Content of the collagen and elastin cross-links pyridinoline and

the desmosines in the human uterus in various reproductive states

Zeenat Gunja-Smith, Ph.D., and J. Frederick Woessner, Jr., Ph.D.

Miami, Florida

During pregnancy the collagen content of the human uterus increases sevenfold and the elastin content

increases fourfold to fivefold. The stable pyridinoline cross-link is found in uterine collagen at a level of 0.11 mol per mole of collagen. The same ratio, or a higher one, is found at the end of pregnancy,

indicating that pyridinoline synthesis keeps pace with the rapid synthesis of collagen. This cross-link would

participate in the maintenance of high mechanical strength of the uterus needed during parturition. Uterine elastin contains 2.4 residues of desmosine plus isodesmosine in 1000 residues of amino acids. This value

falls to 0.95 at term, indicating that synthesis of desmosines does not keep pace with the synthesis of elastin. Therefore, desmosine measurements do not provide an accurate index of elastin changes in

pregnancy. Collagen and elastin contents in nongravid uteri increase with successive pregnancies; the cross-links remain constant during this change. (AM J OBSTET GYNECOL 1985;153:92-5.)

Key words: Cervix uteri, collagen, elastin, pyridinoline synthesis, desmosine, cross-link

In 1963 we published a study of the collagen and elastin contents of the human uterus at term, during involution and in the nongravid state. 1 This study was based on a manual method for the determination of hydroxyproline and an autoclave method for the prep­aration of elastin. We have now reexplored this prob­lem with use of an automatic analyzer to determine the amino acids and with a sodium hydroxide step to in­crease the purity of the elastin fraction.

The most important aspect of the present study is the quantitation of the lysine-derived cross-linking compounds pyridinoline (3-hydroxypyridinium), iso­desmosine, and desmosine known to occur in collagen and elastin. In pregnancy there are very large increases in uterine collagen (sevenfold) and elastin (fourfold to fivefold). About half of this increase occurs in the last trimester.' It would be important to determine whether the cross-linking of these matrix macromolecules keeps pace with their rapid synthesis. Also, one would like to know whether the degradation of these polymeric ma­terials during postpartum involution of the uterus fa­vors those molecules that are not highly cross-linked.

From the Departments of Medicine and Biochemistry, University of Miami School of Medicine.

Supported by Grants AM-30195 and HD-06773 from the National Institutes of Health and from the American Heart Association, Florida Affiliate.

Received for publication January 18, 1985; revised May 14, 1985; accepted May 28, 1985.

Reprint requests: Dr. Zeenat Gunja-Smith, Department of Medicine, R-127, University of Miami School of Medicine, P.O. Box 016960, Miami, FL 33101.

92

Material and methods

Tissue. Human uteri were obtained at surgery and placed immediately in the refrigerator at 4° C. Further processing occurred within I to 48 hours. Each uterus was cut in half and weighed. One half was examined by the pathologist; the other half was used in these experiments if no disease was reported to be present within the body of the uterus. The cervix was removed at the level of the internal os. All further steps were carried out at 4° C.

Preparation of connective tissue fractions. Each uterus was cut into small pieces and a 6 gm portion was homogenized in the VirTis apparatus (VirTis, Gardi­ner, New York) in 60 ml of0.9% sodium chloride. The homogenate was centrifuged in a clinical centrifuge for 20 minutes. Any floating fibrous material was combined with the pellet, and the two were rehomogenized in JO volumes of 10% sodium chloride. After being stirred for 2 hours, the mixture was centrifuged. The pellet was resuspended and stirred overnight in I 0 volumes of 10% sodium chloride and then centrifuged a third time. The resulting pellet was washed twice with dis­tilled water and then twice with acetone over a 24-hour period. During the next 24 hours the pellet was washed with several changes of ethyl ether. The product was spread out to allow the ether to evaporate. It was then dried overnight in a vacuum oven at l 00° C.

Preparation of collagen and elastin. Samples of 50

to 150 mg of dry material, based on the previous es­timates of elastin content, were extracted twice in 25 ml water in an autoclave at 20 pounds per square inch

Volume 153 1'umber I

Collagen and elastin cross-links in human uterus 93

Table I. Collagen, elastin, and their cross-links in the human uterus

Collagen

µ,mol of Age Parity* Wet weight Collagen pyridinoline/

Uterus (yr) (n) (gm/uterus) (gm/uterus) uterus

Term 32 5 880 22.78 25.4 Term 29 3 970 23.96 9.3 Term 35 4 821 25.55 7.9 8 days post- 26 4 393 9.76 7.2

part um 9 days post- 26 4 163 8.71 3.04

part um Non gravid 27 0 44 2.18 1.33 Non gravid 18 2 57 3.21 I.OJ Non gravid 29 3 40 2.52 0.90 Non gravid 34 3 50 3.15 1.17 Non gravid 23 4 45 3.62 0.98 Nongravid 25 5 88 3.33 1.33 Nongravid 33 9 91 6.21 1.56 Post-meno- 80 52 0.76 0.32

pausal

*Term pregnancies. tDID = Desmosine + isodesmosine.

