Non-helical Regions Research Articles

Purification and characterization of a rabbit bone metalloproteinase that degrades proteoglycan and other connective-tissue components.

A metalloproteinase, 'proteoglycanase', that degrades proteoglycan and insoluble type IV collagen as well as casein was purified to homogeneity from rabbit bone culture medium. The major form of this proteinase had a final specific activity of 2400 micrograms of casein degraded/min per mg of enzyme protein, and Mr 24 500 by sodium dodecyl sulphate/polyacrylamide-gel electrophoresis or 12 500 by gel-filtration chromatography. It was active over the pH range 5.0-9.0 against a number of substrates, and the rates of degradation were almost constant over the whole of this range. The products generated from proteoglycan-aggregate degradation by this enzyme indicated cleavage at multiple chondroitin sulphate-binding sites along the protein core. In a new assay to detect degradation of insoluble type IV collagen, the proteoglycanase generated large fragments, probably by cleavage in the non-helical regions. The enzyme degraded laminin, fibronectin and procollagen, removing the extension peptides of the last-mentioned. It also cleaved the 'weak region' of the type III collagen helix in a manner analogous to trypsin. The synthetic substrate 2,4-dinitrophenyl-Pro-Leu-Gly-Ile-Ala-Gly-Arg-NH2 was cleaved exclusively at the Gly-Ile bond. The proteoglycanase was inhibited by tissue inhibitors of metalloproteinases from rabbit bone culture medium, human amniotic fluid and bovine nasal-cartilage extracts, forming essentially irreversible inactive complexes. The importance of this tissue-derived enzyme, with such a wide-ranging degradative capacity, in normal and pathological connective-tissue matrix degradation is discussed.

ウシ皮膚コラーゲンの非還元性架橋Ｈｙｄｒｏｘｙａｌｄｏｌｈｉｓｔｉｄｉｎｅの分子内位置

A three-chained peptide, cross-linked by hydroxyaldolhistidine, was isolated from a tryptic digest of insoluble bovine skin collagen. The cross-link was cleaved specifically at its double bond by using 0s04/NaI04 into a two-chained cross-link peptide and a single peptide. After the separation of these products by gel chromatography, the two-chained peptide (containing the new bifunctional cross-link) was sequenced. This cross-link consists of an aldehyde derived from hydroxylysine 87 in al-CB5 and an aldehyde derived from lysine in the C-terminal nonhelical region of αl-CB6. The original three-chained peptide did not contain any glycosylated hydroxylysine nor glycosylated hydroxyaldolhistidine. The lack of glycosylation of hydroxylysine 87 in αl-CB5, which is usually glycosylated, may allow the formation of the aldehyde. This suggests that the role of glycosylation of the specific hydroxylysine residues is to prevent their oxidative deamination to the aldehyde, thereby preventing the formation of complex stable cross-link. In addition, it was found that the αl-CB6 portion of the peptide showed a sequence inversion Ala-Lys-His instead of normal sequence Lys-Ala-His. It is thought that the combination of these unusual factors generates the formation of this nonreducible stable cross-link hydroxyaldolhistidine.

Collagen structural microheterogeneity and a possible role for glycosylated hydroxylysine in type I collagen.

A three-chained peptide from type I collagen, crosslinked by hydroxyaldolhistidine, has been isolated from a tryptic digest of 5 M guanidine.HCl-insoluble bovine skin collagen (a small but as yet unknown percentage of the total collagen in whole skin). OsO(4)/NaIO(4) specifically cleaved the crosslink at its double bond into a two-chained crosslink peptide and a single peptide. The sequence of the two-chained peptide containing the bifunctional crosslink was determined after amino acid analysis of the separated peptides. The crosslink consists of an aldehyde derived from hydroxylysine-87 in the aldehyde-containing cyanogen bromide fragment alpha1CB5(ald) and an aldehyde derived from the lysine in the COOH-terminal nonhelical region of the alpha1CB6(ald) fragment. The alpha1CB6(ald) portion of the peptide exhibited structural microheterogeneity, containing the inverted sequence Ala-Lys-His instead of the normal sequence Lys-Ala-His. This indicates that another structural gene exists for alpha1(I) chain. The original three-chained peptide did not contain any glycosylated hydroxylysine or glycosylated hydroxyaldolhistidine. The lack of glycosylation of hydroxylysine-87 in alpha1CB5, which is usually glycosylated, allowed formation of the aldehyde, and this, coupled with the sequence inversion, may have allowed formation of the nonreducible crosslink hydroxyaldolhistidine. We suggest that the role of glycosylation, a posttranslational modification, of specific hydroxylysine residues is to prevent their oxidative deamination to aldehydes, thereby precluding formation of complex stable crosslinks. Complex crosslinks would decrease the rate of collagen turnover. The decrease, with time, would increase the population of stable crosslinked collagen molecules, which would eventually accumulate with age.

