Proc. Nat. Acad. Sci. USA Vol. 69, No. 9, pp. 2594-2598, September 1972

Reducible Crosslinks in Hydroxylysine-Deficient of a Heritable Disorder of Connective Tissue (skin/bone/cartilage/aminoacid analysis)

DAVID R. EYRE AND MELVIN J. GLIMCHER* Department of Orthopedic Surgery, Harvard Medical School, Children's Hospital Medical Center, Boston, Massachusetts 02115 Communicated by Francis 0. Schmitt, July 3, 1972

ABSTRACT Reducible compounds that participate don, bone, and cartilage collagens (3-11). Each tissue reveals in crosslinking were analyzed in hydroxylysine-deficient a unique distribution of these reducible crosslinks that changes collagens of patients with a heritable disorder of connec- tive tissue. After treatment with [3H1sodium borohydride, as the tissue matures and ages (8, 10). new compounds, as well as a totally different pattern of Since connective tissues of patients with this disorder are tritiated compounds, were found in hydroxylysine-de- deficient in hydroxylysine, a crosslink precursor, it seemed ficient from skin as compared with age-matched likely that the reducible crosslinks would either be absent or controls. The amount of detected indicated collagen cross- that more elastin was present in abnormal skin than in abnormal. Such a deficiency or abnormality in control skin. linking might be responsible for changes in the solubility Bone collagen, which was not as deficient in hydroxy- characteristics of the collagen (1), and for changes in the as skin collagen, had the same compounds as normal structural properties of the tissues and the consequent skeletal bone collagen, but their relative proportions were altered, and connective tissue abnormalities. consistent with a deficiency of hydroxylysine, a precursor of the crosslinks. Although the content of hydroxylysine Biopsies of skin, bone, and cartilage removed during sur- in collagen of cartilage is essentially normal in these pa- gery were analyzed for the reducible components of collagen, tients, analysis after reduction revealed a different pat- and the results were compared to those obtained from the tern of reduced compounds from that of normal cartilage. It is speculated that Type II collagen, the major collagen component in cartilage, contains a normal amount of hydroxylysine, while Type I collagen, which is the major source of the crosslinks, is hydroxylysine-deficient. This Hydroxyysine - Defitient Skin distribution would explain the findings of an abnormal profile of reducible compounds despite an almost normal total hydroxylysine content. The finding that the deficiency of hydroxylysine in the 5 _ collagen of these patients is accompanied by changes in number, chemical nature, and, probably, distribution of ? crosslinkages, and the previously reported alterations in the solubility characteristics, suggest that at least some skeletal and connective tissue abnormalities are directly related to underlying molecular pathology. I.. 4 r Control SkinPost-Histt.

Aminoacid analyses of biopsies from two siblings with a heri- -4 V~~~~~~~~~~~LNL table disorder of connective tissue revealed the presence of '4.. hydroxylysine-deficient collagen (1). Hydroxylysine was '4.. present in skin collagen at about 5% of the amount found in collagen of normal skin, and at higher, but still markedly reduced, amounts in fascia and bone, while the content in 4 Control Skin 'Post-Iistidinen amount. The basic cartilage was about 90% of the normal 14,4~HLNL~~~~~~~~ biochemical defect is an abnormally low activity of the en- zyme lysyl-protocollagen hydroxylase in cells that synthesize I0Z,0 collagen in the affected connective tissues (2). Hydroxylysine is a precursor in the biosynthesis of certain collagen crosslinks (3, 4). Borohydride-reducible crosslinks, which are apparently intermediates in the biosynthesis of more stable bonds, have been identified in normal skin, ten- 00 20 40 60 80 100 120 140 160 180 200 FRACTION NUMBER Abbreviations: HLNL, hydroxylysinonorleucine; HLHNL, hy- FIG. 1. Comparison of the elution profiles of tritiated com- droxylysinohydroxylnorleucine; DHNL, dihydroxynorleucine. pounds in hydrolysates of normal and hydroxylysine-deficient * Reprint requests to: Melvin J. Glimcher, M.D., Department of human skin collagens treated with [3H]borohydride. 10 mg dry Orthopedic Surgery, Children's Hospital Medical Center, 300 weight of each material were hydrolyzed in acid and applied Longwood Ave., Boston, Mass. 02115. to the aminoacid analyzer. HLNL, hydroxylysinonorleucine. 2594 Downloaded by guest on September 30, 2021 Proc. Nat. Acad. Sci. USA 69 (1972) Crosslinks in Hydroxylysine-Deficient Collagens 2595