(1.36 atmospheres) for 3 to 4 hours each time. The hot mixture was briefly centrifuged, and the pellet was washed twice with hot water to remove the last traces of collagen and soluble proteins. The soluble extracts and washes were combined and lyophilized to provide the collagen fraction. The insoluble residue was also lyophilized. The dry residue was suspended in 5 to 7 ml of 0.1 mol/L of sodium hydroxide and heated to 98° C in a water bath with stirring for 30 min. The residue was washed with water until the washes were neutral and then lyophilized to provide the elastin fraction.

Determination of pyridinoline (3-hydroxypyridi­nium) and collagen. The dried collagen residues were hydrolyzed in 6 mol/L of hydrochloric acid for 24 hours at 105° Cina sealed tube under nitrogen. After removal of hydrochloric acid (by rotary evaporator), the residue was dissolved in water and applied to a phosphocel­lulose (H • form) column ( 15 cm by 1.5 cm) and eluted (2.5 ml fractions) with a linear gradient of water and 0.5 mol/L of hydrochloric acid.' Fractions were col­lected and scanned for fluorescence (excitation maxi­mum at 295 nm and emission maximum at 395 nm) in an Aminco Bowman spectrofluorometer. The appro­priate fluorescent fractions (Nos. 25 to 30) were evap­orated and quantitated on the cation exchange column with the Beckman Model 120 C amino acid analyzer.' The pyridinoline content was calculated as nanomoles by dividing the leucine equivalents by the factor 2.4. 5

Collagen content was determined by measuring hy­droxyproline in a separate run (in amino acid analyzer) and multiplying the weight of this amino acid by the factor 7.46.'

Elastin

mol of mol of D!Df/1000 pyridinolinel Elastin µ,mot of residues of

mol of collagen (gm/uterus) D!Dtluterus amino acids

0.33 3.96 36.0 0.92 0.12 2.40 23.6 0.96 0.09 2.53 24.1 0.95 0.22 1.48 25.5 1.73

0.10 I.OJ 18.I 1.80

0.18 0.23 4.08 1.64 0.09 0.72 15.4 2.09 0.14 0.58 13.9 2.40 0.11 0.62 13.6 2.15 0.08 0.60 16.6 2.80 0.12 0.80 19.2 2.38 0.08 1.32 32.5 2.44 0.12 0.33 7.07 2.16

Determination of desmosines and elastin. The elas­tin residue was weighed and hydrolyzed for 48 hours at 110° C in 6 mol/L of hydrochloric acid in a sealed tube under nitrogen. After the removal of hydrochloric acid, the residue was dissolved in 1.0 ml of water, fil­tered, and a portion analyzed on the amino acid ana­lyzer with use of the five-buffer system previously described6 for the separation and quantitation of des­mosine, isodesmosine, and hydroxyproline. In six of the samples, the desmosine peaks had shoulders that indicated the presence of interfering substances. Fur­ther aliquots of these samples were applied to a phos­phocellulose column (H • form) as described for the isolation of pyridinoline with one modification: after the completion of the water and 0.5 mol/L of hydro­chloric acid gradient, the column was eluted further with 0.5 mol/L of hydrochloric acid to bring out the desmosines (fractions 30 to 40). Appropriate fractions were rerun on the amino acid analyzer and symmetric peaks of desmosine and isodesmosine were found. Hy­droxyproline content was multiplied by the factor 62.5' to give an estimate of total elastin content. Desmosine and isodesmosine were calculated as nanomoles by di­viding the color expressed as leucine equivalents by the factor 3.6.7

Results The major findings are presented in Table I. Thir­

teen uteri were analyzed, and their collagen and elastin contents were compared to those found in the previous study.' The collagen values averaged 89% of those pre­viously reported-a difference attributed to the use of the automated analysis of hydroxyproline instead of

94 Gunja-Smith and Woessner

Table II. Ratios of amino acids in sodium

hydroxide residue of uterus at term compared to the ratios in authentic elastin and collagen*

Sodium hvdroxide

Ratio -residue Elastin Collagen

Proline/hydrox- 11.00 10.01 1.37 proline

Glycine/alanine 1.18 1.25 3.00 Glycine/valine 1.85 2.19 13.52 Valine/proline 1.06 1.08 0.19

*Comparisons are made for the term uterus at age 35 with use of the ratios of the numbers of residues. Data for elastin are the average of two determinations of elastin from human aortas of women of ages 19 and 44". Data for collagen are for type I collagen of human skin. 10

the manual method with dimethylaminobenzaldehyde.

