Biochem. J. (1978) 173, 617-625Printed in Great Britain

A Structural Model for Desmosine Cross-Linked Peptides

By ROBERT P. MECHAM* and JUDITH A. FOSTERDepartment ofBiochemistry, Interdepartmental Medical Biochemistry Section,

Boston University School ofMedicine, Boston, MA 02118, U.S.A.

(Received 28 November 1977)

Desmosine-enriched peptides were isolated from a thermolysin digest of bovineligamentum nuchae elastin and a partial sequence was determined. A 'two-cross-link'model is proposed in which a second cross-link, perhaps lysinonorleucine, joins twopeptide chains approx. 35 amino acid residues removed from the desmosine. Implied inthis model is a certain asymmetry or directionality which places restrictions on the 'sense'of the peptide chains (either always parallel or anti-parallel) in order to align the cross-linking sites. Imposing such restrictions raises the possibility of specific alignment ofelastin precursor molecules by microfibrillar proteins and/or aligning peptides on theprecursor molecules themselves.

The lysine-derived cross-links in elastin (Franzblau,1971) in large measure contribute to the protein'selastic mechanical properties. Their formationinvolves enzymic oxidation of certain lysine c-aminogroups, forming a-aminoadipic acid 5-semialdehyde(allysine) (Lent et al., 1969). Once formed, thesereactive aldehyde groups are thought to condensespontaneously with another allysine molecule byan aldol condensation forming allysine aldol, or withthe E-amino group from an unoxidized lysine residuevia a Schiff-base reaction forming dehydrolysino-norleucine.The most complex of the elastin cross-links are

desmosine and isodesmosine (Partridge et al., 1963;Thomas et al., 1963; Pax et al., 1974). At present, thebiosynthetic route for these isomers is not clearlydefined, although synthetic processes involvingcondensation of dehydrolysinonorleucine with al-lysine aldol (Piez, 1968; Gray et al., 1973; Fosteret al., 1974) and a route involving dehydromero-desmosine as an intermediate (Francis et al., 1973)have been proposed.

Elastin cross-links are found in alanine-enrichedareas of the molecule and are formed from lysinemolecules separated by two or three alanine residues(Sandberg et al., 1971, 1972; Foster et al., 1974).It has been proposed by Gray et al. (1973) that thisclustering of alanine and lysine residues favours ana-helix conformation with the lysine side chainsappositely positioned on one side ofthehelix, allowing

Abbreviations used: SDS, sodium dodecyl sulphate;dansyl, 5-dimethylaminonaphthalene-1-sulphonyl.

* To whom reprint requests should be addressed.Present address: Pulmonary Division, Department ofMedicine, Washington University School of Medicineat the Jewish Hospital of St. Louis, St. Louis, MO 63110,U.S.A.

Vol. 173

for easy condensation of the juxtaposed cross-linkprecursors after enzymic oxidation. Circular-dichro-ism studies suggest that desmosine-enriched peptides,although enriched in alanine, are in an extendedhelix rather than an a-helix (Foster et al., 1976).Although the extended helix differs slightly from ana-helix, the separation of the lysine residues does notseem too distant for condensation of cross-linkintermediates (Wender et al., 1974).The present study was undertaken to determine the

primary sequence around elastin cross-linking sitesand to relate this information to possible structural'signals' which might modulate or otherwise directcross-link formation. Since cross-link formationappears to be an irreversible step in the formation ofmature elastin from soluble precursor molecules,information of the type obtained in this study is vitalto our understanding of elastin fibrogenesis.

ExperimentalMaterials

Chemicals and enzymes were obtained as follows:NaB3H4, New England Nuclear, Boston, MA,U.S.A.; thermolysin (three times crystallized; Agrade), Calbiochem, La Jolla, CA, U.S.A.;Aminex 50W-X4 ion-exchange resin, Bio-RadLaboratories, Richmond, CA, U.S.A.; type PChromobeads, Technicon Corp., Elmhurst, IL,U.S.A.; N-ethyl-N'-(3-dimethylaminopropyl)carbo-di-imide hydrochloride and 2-aminonaphthalene-1,5-disulphonic acid, Aldrich Chemical Co., Milwaukee,WI, U.S.A.

Preparation of elastinElastin from bovine ligamentum nuchae was

prepared by the autoclave method of Partridge &

617

R. P. MECHAM AND J. A. FOSTER

Davis (1955). The amino acid analysis (as describedbelow) of the purified product is given in Table 1.