collagens of site-matched and age-matched samples of normal Hydroxylysine- Deficient Skin tissues. METHODS Skin biopsies were scraped free of fat and stirred in 2 M NaBr for several hours to facilitate removal of epidermis (1). The dermis was washed thoroughly by stirring in saline and then water at 4°. Samples of cortical bone from spinous processes of thoracic vertebrae were washed by stirring in saline and then demineralized in 0.2 M EDTA (pH 7.9), at 40 for several days, followed by further washing in distilled water. Rib cartilage was stirred in saline and then water. Control Skin Samples of the washed tissues were cut into small pieces, HIS stirred in 0.1 M sodium phosphate (pH 7.4), and reacted with tritiated sodium borohydride (10 Ci/mol) at room tem- perature (250) for 45 min. Acetic acid was added to bring the solution to pH 3.0, and after 5 min of further stirring, the tissue pieces were thoroughly washed in several changes of distilled water. After the tissue was dried over P205, it was hydrolyzed in 3 N HCl in an evacuated, sealed tube for 48 hr at 1050, conditions that prevent certain participants of crosslinking from being destroyed (12). Amino acids were I_e--- r------?- . fractionated on an aminoacid analyzer equipped for stream 155 160 165 170 175 180 splitting. Tritium activity was measured in a portion of the FRACTION NUMBER column effluent, and radioactive peaks were identified as described (13). FIG. 2. Regions of the two chromatograms depicted in Fig. 1 Site-matched and age-matched samples of skin, bone, and comparing the elution profiles of tritium activity and of ninhy- cartilage were taken at biopsy or autopsy from normal con- drin-positive material. - , absorbance at 570 nm; nective tissues and analyzed by the same procedures. 'H activity.

RESULTS from the abnormal skin collagen eluted from the column later Skin than the major radioactive peak found in normal skin collagen, The elution profile of tritiated compounds from a hydrolysate indicating they were not the same compounds. of reduced collagen of abnormal skin was markedly different The radioactive peak eluting at fraction number 130 was from that of normal skin (Fig. 1). Three independent age- also of interest. This peak was much larger in borohydride- matched samples of normal skin revealed essentially the treated samples of diseased skin than in normal skin. It eluted same patterns. Skin biopsies from each of the two siblings in an identical position to the reduced peak from with the disease gave virtually the same abnormal elution a hydrolysate of reduced elastin. Identification as reduced profiles of tritium activity. The major peak from normal desmosine on this basis alone was not reliable since several skin collagen was apparently the material described as post- minor radioactive peaks also eluted in this region on analysis histidine compound (14, 15) or Fraction C (6, 9, 10). The of bone collagen. However, conventional aminoacid analysis, next most abundant compound was hydroxylysinonorleucine by reaction with ninhydrin, of a hydrolysate of a relatively (HLNL), derived by reduction of the aldimine crosslink, large sample (10 mg dry weight) of dermis, which had not dehydrohydroxylysinonorleucine, formed by addition between been treated with borohydride, confirmed that more of the a residue of hydroxylysine and a residue of allysine, the alde- elastin crosslinks, desmosine and isodesmosine, were present hyde from an enzymic oxidative deamination of a lysyl res- in abnormal skin than in a sample of control skin. There idue. The abnormal skin, however, did not reveal any of appeared to be four times the concentration of desmosines the normal "post-histidine" compound, and HLNL was per unit weight of collagen in one of the samples of diseased much diminished and possibly absent since its elution posi- skin compared to normal skin. Collagen was estimated by tion was now obscured by several minor peaks. Instead, a content. new major peak of tritium activity eluted later than the normal "post-histidine" peak, and appeared as a partially Bone resolved double peak suggesting that more than one com- The hydroxylysine content of samples of diseased bone varied pound was present. A ninhydrin-positive double peak coin- considerably. Spongy bone from the interior of a vertebra cided precisely with the radioactive peak (Fig. 2). In terms of contained collagen with about 50% of the normal amount of lysine equivalents, this double peak represented in total one hydroxylysine (1). However, histological examination after to two lysine residues per molecule, or about 0.5 lysine res- staining with Safranin 0 showed that pockets of cartilage idue per a chain. It was not detected in the collagen without were included in the bone sample. In the present study, corti- prior borohydride treatment, indicating that the compound cal bone from a spinous process of a thoracic vertebra was before reduction was labile to acid hydrolysis and that little analyzed. Much less hydroxylysine was present in the collagen or no reduction of this material had occurred in vivo. Fig. 2 of this tissue (10-20% of the normal value), and cartilaginous also illustrates more clearly that the radioactive double peak pockets were not detected histologically. Downloaded by guest on September 30, 2021 2596 Biochemistry: Eyre and Glimcher Proc. Nat. Acad. Sci. USA 69 (1972)

