stability of the triple helix assayed by rapid protease digestion. The results indicated that deletion of the central 468 amino acids from the major triple helix lowered the thermal stability of the protein by 2 to 4 degrees C. The results therefore begin to define regions of the molecule that vary in their contributions to helical stability.

Role of the V2, V3, and CD4-binding domains of GP120 in curdlan sulfate neutralization sensitivity of HIV-1 during infection of T lymphocytes.

A sulfated polysaccharide, curdlan sulfate (CRDS) with 1,3-beta-D-glucan as a main chain, inhibits HIV-1 infection of human peripheral blood lymphocytes (PBLs) by binding to the V3 region of gp 120. We previously showed that T cell (T)-tropic HIV-1 isolates are over 10-fold more sensitive to neutralization by CRDS than macrophage (MT)-tropic viruses, which possesses a relatively less charged amino acid composition in the V3 sequence. To analyze the interaction of CRDS with V3 and its association with neutralization sensitivity of HIV-1 isolates, we examined the effect of CRDS on the binding of neutralizing antibodies to monomeric and oligomeric gp 120 mutants of T- and MT-tropic HIV-1 clones in which the V3 loop was either deleted or substituted by V3 of another isolate. Our results showed that the presence and the amino acid composition of the V3 loop appears to determine the extent of interaction of CRDS with the V2 and CD4-binding regions on native gp 120 monomers; however, the positive charge of V3 has less effect on this interaction on oligomeric gp 120. Furthermore, our results established that only the CRDS-induced masking of V3 on oligomeric gp120 appears to be associated with the anti-HIV-1 activity of CRDS in vitro. Our findings underline the usefulness of CRDS for understanding the structural constraints on gp 120 that drive the transition from MT- to T-tropic isolates in vivo and enable the virus to use multiple fusion cofactors.

Assembly in vitro of thin and thick fibrils of collagen II from recombinant procollagen II. The monomers in the tips of thick fibrils have the opposite orientation from monomers in the growing tips of collagen I fibrils.

Human type II procollagen was prepared in a recombinant system and cleaved to pC-collagen II by procollagen N-proteinase. The pC-collagen II was then used as a substrate to generate collagen II fibrils by cleavage with procollagen C-proteinase at 37 degrees C. Electron microscopy of the fibrils demonstrated that, at the early stages of fibril assembly, very thin fibrils were formed. As the system approached equilibrium over 7-12 h, however, the thin fibrils were largely but not completely replaced by thick fibrils that had diameters of about 240 nm and a distinct D-period banding pattern. One typical fibril was photographed and analyzed in its entirety. The fibril was 776 D-periods (52 microM) long. It had a central shaft with a uniform diameter that was about 516 D-periods long and two tips of about 100 D-periods each. Most of the central shaft had a symmetrical banding pattern flanked by two transition regions of about 30 D-periods each. Measurements by scanning transmission electron microscopy demonstrated that the mass per unit length from the tips to the shafts increased linearly over approximately 100 D-periods from the fibril end. The linear increase in mass per unit length was consistent with previous observations for collagen I fibrils and established that the tips of collagen II also had a near paraboloidal shape. However, the orientation of monomers in the tips differed from the tips of collagen I fibrils in that the C termini instead of the N termini were directed toward the tips. The thin fibrils that were present at early stages of assembly and at equilibrium were comparable to the collagen II fibrils seen in embryonic tissues and probably represented intermediates on the pathway of thick fibrils formation. The results indicated that the molecular events in the self-assembly of collagen II fibrils are apparently similar to those in self-assembly of collagen I fibrils, but that there are also important differences in the structural information contained in collagen I and collagen II monomers.

The C-proteinase that processes procollagens to fibrillar collagens is identical to the protein previously identified as bone morphogenic protein-1.

Bone morphogenic protein-1 (BMP-1) was originally identified as one of several BMPs that induced new bone formation when implanted into ectopic sites in rodents. BMP-1, however, differed from other BMPs in that it its structure was not similar to transforming growth factor beta. Instead, it had a large domain homologous to a metalloendopeptidase isolated from crayfish, an epidermal growth-factor-like domain, and three regions of internal sequence homology referred to as CUB domains. Therefore, BMP-1 was a member of the "astacin families" of zinc-requiring endopeptidases. Many astacins have been shown to play critical roles in embryonic hatching, dorsal/ventral patterning, and early developmental decisions. Here, we have obtained amino acid sequences and isolated cDNA clones for procollagen C-proteinase (EC 3.4.24.19), an enzyme that is essential for the processing of procollagens to fibrillar collagens. The results demonstrate that procollagen C-proteinase is identical to BMP-1.

Targeted insertions of two exogenous collagen genes into both alleles of their endogenous loci in cultured human cells: the insertions are directed by relatively short fragments containing the promoters and the 5' ends of the genes.