Reduction with NaB3H4

Ligamentum nuchae elastin (5 g) was reduced bysuspending the protein in 250ml of 0.001 M-EDTA(pH9) and adding a mixture of 182mg ofNaBH4 and27mg of NaB3H4 (1OOmCi). The pH was maintainedat 9 with 0.01 M-HCl. After 2h, excess of NaB3H4was destroyed by adjusting the pH to 3 with 50%(v/v) acetic acid. The protein was filtered and washedthoroughly with 0.01 M-HCl, water, ethanol and ether,and allowed to dry.

Thermolysin digestion

Reduced elastin (5g) was suspended in 100ml of0.1 M-ammonium acetate (pH7.5) and 50mg ofthermolysin added. The mixture was allowed to digestfor 2-3h at 37°C, whereupon another 50mg ofthermolysin was added and the mixture leftovernight (37°C). The elastin had been completelysolubilized by the thermolysin at this point. Digestionwas stopped by adjusting the pH to 3 with acetic acid,at which point a white flocculent precipitate appeared,which was removed by centrifugation (l0000g,20min, 4°C).

Amino acid analysis

Samples for analysis were hydrolysed undervacuumfor 16-22h at 110°C with constant-boiling-pointHCl. Amino acid analyses were routinely performedby using a JLC 6AH automatic amino acid analyserwith a modified program to include a 0.4M-sodiumcitrate buffer, pH 5.3, for the resolution of the elastincross-linking amino acids desmosine, isodesmosine,merodesmosine and lysinonorleucine. Isodesmosine,reduced isodesmosine and reduced desmosine allco-chromatograph in this system. Since all cross-link-containing peptides reported in this work werederived from reduced elastin, all values for isodesmos-ine represent contributions from any or all of thesethree species. The values reported therefore referto all three species as 'desmosine', collectively.The cross-linking amino acids in elastin originate

from two to four lysine residues. Therefore, theirninhydrin colour yield must be calculated accord-ingly. In this study, allysine aldol, lysinonorleucineand dehydrolysinonorleucine are expressed as leucineequivalents divided by 2, merodesmosine anddehydromerodesmosine as leucine equivalents dividedby 3, and desmosine and isodesmosine as leucineequivalents divided by 4.

Alkaline hydrolysis was used to detect certaincross-links or cross-link precursor amino acids thatare destroyed by acid hydrolysis. The samples were

hydrolysed in alkali-resistant glass tubes with 0.2mlof 2M-NaOH at 110°C for 16-22h, after which thesample was diluted with water so that the Na+concentration was 0.2M. The pH was adjusted to 2with concentrated HCl and the sample applied to anion-exchange column (0.6cmx 150cm) packed withChromobeads typeA (Technicon Corp.). The columnwas equipped with a stream-splitting device whichallowed one half of the sample to be diverted toa fraction collector where 1 ml samples werecounted for radioactivity in a Packard model 3310liquid-scintillation counter, with Aquasol (NewEngland Nuclear) as the scintillation fluid. Theremaining sample was analysed with ninhydrin.The column was eluted with a sodium citrate gradientdescribed by Burns et al. (1965).

Peptidefractionation by ion-exchange chromatography

Aminex 50W-X4 resin (20-30 micron, H+ form),or type-P Chromobeads, was washed in a Buchnerfunnel successively with 5 or 6vol. each of 1 M-NaOH,water, 2M-HCl and water. Before use, the resin wasconverted into the pyridinium form by washing withseveral volumes of 2M-pyridine followed by a finalequilibration with 0.02M-pyridine/acetate, pH2.7.A 0.9cmx 60cm column was packed under anitrogen pressure of 75.7-113.6N/m2 (20-301b/in2)at 55°C. Samples were applied to the column instarting buffer and eluted with a continuous six-buffer pyridine/acetate gradient (Schroeder, 1967).The column pressure. was 75.7-113.6N/m2 with aflow rate of approx. 60ml/h; 5min fractions werecollected. At the end of the run, the column waswashed with 8.5 M-pyridine/acetate, pH 5.6.The strong u.v. absorption by the pyridine/

acetate buffer at wavelengths commonly used todetect proteins or peptides made it impossible con-tinuously to monitor the column effluent. Thereforepeptides were detected by ninhydrin after alkalinehydrolysis as described by Foster et al. (1974).