Hydroxylysine-Deficient Bone The pattern of reduced compounds isolated from abnormal bone collagen treated with borohydride was different from that derived from control samples of normal bone. Fig. 3 t' ,,,Il ,, f , Is 'F zt4R it IN I shows the elution profiles of tritium activity for the abnormal HLNL bone and a sample of control bone. The main difference be- HLHNL tween the two patterns is the marked decrease in the relative proportion of hydroxylysinohydroxynorleucine (HLHNL) recovered from hydroxylysine-deficient bone collagen com- k pared to control bone collagen. HLHNL accounted for 33% ?1 of the tritium activity recovered in the column effluent from a treated with IaJ J--\JI\> hydrolysate of normal bone collagen [3H]- borohydride, while only 13% of the activity in the column

tn- Control Bone HLHNL effluent from the abnormal bone collagen was represented by this compound. The radioactive peak identified by its 10 14J elution position as dihydroxynorleucine (DHNL) was much HLNL more pronounced on analysis of the diseased bone compared 6 to control bone. It accounted for 15% of the activity recovered 5 from diseased bone collagen, but only 3% of activity recovered 4 from the control bone collagen. on 3 The "post-histidine" peak of activity observed analysis Post-Histidine of control skin collagen was also present as a minor component 2 in the elution of control bone DHNL profile collagen. Interestingly, similar to the diseased skin, this material was absent on ana- HNL lysis of diseased bone; instead, a peak of activity eluted later 0 20 40 60 80 100 1201 140 160 180 200 in the chromatogram in an identical position to the major FRACTION NUMBEAe peak of activity recovered from borohydride-treated diseased skin. FIG. 3. Comparison of the elution pirofiles of tritiated com- C pounds in hydrolysates of normal and hydroxylysine-deficient Cartilage human bone collagen treated with [3EI] borohydride, DHNL, Rib cartilage biopsied from one of the patients also revealed a dihydroxynorleucine; HNL, hydroxynor rleucineo HLHNL, hy- deficiency in HLHNL relative to control human cartilage droxylysinohydroxynorleucine. when analyzed for radioactive compounds after treatment with tritiated sodium borohydride. Fig. 4 compares the elu- tion profiles of tritium activity for abnormal and control tissues treated with borohydride. Little can be concluded from the difference in total activities incorporated by unit Hydroxylysine - Deficient Rib Cartilage weight of abnormal and control tissues since independent samples of normal rib cartilage did themselves vary markedly Asti 4w9$ <8t 5 in the amount of activity they incorporated on reaction with ,lu a,,,,,, I borohydride. However, three independent age-matched con- trol biopsies of normal rib cartilage clearly revealed HLHNL as the major reducible compound. HLNL Aminoacid analyses of soluble collagen extracted as gelatin 4.. from abnormal cartilage showed a much lower amount of hydroxylysine than collagen of intact or the insoluble t4J tissue dissected to be free of non- '4. residue. Washed tissue, carefully HLHINL material, was stirred in 4 M (pH 7.4) io- Control Rib Cartilage cartilaginous CaCl2 for 24 hr at room temperature (250). About 5% of the total 9 collagen in the cartilage was solubilized as the gelatin by 6 _ this procedure. The extract was dialyzed exhaustively against distilled water, freeze-dried, and analyzed after hydrolysis 5 in acid. Although disc electrophoretic analysis (16) of the 4 extracted collagen showed mainly a chains with a predom- 3 inance of the al component, there was a significant amount HLNL of a2 detected, suggesting that the extract was enriched 2 II in Type I collagen as compared to the whole tissue (17-19). DHNL The amount of hydroxylysine in the extract, 6 residues per