Covalent crosslinking between molecules of type I and type III collagen. The involvement of the N-terminal, nonhelical regions of the alpha 1 (I) and alpha 1 (III) chains in the formation of intermolecular crosslinks.

Several peptides containing intermolecular crosslinks were isolated from tryptic digests of the insoluble matrix of human leiomyoma and calf aorta. From leiomyoma type III collagen, two crosslinked peptides were isolated. The peptide [TN(III)]2, is a dimer which contains the 73 amino‐terminal residues of two, zero‐D staggered, crosslinked α1(III) chains (D being the distance by which molecules or chains are staggered relative to each other). After de‐blocking with pyroglutamate aminopeptidase, its amino‐terminal amino acid sequence was determined to be < Glu‐Tyr‐Asp‐Ser‐Tyr‐Asp‐Val‐Xaa‐Ser‐Gly‐Val‐Ala‐Val‐Gly‐Gly‐Leu‐Ala‐Gly‐Tyr‐Hyp‐Gly‐. The involvement of [TN(III)]2 in the formation of 4D‐staggered intermolecular bonds was demonstrated by sequence analysis of [TN(III)2]×T(III). For this three‐chained peptide, a double sequence which corresponded to TN(III) and T(III), a nine‐residue peptide from the carboxy‐terminal helical crosslinking site of type III collagen, was found.Crosslinks between type I and type III collagens were identified by isolating a peptide TN(III)×TN(I) from leiomyoma. This peptide was characterized, first by digesting it with collagenase giving rise to ColN(III)×ColN(I) and then chymotrypsin giving C(III) and C(I). C(I) and C(III) are peptides characteristic for type I and type III collagens, respectively. Further, three‐component crosslink‐containing peptides, from both type I and type III collagen, were isolated from aorta and leiomyoma. The peptides ColN(III)× ColN(I)× T(III) from leiomyoma and Col×N(III) ColN(I)× T(I) from aorta are crosslinked amino‐terminal regions of 0D‐staggered type I and type III molecules, joined either to an adjacent 4D‐staggered type III or type I molecule. Crosslinked peptides involving only type III collagen molecules were also isolated from calf aorta [colN(III)]2 and [CoIN(III)2]×T(III)].As intramolecular bonds between type I and III collagens can be excluded the isolation of these crosslinked peptides demonstrated that type I and type III collagen molecules are present within the same collagen fibrit with OD stagger, bound by intermolecular crosslinks. The significance of these findings with respect to the proposed models for collagen fibres and the physiological properties of the extracellular matrix, are discussed.

The chemistry of the collagen cross-links. Purification and characterization of cross-linked polymeric peptide material from mature collagen containing unknown amino acids.

A polymeric form of the alpha 1-chain C-terminal peptide alpha 1 CB6 (poly-alpha 1 CB6) was purified from CNBr digests of insoluble bovine tendon type-I-collagen by gel filtration and ion-exchage chromatography. The purified material had a molecular weight of 1.5 x 10(6)-5 x 10(6) on gel filtration and an amino acid content virtually identical with that of monomeric peptide alpha 1 CB6. The material could be adsorbed on affinity gels containing immobilized anti-(alpha 1 CB6-peptide non-helical region) antibodies and was an inhibitor of haemagglutination by the same antibodies of alpha 1 CB6-peptide-coated sheep erythrocytes. Periodate treatment of the material had no effect. Alkali hydrolysates were shown to contain two unknown amino acids, which were purified by gel filtration and ion-exchange chromatography in volatile buffers and are believed to be components of the mature cross-link of collagen.

Structure of epidermal keratin and variations in its polypeptide composition.