Gel filtrationGel-filtration chromatography on columns (1.5cm

xlOOcm) of Sephadex G-50 or G-25 (Pharmacia,Uppsala, Sweden) or Bio-Gel P6 or P2 (Bio-RadLaboratories) eluted with 0.1 M-acetic acid was usedto desalt and provide further separation of peptidecomponents after initial fractionation by ion-exchange chromatography. A column (1.5cm x60cm) of Sephadex LH-20 was also used with anelution buffer of 50% ethanol containing 0.5%benzene.

Purity of isolatedpeptidesSeveral methods were utilized to monitor the purity

of peptides throughout the isolation and purification

1978

618

DESMOSINE PEPTIDES

procedures. One of the methods was N-terminaldetermination by dansyl chloride (Gray, 1972). Dansy-lated end groups were analysed by t.l.c. on poly-amide layers.

SDS/polyacrylamide-gel electrophoresis was alsoused to monitor peptide purification. The methodof Weber & Osborn (1969) was followed, except thatthe gel buffer contained 0.5M-urea (Goldberg et al.,1972). Gels contained 10% acrylamide. Electro-phoresis was carried out at 2-3 mA/gel overnight.Gels were stained with Coomassie Brilliant Blue anddestained in 10% (v/v) methanol/7% (v/v) aceticacid.A peptide-'mapping' procedure was also used

which involved dansylating the peptide mixture bythe method described above for dansyl-end-groupdetermination and chromatographing the dansylatedpeptides on polyamide thin-layer plates. Afterdevelopment, the thin-layer plate was sectioned intogrids and the polyamide layer in each grid scrapedfrom the plate and counted for radioactivity inAquasol.

Sequencing procedureAutomated sequence analysis was carried out with

the Beckman model 890C sequencer by the procedureof Edman & Begg (1967). The resultant amino acidthiazolinone derivatives were converted into themore-stable amino acid phenylthiohydantoins byincubation in hot 1.OM-HC1 and extracted into ethylacetate. A portion (usually 10%) of each sequencerfraction in ethyl acetate was analysed by using aBeckman model 65 gas-liquid chromatograph bythe method of Pisano & Bronzert (1969). In mostcases, samples from the ethyl acetate and aqueousphases were also counted for radioactivity inAquasol. The remaining sample from both phaseswas then transferred to acid-cleaned hydrolysis tubesand hydrolysed to free amino acids by using thehydroiodic acid method of Smithies et al. (1971). Thesamples were then analysed with the amino acidanalyser.To minimize the extreme losses of the hydrophobic

elastic peptides during sequencing extraction pro-cedures (Braunitzer et al., 1970), the film-formingproperties and polarity of the peptides beingsequenced were increased by the attachment of2-aminonaphthalene-1,5-disulphonic acid to the C-terminal residue via a water-soluble carbodi-imideby the method of Foster et al. (1973).

Results

Characterization of the thermolysin digest

N-Terminal analysis of the thermolysin digest ofelastin with dansyl chloride revealed a multiplicityof amino groups, with glycine, leucine, alanine and

Vol. 173

valine being the most prevalent. Present in lower butnevertheless significant amounts were phenylalanineand isoleucine.

SDS/polyacrylamide-gel electrophoresis of thedigest indicated a mol.wt. of less than 15000 for mostof the peptides. Gel-filtration chromatography onSephadex G-50 showed peptide material beingeluted over the entire fractionation range of thegel (30000-1000 mol.wt. for peptides and globularproteins), with most of the material eluted near theregion corresponding to mol.wt. of7000 and below.

Acidic peptide precipitate

When the thermolysin digest was terminated by ad-justing the pH to 3 with acetic acid, a white flocculentprecipitate appeared. The amino acid analysis of thisprecipitate is given in Table 1. Quantitatively, theacidic precipitate comprised less than 5% of thetotal elastin digest. SDS/polyacrylamide-gel electro-phoresis of the precipitate showed that most of thematerial had a mol.wt. less than 15000. There were,however, two sharp bands of 48500 and 41000mol.wt.

Table 1. Amino acid analyses of purified ligament elastinbefore and after thermolysin digestion

Abbreviations: N.D., not determined; Hyp, hydroxy-proline; Des, desmosine; LNL, lysinonorleucine.

Amino acid content (residues/1000 residues)

Reduced ligamentelastin before

digestionLys 5His 0Arg 7Hyp 9Asp 8Thr 10Ser 10Glu 17Pro 119Gly 329Ala 221Cys TraceVal 139Met 0Ile 26Leu 62Tyr 5Phe 28Trp N.D.Des 1.4LNL 0.7

Thermolysin-digestedelastin after acidi-

fication and removalof the precipitate

416889916

114333223

Trace1420

26645

28N.D.