- to 12 residues 100 0.- K!l F r- 100 hydroxyproline residues compared per 0 20 40 60 80 100 120 140 160 180 200 hydroxyproline residues in whole tissue, seemed unusually FRACTION NUMBER low even for pure Type I collagen of cartilage (17). However, FIG. 4. Comparison of the elution prc files of tritiated com- the extracted collagen apparently was not pure Type I, but a pounds in hydrolysates of normal and Eiydroxylysine-deficient mixture of Type I and Type II, judging by the excess of al human cartilage collagen treated with [3H]Iborohydride. chains seen on disc electrophoresis. Downloaded by guest on September 30, 2021 Proc. Nat. Acad. Sci. USA 69 (1972) Crosslinks in Hydroxylysine-Deficient Collagens 2597

DISCUSSION an elastic fibrillar network (21). Chemical analysis might Abnormalities in the number and chemical nature of the re- therefore be expected to reveal a higher concentration of ducible crosslinks and crosslink precursors have been found in elastin crosslinks in skin biopsies from at least some regions in collagens from several connective tissues of patients with a patients with this disease. Another explanation for the higher heritable disorder in which the basic biochemical lesion is a concentration of desmosine crosslinks detected in the abnormal deficiency of hydroxylysine in collagen, due to a low activity of skin might be that these crosslinks are present in collagen the enzyme lysyl-protocollagen hydroxylase (2). This disease rather than in elastin. This is not unreasonable considering the has been termed "hydroxylysine-deficient collagen disease" similarity between crosslink biosynthesis in collagen and in (1). Skin collagen, which contained the most diminished level elastin, the virtual absence of hydroxylysine from collagen of hydroxylysine of the diseased tissues studied (5% of nor- in the diseased skin, and the fact that elastin crosslinks are mal), gave rise to the most abnormal pattern of reduced com- derived solely from lysyl residues. pounds on analysis after treatment with borohydride. The The elution profile of tritiated compounds isolated from absence, on analysis of the diseased skin, of the "post- abnormal bone collagen after [3H]borohydride treatment histidine" peak detected as the major component on analysis differed from that of control human bone collagen consistent of normal skin collagen, suggests that hydroxylysine is a with a paucity of hydroxylysyl residues being available for precursor of the normal "post-histidine" compound. The participation in crosslinking. Thus, the major reducible new compounds eluting later in the chromatogram may be compound of normal bone, dehydrohydroxylysinohydroxy- analogues of the compound or compounds present in the norleucine (deH-HLHNL), which is derived in collagen from normal "post-histidine" peak, with lysine replacing hydroxy- two residues of hydroxylysine, was much diminished on anal- lysine. Their retarded elution position is consistent with their ysis of the hydroxylysine-deficient bone. Dehydrohydroxy- being hydroxyl-deficient, and therefore less acidic, analogues lysinonorleucine (deH-HLNL), which is derived from a residue of the normal "post-histidine" material. Detection of a rela- of hydroxylysine and a residue of lysine, was the major re- tively large amount of the compounds by ninhydrin analysis ducible compound detected in abnormal bone, as might be ex- (1-2 lysine equivalents per molecule), suggests that they may pected if lysyl residues now occupied some of the crosslinking be borohydride-reduced analogues of crosslink intermediates, sites normally occupied by hydroxylysyl residues. The such as the native form of "post-histidine" compound, that absence of a "post-histidine" compound on analysis of the had accumulated in the defective collagen. abnormal bone collagen and the appearance of about the same However, it cannot be assumed, solely on the basis of the amount of radioactivity in a retarded peak supports the deficiency of hydroxylysine, that the new reducible com- conclusion from