The subunit structure of prekeratin has been studied by examining the products formed after proteolytic digestion and site-specific cleavage at cysteine and cystine residues. The monomer unit of prekeratin consists of 3 polypeptides which contain a helical segment of 34,000 daltons. This segment appears to consist of 2 helical polypeptides separated by a non-helical one, and the helical polypeptides in turn contain 2 smaller segments of about 7,000 daltons. Cysteine residues are not localized to one region of prekeratin, but are evenly distributed in the molecule; this suggests that they may be located in the nonhelical regions. The polypeptide composition of keratins has been studied by SDS polyacrylamide gel electrophoresis, and heterogeneity in the polypeptide components can be demonstrated between species, within a species, and in different areas of skin of a single member of species. The pattern of keratin polypeptides, furthermore, is not fixed but can be altered by changes in the cellular environment.

Structural basis of human alpha-1-antitrypsin deficiency

Two common variants, S and Z, of alpha-1-antitrypsin result in respectively a mild and severe circulating deficiency which predisposes towards the development of emphysema. A feature unique to the Z variant is the impaired secretion from the hepatocyte resulting in intra-cellular accumulation of insoluble, partially glycosylated antitrypsin. Peptide mapping techniques revealed that the S protein differed from the normal by the substitution of a glutamic acid by a valine residue, and the circulating Z protein differed from the normal by the substitution of a different glutamic acid by a lysine residue. Cyanogen bromide fragmentation of antitrypsin gave about ten fragments one of which contained the S and Z mutation sites. This fragment of 109 residues was sequenced and was large enough to allow a computer prediction of secondary structure. The S protein mutation site is predicted to be on a hydrophilic aspect of an alpha-helical section. Hence the presence of the new valine could result in incorrect folding or aggregation of the nascent protein to give intra-cellular degradation and decreased cellular export. The Z mutation has introduced a new lysine adjacent to another lysine in a position of the molecule that is likely to be externally situated and in a non-helical region. Paired basic residues are known to act as molecular markers for protein modification. The pathogenesis of the Z deficiency would be explicable by the new paired basic residues competing with an earlier homologous binding site at the rough endoplasmic reticulum level. Some of the Z protein will be processed but much of it will be arrested at the stage of incomplete glycosylation.

The formation and thermal stability of in vitro assembled fibrils from acid-soluble and pepsin-treated collagens

The role of the non-helical regions of the collagen molecule in fibrillogenesis has been investigated by comparing the kinetics of fibril formation of pepsin-treated acid-soluble collagen, acid-soluble collagen and mixtures of the two and by comparison of the thermal stabilities of the fibrils formed. The acid-soluble collagen was found to aggregate more rapidly than the pepsin-treated collagen under physiological conditions of pH and ionic strength. Variations in ionic strength, at physiological pH, were found to have differing effects on the aggregation of these two forms of soluble collagen. Fibrils formed from the pepsinized-collagen had a lower thermal stability than those formed from the intact collagen. The behavior observed with mixtures of acid-soluble and pepsin-treated collagens was found to be quantitatively consistent with the pepsinized collagen being able to utilize the nuclei formed by the acid-soluble collagen for subsequent growth. However, the use of the acid-soluble nuclei by the pepsinized collagen for growth did not enhance its rate of precipitation during the growth phase, nor did it enhance the thermal stability of the fibrils formed from the pepsinized collagen.

A survey of peptides containing sites of phosphorylation in nonhistone nuclear proteins

32P-labeled peptides obtained by pronase digestion of unfractionated nonhistone nuclear proteins were resolved on columns of Dowex 50, DEAE-Sephadex, Bio-Gel P2, and paper electrophoresis at pH 1.8. Each of 30 peptides analyzed contained predominantly glycine, glutamic acid and proline. The chain length ranged from 7 to 19 residues, including 1 to 4 phosphorylated residues per peptide. These results suggest phosphorylation sites in nonhistones involve a high negative charge density in non-helical regions of these proteins.

Amino acid sequence of the aminoterminal segment of dermatosparactic calf-skin procollagen type I.

The N-terminal procollagen peptide of the pN alpha 1(I) chain from dermatosparactic calf skin contains 139 amino acid residues. For the determination of the amino acid sequence the procollagen peptide was treated with pyroglutamate aminopeptidase, protease from Staphylococcus aureus V8 and trypsin. The fragments obtained were separated by molecular sieve and ion-exchange chromatography and submitted to automated Edman degradation. The procollagen peptide consists of three segments, an N-terminal globular domain which contains all the cysteine residues and most of the hydrophobic residues present in the entire peptide, a triple helical part with a relatively high content of proline and hydroxyproline, and a short nonhelical region which forms the connection to the nonhelical region of the alpha 1(I) chain and which contains the proline-glutamine bond specifically split by the N-terminal procollagen peptidase during conversion of procollagen to collagen.