1.40.8

Acidicprecipi-

tate3223340

17174738025110912585

50518331

N.D.00

619

R. P. MECHAM AND J. A. FOSTER

2.4

2.2

2.0

1.8

1.6

1.4

> 1.2

1.0

0.8

0.6

0.4

0.2

EC)

0

-0

CZ

*5

x

01-

50 70 90 110 130 150 170 190 210 230Tube no.

1 2 3 4 151617 18191101111121 13 1141151161 17 18Fraction no.

Fig. 1. Elution profile ofthermolysin-digested elastin from an Aminex 50 W-X4 column (0.9 cm x 60cm)The column was eluted at 55°C with a continuous pyridine/acetAte gradient (Schroeder, 1967) at a flow rate of60ml/h followed by a column wash with 8.5M-pyridine/acetate (pH5.6). Peptide material was detected by allowing asample from each tube to react with ninhydrin after alkaline hydrolysis, and measuring the A570 ( ). Radioactivity(----) was determined by liquid-scintillation counting. Peptide fractions were pooled as indicated, i.e. 1-18.Fraction 19 (not shown on the Figure) is the material eluted with the column wash.

Ion-exchange chromatographyTo effect better separation of peptides than that

obtained with gel filtration, 50mg of the reducedelastin digest was chromatographed on an Aminex50W-X4 ion-exchange column. In later experiments,the resin was changed to Technicon type-PChromobeads, which afforded slightly better resolu-tion and lower column pressures. Fractions from theAminex column were designated with the prefix'TA' followed by the fraction number. Fractionsfrom the Technicon column were designated 'Th,t'.Fig. 1 shows the typical elution profile of a sample ofthe elastin digest chromatographed on the ion-exchange column. Table 2 gives the amino acidanalyses of the pooled fractions. Amino acid analysisof the starting material plus summation of the totalnmol from each fraction indicated 90-95% recoveryfrom the ion-exchange column. It is noteworthy thatmost of the radioactivity, and hence cross-links(fractions TA-11-TA-19 inclusive), were separatedquite well from the fractions that did not containcross-links (TA-1-TA-10). To obtain enough peptidein each fraction for purification and subsequentanalysis, many runs of 50-100mg of the digest wereseparated on the ion-exchange column and similarfractions between runs pooled.

Desmosine- and lysinonorleucine-enrichedpeptides

Peptides enriched in desmosine, isodesmosine,merodesmosine and lysinonorleucine were elutedfrom the ion-exchange column over a relativelybroad range at the high-salt end of the pyridine/acetate gradient (fractions TA-16-TA-19 and Th,t-12-Th,t-14). The radioactivity profile showed minorbroad peaks which served as guides for poolingfractions, which were then freeze-dried and chromato-graphed on Sephadex G-50, G-25, LH-20, or Bio-GelP6 or P2. Fractions containing the highest content ofdesmosine were pooled and rechromatographedseparately. The compositions of the desmosine-enriched peptides are given in Table 3.

Sequence analysis of desmosine-enrichedpeptides

Peptide TA-17,G-50-II and a sample from peptideTA-1 8,G-50-IV,LH20-II were sequenced withoutmodification, but yielded no reliable results owing topeptide washout. The results of sequencing themodified peptide Th,t-11,P6-1,P2-1 is shown in Table4. An indication of the purity of this peptide wasascertained by dansylatlion and subsequent chromato-graphy as described in the Experimental section,which yielded only -one radioactive and fluorescent

1978

620

DESMOSINE PEPTIDES

r- en en eq r 7N 0 (7o a m woo "N £0%£ t_ _ N o _rc-. I : oi0 _ Cd

1. tn (7o -t- en "o O ,_ Nt r 08 r o

oo 11 enz \0mP-_o ff I- tn N W s o r mOr 6% -t r-bo-oo en F Oo I

00 _r.***e a ) I en d

~I_0 t

C1 'IC> tn aE Mt- l-tenl- C' t- t--0- t \£ _ \, WI\£ 06 l- 00 Ct~ _: CN 0s r 42: o t-- Ct 4 oo _-: 1r 4 _.I"It - N.0' ftt (7~00_% eTe~~o6r-=-4 4