analysis of the abnormal skin, that the new pounds are crosslinks in which lysine has replaced hydroxy- compound or compounds are hydroxyl-deficient analogues of lysine. Crosslinks derive only from a small fraction of the the "post-histidine" material detected in normal bone and total* lysyl and hydroxylysyl residues in collagen. Conse- skin collagens. quently, a deficiency, but not a complete absence, of hydroxy- The relatively high level of DHNL in abnormal bone colla- lysine need not impair normal crosslink synthesis. Interest- gen treated with borohydride suggests that more of the alde- ingly, a recent report (20) described an abnormal pattern of hyde derived enzymatically from hydroxylysine (hydroxy- reducible bonds in skin collagen from a patient with Ehlers- allysine) has accumulated in defective collagen than in normal Danlos syndrome, a disease having certain clinical features in bone collagen. Such a build-up of this crosslink precursor common with the "hydroxylysine-deficient collagen disease" might occur if insufficient hydroxylysyl residues were avail- (1, 2). This pattern showed some similarities, but was not iden- able at the correct sites on collagen chains for reaction with tical, to that of the abnormal skin collagen in the present study. aldehydes on adjacent chains. Although the hydroxylysine content of the skin collagen was The pattern of reduced compounds derived by analysis of within normal limits (G. Mechanic, personal communication), normal human cartilage treated with borohydride is similar to the basic biochemical defect again manifested itself as an ab- that reported by Bailey and coworkers in their analyses of normality of collagen crosslinking. However, the results refer collagen in vertebrate cartilages (10, 11). As in the present to one patient and may not be a universal feature of what is study, one major peak of activity, which is presumably described as Ehlers-Danlos syndrome, which may be classified HLHNL, was isolated from [3H]borohydride-treated tissue. into as many as five different subtypes (1). However, analysis of bovine articular cartilage from animals The elevated concentration of desmosines per weight of of various ages showed that by 6 years of age this compound collagen observed in a sample of the skin of one of the siblings had virtually disappeared and two unidentified peaks were used in this study in comparison to site-matched samples of now the main features on analysis of the cartilage collagen control human skin suggests that there is more elastin per unit (10). Such age-related changes in the distribution of reducible volume in the abnormal skin. Since samples of normal skin components of collagen also occur in human connective tissues from different regions of the body presumably have different (10), cartilage presumably included. However, in the present elastin contents, it cannot be conclusively established from study, analysis of biopsies of rib cartilage from patients these rang- results that the skin of patients with hydroxylysine- ing in age from 9-14 years all revealed HLHNL as the major deficient collagen disease has a higher content of elastin than borohydride-reduced component of collagen. Thus, in these normal human skin in general. Indeed, in the original report subjects, aging had not progressed sufficiently for this com- (1), samples of skin from the patients contained desmosines pound to disappear from rib cartilage, as it might do in older within the normal range when analyzed by conventional subjects. However, HLHNL was not the major compound de- aminoacid analysis. However, electron microscopic studies tected on analysis of cartilage from the 9-year-old patient with have indicated that skin samples from several patients with the disease. Even though 90% of the normal complement of various subtypes of Ehlers-Danlos syndrome were enriched in hydroxylysine was present in this cartilage, an abnormal Downloaded by guest on September 30, 2021 2598 Biochemistry: Eyre and Glimcher Proc. Nat. Acad. Sci. USA 69 (1972)