Isolation of Crosslinked Peptides from Insoluble Human Leiomyoma

Two collageneous crosslinked peptides have been isolated from tryptic digests of sodium cyanoborohydride(NaBH3CN)-reduced insoluble matrix of human leiomyoma. The amino acid composition of both peptides demonstrates that the N-terminal, non-helical region of type III collagen is involved in crosslinking. The monomer (Col1) and dimer (Col1)2 N-terminal, non-helical peptides, consisting of 21 and 42 residues respectively, were found to be cross-linked to a tryptic peptide ('T9′, 9 residues) from the helical region of an adjacent type III molecule, forming the crosslinked peptides Col1 × T9 and (Col1)2× T9. The location of peptide T9 in the C-terminal, helical crosslink region [residues 922–930 of the α1(III)chain] was confirmed by the following observations. a) Its amino acid composition, which was deduced by calculation of the composition difference between the crosslinked peptide (Col1)2× T9 and the non-crosslinked peptide (Col1)2. This was consistent with the corresponding sequence of calf calf α1(III) when methionine is replaced by hydroxyproline. b) Cleavage of the crosslink of Col 1 × T9 by periodate, which released a peptide with the amino acid composition of T9. c) Sequence determination of the crosslinked peptide (Col1)2× T9; since the N-terminus of Col1 is blocked by pyroglutamic acid, sequence determination reveals the N-terminal sequence of T9: The chemical nature of the reduced crosslinks was shown to be different in the two crosslinked peptides. Col1 × T9 contains hydroxylysinonorleucine o5Lys(ω Nle) whereas (Col1)2× T9 contains a trifunctional crosslink, which is different from the trifunctional crosslinks that have been found in type I collagen.

Isolation and characterization of a double chain intermolecular cross-linked peptide from insoluble calf bone collagen.

A double chain peptide containing the sodium borohydride-reduced intermolecular cross-link, hydroxylysinohydroxynorleucine, was isolated following sequential cyanogen bromide digestion and limited alkaline hydrolysis of insoluble calf bone collagen. Amino acid composition and NH2-terminal sequence analysis indicated that the peptide was highly acidic and consisted of 19 amino acid residues including the cross-link. Amino acid composition and automated sequence analysis of this peptide before and after cleavage of the cross-link, using periodic acid, provided the data from which the following structure was deduced. (formula: see text). The sequence of the larger peptide is identical with that of residues 8c to 19c in the COOH-terminal nonhelical region of the homologous skin collagen alpha1 chain. The hydroxylysine residue located at position 17c in the alpha chain of type I collagen appears to be a predominant site for intermolecular cross-link formation. Assignment of the smaller peptide component within the known primary structure of the collagen molecule currently cannot be made.

Amino acid sequences of alpha-helical segments from S-carboxymethylkerateine-A. Tryptic and chymotryptic peptides from a type-II segment.

1. Amino acid-sequence studies were done on a peptide of mol.wt. approx. 12500 that was isolated from the highly helical fragments obtained by partial chymotryptic digestion of the low-sulphur proteins (S-carboxymethylkerateine-A) from wool. 2. The peptides obtained by tryptic and chymotryptic digestion of this large peptide were separated by ion-exchange chromatography on DEAE-cellulose at pH8.5 with an (NH4)(2)CO(3) concentration gradient and, where necessary, purified further by paper electrophoresis. 3. Determination of the sequences of many of these peptides showed that a high proportion of the cationic residues occurs in pairs. 4. Although two of the four S-carboxymethylcysteine residues are located in what appears to be a non-helical region near the N-terminus the other two S-carboxymethylcysteine residues occur in or near sequences suggesting a helical conformation. 5. Some peptides were obtained, in low yields, that appeared to be homologues of more major ones. These suggest either homologies in the helical portions of the low-sulphur proteins or the presence of closely related amino acid sequences in helical regions of completely different origins. 6. A partial sequence of the complete peptide is proposed.