InN_

4 I" I

t_0 N N oo

z-Cl_lr

t- o0 t-en000V

z061 1 2 O

e I'D 00 C4 enoI I I I el;a4 I

I I

~~~~~~~~~~~~~~~v Fb00 ~v e

en

b imt t iI I II tt

_ 0 0,_ t _

en oen tn in nIIII_

N m Naf m

cli CN, el; 06 W;Cl0 0 en ON04

0% Cl4- "o

'%0 Cl000 -0I '44V , 6 1r- C ~I, II_- tn lt q a-, rl 'I

ooN IemnI"

0a 0 m N 0s C m

as t- o -^o

encr Ct ,

I'd,0000 I'llN nNVC A00-4

tN j_e" NIt r. r.o6\_C4on

N e 00 - N

III I1 , eo00 ON or) en

Ico~-~ro6--~14I116 1 1 111Ien"

o_ o IkD II'oo .0Cl |_|'

I I I<DO tWI tef t

e8 ClN0%fi o0~O~-It enI00 e o_III

O~~~~o ,6 o6 een r-6 4 f

I III m-ob I vt}oo-e Ne 0t -0 O0t 0NO-

et~ ~ ~ , 00mr

- iii lia-ii Ci -iW;0

Itr-a I' a)a0>- 0 '

.~ ~~~~ir- C a);- - 0 40c ~ ~ ~ ~ cq '.e-Oc t q "

621

al t- 00 0 "l-

tn 't W)£- 's-t a W

I _N000C N- t I -00-_ N en

a)

cdc

C0e7o

rA '

Cd 0

'0

C.i0

E

0

a)0

'0

>~a

Vol. 173

I

I

II

R. P. MECHAM AND J. A. FOSTER

Table 3. Amino acid compositions ofdesmosine-enrichedpeptidesValues for desmosine (Des) are leucine equivalents divided by 4, lysinonorleucine (LNL)equivalents divided by 2. For peptide nomenclature, see the text.

Amino acid content (residues/peptide, assuming Des = 1)

LysArgSerGluProGlyAlaValLeuTyrPheLNLDes

TA-17,G-50-II0.70.60.10.42.95.3

19.71.01.20.80.41.01.0

TA-18,G-50-4,LH20-II

0.40.30.80.42.73.7

12.91.01.30.90.50.41.0

Th,t- 1,P6-1,P2-11.10.50.71.03.08.3

20.21.52.41.20.30.81.0

Th,t-12,P2-1,G-50-II

0.60.30.60.54.07.1

21.01.61.81.00.51.01.0

values are leucine

Th,t-13,P2-11.70.72.61.68.1

14.328.63.72.91.81.20.91.0

Table 4. Sequential degradation of desmosine-enrichedpeptide Th,t-1 1,P6-1 ,P2-1

Desmosine was determined qualitatively by theappearance of radioactivity at step 6. Quantificationof the phenylthiohydantoin derivative of reduceddesmosine(*) proved to be a difficult problem (see theResults section). Total desmosine in the startingmaterial was 360nmol.

Sequencer step no.1

2

34

5

6

7

Amino acid released (nmol)AlanineLeucineAlanineGlycineAlanineAlanineGlycineAlanineGlycine

69032040225835030014016078

Alanine 40Desmosine *Tyrosine 30Alanine 22

Non-cross-linked peptidesMost of the ninhydrin-positive material was eluted

from the ion-exchange column as sharp peaks in theearly stages of the pyridine/acetate gradient (see Fig.1). Amino acid analysis of these fractions (TA-4-TA-10 and Th,t-2-Th,t-7) indicated high values forglycine, alanine, valine, proline and, in some cases,leucine. There was very little radioactivity in this partof the effluent, except for a radioactive peakprimarily in fractions TA-6 and TA-7, which wasanalysed and found to contain allysine. Sequenceanalysis of one of the peptides in TA-4 indicated apartial sequence of Val-Gly-Val-X-Pro/Hyp-Gly . . .,where X signifies an unknown residue and Pro/Hypindicates a partial hydroxylation of proline residues.The size of these non-cross-linked peptides is

small, since most of the material ran with thetracking dye in polyacrylamide-gel electrophoresis(10% gel). Gel-filtration chromatography indicatedthat most of the peptide material had mol.wts. lessthan 5000.

spot. Each sequencer fraction was counted for radio-activity (both the ethyl acetate and acid phases),analysed by g.l.c. and by using the amino acid analyserafter back-hydrolysis to free amino acids withhydroiodic acid. In those fractions where cross-linkamino acids were indicated by an increase in radio-activity, the effluent from the amino acid analyser wascollected and counted for radioactivity so that onecould relate the counts to the desmosine peak on thechromatogram.

Discussion

One of the major obstacles to a thorough study ofelastin has been the protein's extreme insolubilityand relative resistance to solubilization by enzymeswith limited specificities, such as chymotrypsin andtrypsin. Relatively harsh chemical methods or non-specific enzymes have of necessity been used to effectsolubilization.

In this study, ligamentum nuchae elastin wascompletely solubilized by thermolysin, an enzymehaving activity directed towards the N-terminalpeptide bond of amino acids with hydrophobic sidechains, such as leucine, isoleucine, valine, phenyl-