The elastin values were lower, averaging only 80% of

those previously reported. This is attributed to the fur­

ther purification of the elastin fractions in the present

study by the use of hot sodium hydroxide to remove

remaining traces of collagen. This is a critical point,

since the elastin content is calculated from the hydroxy­

proline value. The collagen and elastin changes during pregnancy

and involution are essentially the same as those previously reported. 1 Three uteri at term contained

24. l ± l .4 gm (mean ± SD) of collagen compared to

3.5 ± l .3 gm for seven nongravid uteri; this corre­

sponds to a sevenfold increase in collagen during preg­

nancy. The elastin content increases from 0.69 ± 0.32

gm to 2.96 ± 0.86 gm at term or a 4.3-fold increase.

The resorption of the matrix is more than 60% com­

plete by 8 to 9 days after delivery. The major new findings concern the cross-linking

compounds. The fluorescent peak isolated from the hydrolysates of the collagen fraction (see Material and

methods) is found to have fluorescence excitation and

emission spectra identical with those of a pyridinoline

standard obtained from Achilles tendon collagen. This

is the first report of the presence of pyridinoline in

human uterus. It has been reported in the literature that free pyr­

idinoline is partially destroyed by autoclaving." We have

verified this but have also shown that pyridinoline in­

corporated in collagen is not destroyed. This was shown

by comparing the pyridinoline content of Achilles ten­

don subjected to autoclaving with that of tendon hy­

drolyzed directly in hydrochloric acid. Pyridinoline is found to increase in pregnancy in par­

allel with, or even faster than, the deposition of new

collagen. There were 14.2 ± 9.7 µmol of pyridinoline

per uterus at term compared to l.18 ± 0.23 µmol

in the nongravid state, or a twelvefold increase (p < 0.00 l, Student's t test). The pyridinoline content, ex-

September I, 1985 Am J Obstet Gynecol

pressed as moles per mole of collagen, was not signif­

icantly different at term from the nongravid value. This

indicates that the cross-link forms at least as rapidly as

the new collagen is deposited, maintaining a constant

proportion of cross-link regardless of the reproductive

state of the uterus. The content of desmosines (desmosine plus isodes­

mosine) behaves somewhat differently. It increases

from 13.8 ± 5.2 µmol per uterus for nongravid tissue

(omitting the sample of parity 9) to 27.9 ± 7.0 µmol

per uterus at term. This is only a twofold increase

(p < 0.0 I), whereas the elastin increased 4.3-fold based

on its hydroxyproline content. Thus cross-linking of

elastin did not keep pace with new synthesis during

pregnancy. It may be argued that the elastin at term

was not sufficiently pure and might have contained

excess hydroxyproline from traces of collagen, which

would give a calculated value for elastin that is erro­

neously high. This point was tested by looking at several

other amino acids (Table II). It can be seen that the

ratios of various amino acids to one another are very

close to those for elastin. If there were any admixture

of collagen, lower ratios of proline/hydroxyproline and

valine/proline would be sensitive indicators of this. It

must be concluded that the content of 0.95 residues of

desmosines per I 000 amino acid residues of elastin is valid and represents diminished cross-linking. The

seven nongravid uteri had an average of 2.3 residues

per 1000 residues, which may be compared to the 2.7

and 1.8 residues per 1000 residues reported for elastin

from the human female aorta9 and human cervix, 10

respectively. The effects of successive pregnancies can be deduced

by examining the seven nongravid specimens of Table

I. There was a generally increasing trend in weight

(twofold), collagen content (twofold to threefold) and elastin (fivefold to sixfold) as one goes from the nullip­

arous uterus to one with nine full pregnancies. This is

in agreement with the earlier findings. 1 The collagen

and elastin cross-links did not change with parity. A single 80-year-old uterus was examined. It showed

the typical postmenopausal loss of collagen. 11 However,

the cross-links of collagen and elastin were unaffected

by age. They neither increased in amount nor were

they selectively lost during the postmenopausal invo­

lution.