Trichotoxin A-40, a new membrane-exciting peptide. Part A. Isolation, characterization and conformation

The new polypeptide antibiotic trichotoxin A-40 is isolated by chloroform/methanol extraction from the dry mycelium of Trichoderma viride NRRL 5242. The lipophilic peptide is purified by chromatography on Kieselgel H-60 and reverse-phase chromatography on Lichrosorb RP-8. The new antibiotic differs in amino acid composition and various chemical and physicochemical properties from similar peptides such as trichotoxin A, the suzukacillins or alamethicins. The amino acid composition is (Pro) 1 (Gly) 1 (Ala) 2 (Leu) 2 (Aib) 10 (Glx) 2. (Aib, α-aminoisobutyric acid.) The antibiotic has a carboxyl group which can be esterified by diazomethane, which results in slightly enhanced membrane-modifying activities. The peptide exhibits a right-handed α-helical conformation increasing about two-fold from aqueous to lipophilic media as shown by solvent-dependent circular dichroism measurements. Most of the 13C-NMR resonances can be assigned unequivocally and amino acids situated in the α-helical part show characteristic shift differences from those in the non-helical regions. No β-phenylalaninol residue could be identified by 13C-NMR and ultraviolet spectroscopy, as can be for alamethicins and suzukacillins. A pronounced hemolytic action is found on human erythrocytes, which develops at micromolar concentrations. Trichotoxin A-40 induces a voltage-dependent ionic conductance in bilayer lipid membranes and it can serve as a new pore-forming model system for structure/activity studies in membrane excitation by peptides.

Synthesis of hapten-polypeptide conjugates as antigen models for the N-terminal region of the alpha-2-chain of rabbit skin collagen.

Synthesis of derivatives of the peptide sequence L-pyroglutamyl-L-phenylalanyl-L-aspartyl-glycyl-L-lysyl-glycyl-glycyl-glycine as the antigenic determinant representing the N-terminal non-helical region of the α-2-chain of rabbit skin collagen, and conjugation to two different polypeptide carriers, are described.

31P Magnetic relaxation studies of yeast transfer RNAPhe

AbstractThe molecular mechanism of thermal unfolding of yeast tRNAPhe in 20 mM NaCl, 1 mM EDTA, and 10 mM MgSO4, pH 7.1 ± 0.1, has been examined by 31P magnetic relaxation and the nuclear Overhauser effect methods at 40.48 MHz in the temperature range of 22.5–80°C. Two partially resolved 31P resonance peaks of yeast tRNAPhe have been found to behave distinctively different in their longitudinal relaxation times. Individual intensities of the two partially resolved peaks have been quantitatively estimated by the use of relaxation data and the nuclear Overhauser effect as a function of temperature. The results of these observations largely support the earlier suggestion by Guéron and Shulman that the high‐ and low‐field parts of the main 31P resonance cluster originate from phosphorus nuclei belonging to the double‐helical and nonhelical regions of the tRNA, respectively. The spin‐lattice relaxation of the phosphorus nucleus has been found to be determined dominantly by the dipolar interaction with the surrounding ribose protons at this observing frequency. Rotational correlation times for the two portions of the ribose‐phosphate backbone of the tRNA have been separately deduced from the quantitative treatment of the 31P nuclear spin‐lattice relaxation times (T1) and the nuclear Overhauser effect. The result indicates that the two portions undergo internal motions at distinctively different rates of 108–1010 sec−1 order in the temperature range of 22.5–80°C, and that the thermal activation of these motions occurs at least in three distinctive steps, i.e., 22.5–31, 31–40, and 40–80°C. The rates of the internal motions and the associated activation energies in respective steps give some insight into the thermo‐induced change of the yeast tRNAPhe structure.

Onset of helical order

Renormalization group methods are used to describe systems which model critical phenomena at the onset of helical order. This onset is marked by a change in the “bare propagator” used in perturbation theory from a k2-dependence to a more general form. We consider systems which in the non-helical region exhibit O simultaneously critical phases. Results are given to first order in an ϵ-expansion. For the isotropic case of k2L dependence and O = 2, we give η to first order in 1/n for d- ⩽ d ⩽ d+ where d+- are upper and lower borderline dimensions.

Structural domains of transfer RNA molecules.