1978

622

DESMOSINE PEPTIDES

alanine and alanine. Most of the elastin peptidesgenerated by thermolysin are small, however, havingmol.wts. less than 7000.An important question about the primary structure

of desmosine-cross-linked peptides is the relativeposition of the two isomers desmosine and iso-desmosine in the sequence. Foster et al. (1974)showed that both desmosine and isodesmosine canoccur in the same sequence position. Davril & Hahn(1974), however, suggest that they occur at separatesites close together on the peptide chains. Data in thepresent paper do not help discern a correct answer,since the cross-linked peptides being studied have beenreduced, and, as mentioned above, reduced desmosineand isodesmosine co-chromatograph on the aminoacid analyser, making positive identification of theisomers impossible.

Sequencing cross-linkedpeptidesA major difficulty encountered in sequencing

elastin peptides is the drastic loss of peptidematerial from the automated sequencer reaction cupduring the various sequencer extraction steps. Suchwashout is a common occurrence for hydrophobicpeptides, which are somewhat soluble in the organicsolvents used in the sequencer (Braunitzer et al.,1970). In a previous report, washout reported whensequencing non-reduced desmosine-enriched peptideswas not as great as that encountered in the presentstudy (Foster etal., 1974). It appears that reduction ofthe pyridinium nucleus of the desmosine cross-link,and hence neutralization ofthe positive charge on thering, increases the hydrophobicity ofthe peptide.

In addition to the peptide losses caused bywashout, resulting in very low repetitive yields,sequencing cross-linked peptides presents otherdifficulties. When two or more chains covalentlylinked in one or more locations are sequencedsimultaneously, the cross-linking loci may not bedetectable as a phenylthiohydantoin derivative untilthe final attachment site is cleaved. Also, if the cross-link is linking two or more chains, the interpretationof the sequencing data will be complicated by twoor more amino acids being sequenced per step,compounded by the degree of carry-over and thefact that different peptides will be released from thecross-link at different times. It appears that peptidelosses through the extraction cycles are increasedonce a peptide segment is released from the cross-link.For a discussion of the problems and strategies in-volved in sequencing cross-linked peptides, seeFoster et al. (1974).

Desmosine-enrichedpeptidesSince desmosine and isodesmosine originate from

four lysine residues, it is possible for them to linkfour, three or two polypeptide chains. Sequence data

Vol. 173

indicate, however, that desmosines links only twochains, not three or four (Foster et al., 1974; Gerber& Anwar, 1974, 1975).Table 3 summarizes the amino acid compositions

3.0

._

'- 2.00

u

41.80

1 2 3 4 5Sequence step number

6 7

Fig. 2. Sequencer repetitive-yield plot for cross-linked andnon-cross-linked elastin peptides

The repetitive yield, calculated from the slope of theline generated in a plot of the log of the sequencerrecovery at step n against n, for the two arms of thecross-linked peptide Th,t-l l,P6-l,P2-1 is 830' (o)and 84% (o). The coefficient of determination forthe least-squares linear regression fit (r2) is 0.89 and0.98 for the two peptide chains respectively. For thenon-cross-linked peptide TA-4 (A), the repetitiveyield is 80% and r2 = 0.95. The segmented line showsthe fall in yield after the desmosine cross-link isreleased from the peptide chain. Pro/Hyp representsproline and hydroxyproline occurring at the samesequence position.

623

R. P. MECHAM AND J. A. FOSTER

Sequence Sequence

Ala-Ala-X-Ala-Ala-X Leu-GI

\ ~/Desmosin

Ala-X-Ala-Ala-Ala - Ala-Ala-(

Sequence

Fig. 3. Proposed modelfor desmosine peptide in which a second cross-link joins the two peptide chains approx. 35 amino acidresidues removedfrom the desmosine