Previous studies demonstrated that type II procollagen is synthesized by HT-1080 cells that are stably transfected with constructs of the human COL2A1 gene that contain the promoter and 5' end of either the COL2A1 gene or the human COL1A1 gene. Since the host HT-1080 cells were from a human tumor line that synthesizes type IV collagen but not type II or type I procollagen, the results suggested that the constructs were integrated near active enhancers or promoters. Here, however, we demonstrate that a 33-kb construct of the COL2A1 gene containing a 5' fragment from the same gene was inserted into both alleles of the endogenous COL2A1 gene on chromosome 12, apparently by homologous recombination by a nonconservative pathway. In contrast, a similar construct of the COL2A1 gene in which the 5' end was replaced with a 1.9-kb fragment from the 5' end of the COL1A1 gene was inserted into both alleles of the locus for the COL1A1 gene on chromosome 17. Therefore, targeted insertion of the gene construct was not directed by the degree of sequence homology. Instead, it was directed by the relatively short 5' fragment from the COL1A1 gene that contained the promoter and the initially transcribed sequences of the gene. After insertion, both gene constructs were expressed from previously inactive loci.

Self-assembly into fibrils of collagen II by enzymic cleavage of recombinant procollagen II. Lag period, critical concentration, and morphology of fibrils differ from collagen I.

A recently developed recombinant system for synthesis of human procollagen II by stably transfected host cells was used to prepare adequate amounts of protein to study the self-assembly of collagen II into fibrils. The procollagen II was cleaved to pCcollagen II by procollagen N-proteinase (EC 3.4.24.14), the pCcollagen II was chromatographically purified, and the pCcollagen II was then used as a substrate to generate collagen II fibrils by cleavage with procollagen C-proteinase. The kinetics for assembly of collagen II fibrils were similar to those observed previously for the self-assembly of collagen I in that a distinct lag phase was observed followed by a sigmoidal propagation phase. However, under the same experimental conditions, the lag time for assembly of collagen II fibrils was 5-6-fold longer, and the propagation rate for collagen II fibrils was about 30-fold lower than for collagen I fibrils. The relatively long lag time for the assembly of collagen II into fibrils made it possible to demonstrate that most of the conversion of pCcollagen II to collagen II occurred in the solution phase. The critical concentration at 37 degrees C for collagen II was about 50-fold greater than the critical concentration for collagen I. The Gibbs free energy change for the assembly of collagen II into fibrils was -40 kJ/mol, a value that was about 14 kJ/mol less than the free energy change for collagen I and about the same as the free energy change for the homotrimer of collagen I. Dark-field light microscopy and negative-staining electron microscopy demonstrated that the collagen II fibrils were thin and formed network-like structures. The results demonstrated, therefore, that the structural information of the monomer is sufficient to explain the characteristically small diameters and arcade-like geometry of collagen II fibrils found in cartilage and other tissues.

Synthesis of recombinant human procollagen II in a stably transfected tumour cell line (HT1080).

Apparently because the biosynthetic pathways involve eight or more highly specific post-translational enzymes, it has been difficult to obtain expression of genes for fibrillar collagens in recombinant systems. Here two constructs of the human gene for procollagen II (COL2A1) were prepared, one with about 0.5 kb of a promoter for a procollagen I gene (COL1A1) and the other with about 4 kb of the promoter for the procollagen II gene. The constructs, together with a neomycin-resistant gene, were transfected into a human tumour cell line (HT1080) that synthesizes the collagen IV found in basement membranes, but does not synthesize any fibrillar collagen. About two per 100 clones resistant to the neomycin analogue G418 synthesized and secreted human procollagen II. Milligram quantities of the recombinant procollagen II were readily isolated from the cultured medium. The recombinant procollagen II had the expected amino acid sequence as defined by nucleotide sequencing of mRNA-derived cDNA and the expected amino acid composition as defined by analysis of procollagen II that was converted into collagen II by digestion with procollagen N- and C-proteinases. Also, analysis of the carbohydrate content indicated that there was glycosylation of some of the hydroxylysine residues but no evidence of post-translational overmodification of the residues. In addition, the protein was shown to have a native conformation as assayed by a series of protease digestions. No essential differences were found between clones transfected with the COL2A1 gene construct containing the COL1A1 promoter and the similar construct containing the COL2A1 promoter in terms of number of clones synthesizing recombinant procollagen II and the levels of expression. With both constructs, the expression of the COL2A1 gene was closely related to copy number. The results demonstrated therefore that it is not essential to use a promoter for a gene normally expressed in a host cell in order to obtain gene copy-number-dependent expression of an exogenous collagen gene in stably transfected cells.

Deletion of a large domain in recombinant human procollagen II does not alter the thermal stability of the triple helix.

A construct of the human gene for procollagen II (COL2A1) was prepared with an internal deletion of 5 kilobases that removed 12 exons coding for 291 amino acids from near the NH2 terminus of the triple helix. The construct was then used to transfect stably a human tumor cell line (HT-1080), and clones secreting internally deleted pro alpha 1(II) chain of procollagen II were isolated. The protein was purified, and the thermal stability of the triple-helical domain was assayed by brief protease digestion. The thermal stability of the internally deleted protein was the same as that of intact collagen II even though the triple helix was 39% shorter. Additionally, the thermal stability of the collagenase A fragment was the same as that of the collagenase A fragment of normal collagen II even though it was 38% shorter. Analysis of the results suggested that the thermal stabilities of large fragments of collagen II depended primarily on their contents of -Gly-Pro-Hyp-triplets corrected for length.