In this article, we have described various detailed features of the conformation of yeast tRNA(Phe) revealed by recent refinement analysis of x-ray diffraction data at 2.5 A resolution. The gross features of the molecule observed in the unrefined version have been largely confirmed and a number of new features found. The unique role of the ribose 2' hydroxyl groups in maintaining a series of nonhelical conformations in this RNA molecule has become apparent. Many of these features are a direct consequence of the geometry of the ribose phosphate backbone of RNA molecules, and these may also be found in structured regions of other RNA species as well. Special attention has been directed toward two conformational motifs revealed by this analysis. These include the striking similarity between the TpsiC and anticodon hairpin turns in the polynucleotide chain, which are stabilized by the participation of uridine in the U turn. In addition, there is frequent occurrence of an arch conformation in the polynucleotide chian which is stabilized by hydrogen bonds from 2' hydroxyl residues to phosphate groups across the base of the arch. The importance of the 2' hydroxyl interactions in defining tertiary structure is illustrated by the fact that, in the nonhelical regions, almost half of the ribose residues are involved in O2' hydrogen-bonding interactions which stabilize the conformation of the molecule.

Conformation of Escherichia coli glutamic acid tRNA II as studied by hydrogen-tritium exchange catalyzed by cysteine methyl ester.

Incubation of CMP in 2H2O with 0.5M cysteine methyl ester at p2H 5 and 37 degrees C for 24 h resulted in 43% exchange of 5-H to 5-2H. No deamination of the cytosine nucleus was noted during this treatment. Native and denatured DNA samples from calf thymus were treated in 3H2O with cysteine methyl ester at pH 5 and 37 degrees C for 24 h and incorporation of tritium into each DNA base was determined by enzymic digestion of the treated DNA. The order of the specific radioactivity found was cytosine greater than guanine greater than adenine greater than thymine for denatured DNA and guanine greater than adenine approximately cytosine greater than thymine for native DNA. The ratio of radioactivity for denatured/native was 11.6 for cytosine, 1.5 for guanine, 1.8 for adenine and 1.1 for thymine. Hence the incorporation in cytosine under the reaction conditions is preferential for single-stranded, nonhelical regions of DNA. Escherichia coli glutamic acid tRNA II was treated in 3H2O with 1.24 M cysteine methyl ester at pH 5 and 37 degrees C. The 24-h-treated tRNA was digested with ribonuclease T1 and the fragments were fractionated. Each fragment was then digested with ribonuclease T2 into mononucleotides and the radioactivity distribution among the bases was determined. The average radioactivity found for each of the bases of the four major nucleotides was cytosine greater than guanine approximately adenine greater than uracil. The radioactivity in cytosine varied greatly among the RNase T1 fragments, the ratio of the highest to the lowest radioactivity being 18.7. The corresponding value for guanine was 11.1, for adenine 4.73 and for uracil 3.64. Based on the data obtained, it was deduced that in this tRNA the anticodon loop, the dihydrouridine loop and the extra loop were "exposed" under the conditions employed for the labeling. The 5'-terminal cytosine of the anticodon loop was in a "non-exposed" state, a situation similar to that previously reported for E. coli tyrosine tRNA [Cashmore, A. R., Brown, D. M. & Smith, J. D. (1971) J. Mol. Biol. 59, 359-373] and for E. coli formylmethionine tRNA [Goddard J. P.+Schulman L. H. (1972) J. Biol. Chem. 247, 3864-3867]. Both cytosine 48, located at the 3'-terminal of the extra loop, and guanine 15 in the dihydrouridine loop were in an "emposed" state. This finding does not agree with a tRNA model in which this pair of cytosine and guanine, commonly found in tRNA sequences, forms hydrogen bondings. Positions 30--32, 61--64 and 71, which are located in the stems, were found to be strongly "buried".

Chemistry of the collagen cross-links. Nature of the cross-links in the polymorphic forms of dermal collagen during development

Both the type I and type III collagens present in embryonic dermis are stabilized by the intermolecular cross-link, hydroxylysino-5-oxonorleucine, derived from hydroxylysine-aldehyde, although the type I collagen possesses a significant proportion of dehydrohydroxylysinonorleucine. However, concurrent with the change in the proportion of the two types of collagen during postnatal development there is a change-over with both type I and III collagens to the labile cross-link, dehydrohydroxylysinonorleucine, derived from lysine aldehyde. The results indicate that the change in the nature of the cross-link with development is determined primarily by the change in the extent of hydroxylation of the lysine residues in the terminal non-helical regions rather than being due to the change in the type of collagen.