The second cross-link need not be desmosine and in fact could be lysinonorleucine, which was present on a molar basisequal to desmosine in the peptide that was sequenced. The second cross-link may serve to hold and align the peptidechains, so that desmosine formation can take place at the appropriate site. Three sequencing sites are indicated, whichcorrespond to the peptide sequences described in Table 4. X is a modified lysine residue which constitutes an arm of thecross-link.

of the desmosine-enriched peptides isolated in thepresent study. Of note is the presence of lysinonor-leucine in quantities approximately equal to thedesmosine content in four of the five peptides.

Table 4 contains the sequence data for peptideTh,t-1 1,P6-1,P2-1. Both the acid and ethyl acetatephases from the sequencer conversion and extractionsteps were analysed on the amino acid analyser afterback-hydrolysis to free amino acids. Quantificationof desmosine after this procedure proved to be a

difficult problem in that values from the analyser weremuch lower than expected. Relative desmosineconcentration per sequencer fraction was bestdetermined by radioactivity before hydrolysis.The first step of the Edman degradation should

yield all amino acids with a free N-terminal in an

amount approximately equal to the total amount ofdesmosine. As shown in Table 4, the amount ofN-terminal leucine in step 1 is approximately equal tothe desmosine concentration, whereas alanine ispresent in twice the amount. It appears therefore thatthree chains are being sequenced, two chains withN-terminal alanine, and one with N-terminal leucine.The sequence data agree with those reported by Fosteret al. (1974) for similar desmosine-enriched peptides,except for the presence of the third sequence withleucine as the N-terminal residue.

Gerber & Anwar (1974) have described the partialsequences of seven peptides C-terminal to desmosinefrom bovine ligamentum nuchae and five peptidesfrom pig aortic elastin. The sequence Leu-Gly-Ala-Gly-Gly. . ., which was detected in sequencer steps1-5 in the present study, corresponds to one of thesepeptides. Its presence as a free contaminating peptide

seems unlikely, since it was evident in three of thedesmosine-enriched peptides, all of which werepurified differently. It appears therefore that thepeptide is bound to the desmosine peptide beingsequenced. This is further evident when repetitiveyield (average yield at each step) is considered.Repetitive yield is calculated from the slope of theline in a plot of log recovery at step n against n. Fromthe data in Table 4, the repetitive yield for theputative third chain with N-terminal leucine (seeFig. 2) is approximately equal to the repetitive yieldof the alanine chains (84%, cf. 83 %). If the peptideswere separate, it is very unlikely that their repetitiveyields would be equal, owing to differing degreesof peptide washout. A second factor to be con-sidered is the drastic loss of peptide material fromboth chains when the desmosine residue is removedby sequential degradation at step 6 (see Fig. 2).Such a loss was not evident with the non-cross-linked peptide TA-4.

Fig. 3 depicts our proposed model for the type ofdesmosine cross-link peptide sequenced in this study.The Leu-Gly-Ala-Gly-Gly... peptide is placed C-terminal to the desmosine cross-link, which fits thedata presented for similar peptides by Foster et al.(1974) and Gerber & Anwar (1974). This thereforenecessitates a second cross-link, not necessarilydesmosine, relatively close to the first linking thetwo chains. The N-terminal leucine is generated bythermolysin cleaving one of the intermediate chainson the N-terminal side of leucine directly C-terminalto desmosine.As mentioned above, the second cross-link in this

model need not be desmosine. In fact, it may be

1978

y-Ala-Gly-Gly-( ~)-X- ?

Cross-link

\ 'V 9

624

DESMOSINE PEPTIDES 625

lysinonorleucine, which was present on a molarbasis equal to desmosine in the peptide that wassequenced. The function of the second cross-link maybe to hold and align the chains so that desmosineformation can take place at the appropriate site.Dehydrolysinonorleucine or allysine aldol couldeasily serve such a function, owing to the simplecondensation reactions involved in their synthesis.The distance between the two cross-links can be