Comment

Perhaps the most surprising finding is that the pyr­

idinoline compound did not decrease at term but kept

pace with the increase in collagen content. Current

thinking is that the pyridinoline compound is formed

from the interaction of two reducible ketoamine resi­

dues (dihydroxylysinonorleucine) and that this process

occurs slowly with advancing age. 12 In the uterus at least

Volume 15'.> I\ umber I

half of the increase in collagen takes place in the final trimester of pregnancy. Therefore, the pyridinoline must form rapidly within this same time span. If pyr­idinoline is important for the mechanical strength of collagen, then it is essential that the links form before labor when the greatest mechanical forces will be placed on the collagen framework of the uterus. Since each pyridinoline is postulated to link three molecules of collagen together," a level of 0.1 mol per mole of col­lagen implies that about one third of the collagen mol­ecules of the uterus are involved in pyridinoline link­ages. Other, borohydride-reducible, cross-linking com­pounds such as hydroxylysinonorleucine, dihydroxy­lysinonorleucine, and histidinohydroxymerodesmo­sine, also found in human uterus,"' would further en­hance the mechanical strength.

The failure of elastin to develop the full complement of cross-links at term may be due to a failure of the activity of lysyl oxidase to keep pace with the synthesis of elastin. In embryonic tissue many weeks are required to produce fully cross-linked elastin." On the other hand, it may be advantageous to have a partially cross­linked elastin to facilitate its removal in the postpartum period. As the elastin is resorbed, the weakly cross­linked elastin is removed first so that, by 8 to 9 days, the remaining elastin has a content of desmosines more closely approaching that of the nongravid uterine elas­tin (Table I). The variable content of desmosine in uter­ine elastin indicates that elastin cannot be estimated in the pregnant uterus solely on the basis of desmosine quantitation.

We gratefully acknowledge the excellent technical as­sistance of Ms. Minny Chang and Ms. Marie Selzer.

REFERENCES

I. Woessner JF Jr, Brewer TH. Formation and breakdown of collagen and elastin in the human uterus during

Collagen and elastin cross-links in human uterus 95

pregnancy and post-partum involution. Biochem J 1963;89:75.

2. Morrione TG, Seifter S. Alteration of the collagen content of the human uterus during pregnancy and postpartum involution.J Exp Med 1962;115:357.

3. Fujimoto D, Moriguchi T. Pyridinoline, a nonreducible crosslink of collagen. Quantitative determination, distri­bution and isolation of a crosslinked peptide. J Biochem (Tokyo) l 978;83:863.

4. Gunja-Smith Z, Boucek RJ. Collagen cross-linking com­pounds in human urine. Biochem J 1981; 197:759.

5. Eyre DR. Collagen cross-linking. In: Akeson WH, Born­stein P, Glimcher MJ, eds. American Academy of Ortho­paedic Surgeons symposium on heritable disorders of connective tissues. St. Louis: CV Mosby, 1982:43.

6. Gunja-Smith Z, Boucek RJ. Desmosines in human urine: amounts in early development and in Marfan's syndrome. BiochemJ 1981;193:915.

7. Starcher BC, Galione MJ. A large-scale procedure for purification of desmosine and isodesmosine. Prep Bio­chem 1975;5:4.')5.

8. Fujimoto D, Moriguchi T, Ishida T, Hayashi H. The struc­ture of pyridinoline, a collagen crosslink. Biochem Bio­phys Res Commun 1978;84:52.

9. Sandberg LB. Elastin structure in health and disease. Int Rev Connect Tissue Res 1976; 7: 160.

I 0. Leppert PC, Keller S, Cerreta J, Hosannah Y, Mandi I. The content of elastin in the uterine cervix. Arch Biochem Biophvs I 983;222:53.

11. Woessner JFJr. Age-related changes of the human uterus and its connective tissue framework. J (;crontol 1963: 18:220.

12. Eyre DR, Oguchi II. The hydroxypyridinium crosslinks of skeletal collagens: their measurement, properties and a proposed pathway of formation. Biochcm Biophys Res Comm 1980:92:403.

13. Eyre DR. Collagen: molecular diversity in the body's pro­tein scaffold. Science 1980;207: 1315.

14. Kao KYT, Hitt WE, Leslie JG. The intermolecular cross­links in uterine rnllagens of guinea pig, pig, cow and hu­man being. Proc Soc Exp Biol !\led 1976;151:385.

15. Partridge SM. Biosynthesis and nature of elastin struc­tures. Fed Proc 1966:25: 1023.

16. Bornstein P, Piez KA. A biochemical study of human skin collagen and the relation between intra- and inter-molec­ular cross-linking . .J Clin Invest 1964:43: 1813.