elution profile of tritium activity, relative to controls, was ob- be quantitated. The ratio of glucosylgalactosylhydroxylysine tained. The difference in reducible components between con- to galactosylhydroxylysine in the diseased bone collagen ap- trol and abnormal cartilages might be explained if reducible peared abnormally high (1.04: 1 in one analysis and 1.24: 1 in participants in crosslink synthesis were concentrated in one of a second analysis), compared to 0.39: 1 in a control and the re- the two types of collagen, Type I [(ali)2a2] or Type II ported value of 0.47: 1 for normal human bone (23). [(alul)3], present in cartilage (17, 18), and if only this type were deficient in hydroxylysine. If reducible components were Further analysis, if additional tissue from patients with this present mainly in Type I collagen, which seems to be a minor particular biochemical lesion becomes available, should reveal collagen component of cartilage relative to Type II (17, 18), whether the hydroxylysine deficiency is more pronounced in a then aminoacid analysis of whole tissue would reveal a high, particular polypeptide chain or regions of a polypeptide chain. almost normal, amount of hydroxylysine. However, reduction Such information would contribute to a fuller understanding with [3H]borohydride and analysis of reduced components of substrate specificity of lysine hydroxylase in the various would reveal a profile typical of the hydroxylysine-deficient connective tissues. Type I collagen. This tentative hypothesis explains the present findings 1. Pinnell, S. R., Krane, S. M., Kenzora, J. E. & Glimcher, and is supported by the pronounced abnormality of skin and M. J. (1972) N. Engl. J. Med. 286, 1013-1020. I The 2. Krane, S. M., Pinnell, S. R. & Erbe, R. W. (1972) Proc. bone collagens, which are essentially Type collagen. ge- Nat. Acad. Sci. USA, 69, in press. netic defect responsible for the disease might inhibit hy- 3. Bailey, A. J. & Peach, C. M. (1968) Biochem. Biophys. Res. droxylation of lysyl residues only in Type I collagen, but im- Commun. 33, 812-819. pair this collagen in all connective tissues in which it is located. 4. Mechanic, G. & Tanzer, M. L. (1970) Biochem. Biophys. The presence of a low level of hydroxylysine in collagen ex- Res. Commun. 41, 1597-1604. 5. Bailey, A. J., Fowler, L. J. & Peach, C. M. (1969) Biochem. tracted from abnormal cartilage in 4 M CaCl2 suggests that Biophys. Res. Commun. 35, 663-671. the more readily extractable Type I collagen (17-19) is indeed 6. Bailey, A. J., Peach, C. M. & Fowler, L. J. (1970) Biochem. deficient in hydroxylysine. J. 117, 819-831. Alternatively, it can also be reasoned that a slight de- 7. Tanzer, M. L. & Mechanic, G. (1970) Biochem. Biophys. ficiency in hydroxylysine, as detected in the cartilage collagen, Res. Commun. 39, 183-189. 8. Mechanic, G., Gallop, P. M. & Tanzer, M. L. (1971) Bio- could markedly affect the pattern of reducible components if chem. Biophys. Res. Commun. 45, 644-653. the deficiency were limited to those sites participating in cross- 9. Bailey, A. J., Peach, C. M. & Fowler, L. J. (1970) in Chem- linking of the collagen molecule. istry and Molecular Biology of the Intercellular Matrix, ed. Although an attractive and plausible explanation, it cannot Balazs, E. A. (Academic Press, New York & London), be the present observations that defects in Vol. I, pp. 385-404. concluded from 10. Bailey, A. J. & Shimokomaki, M. S. (1971) FEBS Lett. collagen crosslinks alone are responsible for the clinical mani- 16, 86-88. festations of the disease. In addition to its role in crosslinking, 11. Bailey, A. J. (1971) FEBS Lett. 18, 154-158. hydroxylysine is the site of sugar attachment to collagen mole- 12. Tanzer, M. L., Mechanic, G. & Gallop, P. M..(1970) Bio- cules. Impairment of collagen , which must occur chim. Biophys. Acta 207, 548-552. 13. Eyre, D. R. & Glimcher, M. J. (1971) Biochim. Biophys. in collagen devoid of hydroxylysine, may be equally responsi- Acta 243, 525-529. ble for the clinical defects observed in these patients. Indeed, 14. Kang, A. H., Faris, B. & Franzblau, C. (1970) Biochem. as-pointed out by Prockop (22), it is not clear whether gly- Biophys. Res. Commun. 39, 175-182. cosylation of a minimum number of hydroxylysine residues is 15. Tanzer, M. L. & Mechanic, G. (1968) Biochem. Biophys. essential for the cellular extrusion of collagen at a normal Res. Commun. 32, 885-892. 16. Furthmayr, H. & Timpl, R. (1971) Anal. Biochem. 41, 510- rate. 516. About one-third of the hydroxylysine in collagen of normal 17. Trelstad, R. L., Kang, A. H., Igarashi, S. & Gross, J. (1970) human skin and bone is glycosylated (23). Preliminary anal- Biochemistry 9, 4933-4998. yses were possible on a limited amount of material remaining 18. Miller, E. J. (1971) Biochemistry 10, 1652-1658. 19. Seyer, J. M. & Brickley, D. (1972) J. Bone Joint Surg., from biopsies of diseased skin, bone, and cartilage from one of in press. the patients (Case 1, see ref. 2), for glycosylated hydroxylysine 20. Mechanic, G. (1972) Biochem. Biophys. Res. Commun. residues by the method of Pinnell, Fox, and Krane (23). A 47, 267-272. single analysis showed that the trace of hydroxylysine present 21. Julkunen, H., Rokkanen, P. & Inoue, H. (1971) Ann. Med. in skin collagen was at last 52% glycosylated as the di- exp. Fenn. 48, 201-204. 22. Prockop, D. J. (1972) N. Engl. J. Med. 286, 1055-1056. saccharide derivative, glucosylgalactosylhydroxylysine. Ga- 23. Pinnell, S. R., Fox, R. & Krane, S. M. (1971) Biochim. lactosyl hydroxylysine may have been present, but could not Biophys. Acta 229, 119-122. Downloaded by guest on September 30, 2021