estimated from sequence work done in this laboratoryon copper-deficient insoluble elastin. We were ableto sequence a peptide homologous to the tropoelastintryptic peptide T9c reported by Foster et al. (1973),demonstrating an alanine-enriched region on theC-terminus (Mecham, 1976). Torres et al. (1977),studying peptides homologous to T9c from aa thrombin digest of tropoelastin, found a similarC-terminal alanine enrichment. The significance ofthis is that two possible cross-linking sites (alanine-enriched areas) are indicated in peptide T9c separatedby approx. 35 amino acid residues. It is thereforeproposed that the second cross-link (lysinonor-leucine?) is separated from the desmosine by approx.35 amino acid residues. Assuming two intermittentpeptides of this length, the total molecular weightof the desmosine peptide being sequenced should beapprox. 7000, which agrees with that of the peptidedetermined by gel filtration.The implications of this model are important.

Having to form two cross-links relatively closetogether implies a certain asymmetry or directionalityand places restrictions on the 'sense' of the peptidechains (either always parallel or always anti-parallel)in order to align the cross-linking sites. Imposingsuch a restriction raises the possibility of specificalignment by microfibrillar proteins and/or specificpeptide regions on the elastin precursor molecules.The acidic peptide detected on initial acidification ofthe thermolysin digest may serve such a function.

This work was supported by National Institutes ofHealth Grant HL-15964 and Training Grant HD-00207.The data presented are taken in part from the Ph.D.dissertation of R. P. M. J. A. F. is an EstablishedInvestigator of the American Heart Association.

References

Braunitzer, G., Schrank, B. & Ruhfus, A. (1970) Hoppe-Seyler's Z. Physiol. Chem. 351, 1589-1590

Burns, J. A., Curtis, C. F. & Kacser, J. H. (1965) J.Chromatogr. 20, 310-318

Davril, M. & Hahn, K. (1974) FEBSLett. 43,331-336Edman, P. & Begg, G. (1967) Eur. J. Biochem. 1, 80-91Foster, J. A., Bruenger, E., Hu, C. L., Albertson, K. &

Franzblau, C. (1973) Biochem. Biophys. Res. Commun.53, 70-74

Foster, J. A., Rubin, L., Kagan, H. M. & Franzblau, C.(1974) J. Biol. Chem. 249, 6191-6196

Foster, J. A., Bruenger, E., Rubin, L., Imbermin, M.,Kagan, H., Mecham, R. & Franzblau, C. (1976)Biopolymers 15, 833-841

Francis, G., John. R. & Thomas, J. (1973) Biochem. J.136, 45-55

Franzblau, C. (1971) Compr. Biochem. 26C, 659-711Gerber, G. E. & Anwar, R. A. (1974) J. Biol. Chem. 249,

5200-5207Gerber, G. E. & Anwar, R. A. (1975) Biochem. J. 149,

685-695Goldberg, B., Epstein, E. H., Jr. & Sherr, C. J. (1972)

Proc. Natl. Acad. Sci. U.S.A. 69, 3655-3659Gray, W. R. (1972) Methods Enzymol. 25B, 121-138Gray, W. R., Sandberg, L. B. & Foster, J. A. (1973)

Nature (London) 246, 461-466Lent, R. W., Smith, B., Salcedo, L. L., Faris, B. &

Franzblau, C. (1969) Biochemistry 8, 2837-2845Mecham, R. P. (1976) Ph.D. Dissertation, Boston Uni-

versityPartridge, S. M. & Davis, H. F. (1955) Biochem. J. 61,

11-21Partridge, S. M., Elsden, D. F. & Thomas, J. (1963)

Nature (London) 197, 1297-1298Pax, M. A., Pereyra, B., Gallop, P. M. & Seifter, S. (1974)

J. Mechanochem. Cell Motil. 2, 231-239Piez, K. A. (1968) Annu. Rev. Biochem. 37, 547-570Pisano, J. J. & Bronzert, T. J. (1969) J. Biol. Chem. 244,

5597-5607Sandberg, L. B., Weissman, N. & Gray, W. R. (1971)

Biochemistry 10, 52-56Sandberg, L. B., Gray, W. R. & Bruenger, E. (1972)

Biochim. Biophys. Acta 285, 453-458Schroeder, W. A. (1967) Methods Enzymol. 11, 351-361Smithies, O., Gibson, D., Fanning, E. M., Goodfliesh,

R. M., Bilman, J. C. & Ballantyne, D. L. (1971)Biochemistry 10, 4912-4921

Thomas, J., Elsden, D. F. & Partridge, S. M. (1963)Nature (London) 200, 651-652

Torres, A. R., Alvarez, V. L., Sandberg, L. B. & Gray,W. R. (1977) Adv. Exp. Med. Biol. 79, 267-276

Weber, K. & Osborn, M. (1969) J. Biol. Chem. 244,4406-4412

Wender, D. B., Treiber, L. R., Bensusan, H. B. & Walton,A. G. (1974) Biopolymers 13, 1929-1941

Vol. 173