IN VITRO Vol. 11, No. 5, 1975
COLLAGEN SYNTHESIZED AND MODIFIED BY AGING FIBROBLASTS IN CULTURE MERCEDES A. PAZ1 AND PAUL M. GALLOP The Department of Biological Chemistry and Orthopaedic Surgery, Harvard Medical School and Harvard School of Dental Medicine, and The Children's Hospital Medical Center, 800 Longwood Avenue, Boston, Massachusetts 02115
SUMMARY Collagen is produced by WI-38 diploid human fibroblast cultures throughout their life cycle. It is examined by a sensitive method based on the analysis of specific peptides obtained after digestion with bacterial collagenase. The production and hydroxylation of the collagen is strongly dependent upon the age (population doublings) of the culture and the presence of ascorbic acid. Young cultures (passage 26) produce large amounts of collagen in the absence of ascorbic acid, and this collagen is about 50% hydroxylated compared to that produced by young cultures in the presence of ascorbic acid. Ascorbic acid reduces to about one-half the amount of collagen produced by these young cultures. The young confluent cultures also depend strongly on ascorbic acid for hydroxylation of proline. The dependence declines rapidly with the age of the culture. The collagen produced by young cultures supplied with ascorbic acid is very similar to the type I collagen produced by normal individuals and has about the same degree of hydroxylation of its prolyl residues. The amount of collagen produced by "older" cultures is unaffected by ascorbic acid, but the degree of hydroxylation is normal only if ascorbic acid is present, and is decreased to about 60 to 70% in the absence of the vitamin. "Senescent" cultures showed little, if any, dependency on ascorbic acid, and the collagen produced, with and without the vitamin, is about 80% hydroxylated. The prolyl hydroxylation system of the WI-38 cells and the various controls on the SySr tern are age-dependent. Key words: collagen; fibroblast; hydroxylation; culture; ascorbic acid; age-dependence. Fibroblasts and other cell types such as ehondroblasts, smooth muscle cells, and epithelial cells can synthesize collagen. There is evidence that transformation of certain fibroblastic cells results in established cell lines that continue to produce and hydroxylate collagen (1-5). For this study, we have chosen a human diploid fibroblast strain WI-38 derived from embryonic lung tissue. This system displays the reproducible limited cell doubling phenomenon described by Hayiiick and Moorhead (6). It has been suggested that WI-38 fibroblasts may serve as an experimental model of aging (7, 8). This cell strain and related strains have been employed
for a wide variety of investigations which attempt to elucidate the mechanism of "programmed" death in culture. Several investigators have been considering the possibility that this system may represent an experimental model to test the Orgel hypothesis (9), in which aging is considered to be a consequence of accumulated errors in the protein synthetic apparatus. These errors are presumed to occur in the transcriptional and/or translational processes; some errors could accumulate and thereby promote additional increases in error frequency with each cell doubling. Such errors and their consequences may be expressed in a more or less random manner, or in some specific sensitive loci in the genetic material of the cells. In the latter case, 1 To whom reprint requests should be sent at the Department of Orthopaedic Surgery, Children's they could be considered analogous to somatic Hospital, 300 Longwood Ave., Boston, Mass. 02115. mutations. 302
COLLAGEN AND AGING FIBROBLASTS Modified proteins that are the end products of such events might be expected to show increasing imperfections with cell doublings until a lethal situation is attained at the so-called "error catastrophe stage." This stage is expected at phase I I I in the life of the diploid cell culture. As phase I I I is approached, less perfect, albeit functional, proteins may be found which might be more heat labile, and might also demonstrate a broadened or modified substrate specificity or an altered response to regulatory agents. Heatlabile glucose-6-phosphate dehydrogenase and catalytically inactive lactic dehydrogenase accumulate in an aging diploid cell strain derived from embryonic lung (7, 8), similar to the WL38 strain. This may be a first experimental confirmation of the Orgel hypothesis, providing valuable, but still controversial, evidence supporting the concepts of Hayflick (10) and others (11, 12), which relate limited doubling potential to cellular aging. We have selected as our test system aging WI-38 fibroblasts. After increasing numbers of doublings, we examined the continued potential of these cells to produce enzymes that can correctly modify precursor collagen at various post-translational steps. Collagen is a large structural protein produced by fibroblasts and other cells. Precursor collagen is subject to several specific post-translational modifications that occur both within the cell and after extrusion of the protein from the cell. We have focused on the specific collagen-hydroxylating systems, namely the two mixed function oxidase systems, prolyl hydroxylase and lysyl hydroxylase. Both of these hydroxylase systems require ferrous iron as co-oxidant and molecular oxygen as oxidant, and a-ketoglutarate as an obligatory co-substrate and co-reductant (13, 14). Aseorbate appears here, as in many other mixed function oxidase systems, to be an effective mediator, but its mode of action is not completely understood. In addition to its possible mediator role in the hydroxylation reaction, ascorbate may be an activator of an inactive form or precursor of prolyl hydroxylase (2). In this investigation, collagen synthesis, collagen prolyl residue hydroxylation, and the role of ascorbic acid are examined in "young," "older," and "senescent" cultures. A new and sensitive procedure is introduced, which employs bacterial collagenase for digestion of collagen, and mapping of the peptides with examination
303
of the ratio of two important and specific collagen tripeptide fractions. This can accurately define the level of hydroxylation, and allow one to measure relatively small changes in the levels of collagen hydroxylation. We have examined as a function of age the effect of ascorbate on collagen in the culture medium and collagen which remains within and around confluent cells with regard to synthesis and prolyl hydroxylation. Later experiments will be concerned with hydroxylysine formation and distribution, collagen carbohydration, and cross-linkage maturation in the collagen produced by the aging WI-38 cultures. MATERIALS AND METHODS
The WI-38 diploid fibroblast cell strain em: ployed was initially derived from female embryonic human lung. Starter cultures were obtained from Professor L. Hayflick of Stanford University. These cultures were usually received at the 16th or 17th passage level and were grown in 75-cm2 plastic tissue culture flasks (T-75). Cells were transferred at weekly intervals at a 1:4 split ratio. General procedures were similar to those used by Hayflick and Moorhead (6). The growth medium consisted of BME (basal medium Eagle, Earle's salts; GIBCO), supplemented with 10% fetal calf serum (GIBCO) and 50 ttg per ml of aureomycin (Lederle, product 4691=96 intravenous). Cultures were grown at 37~ in an atmosphere of 5% C02 and 95% air, and were monitored routinely for mycoplasmal contamination by Dr. L. Hayflick. Cells were trypsinized and counted, using a hemocyt0meter, after the cultures had reached conftuency. Total protein and DNA contents were determine d9 just after the cultures had reached confluency and after 1 week at confluency. The procedures of Lowry et al. (15) and a modification of the biuret reaction (16) were used for protein determinations. The diphenylamine reaction (17) was used to determine DNA content. Collagen concentration was determined by amino acid analysis of the collagenase-digested material after hydrolysis with 6 N HC1 at 110~ for 22 hr. The amount of collagen present was calculated from the hydroxyproline value, assuming for a normal hydroxylated collagen an average of 100 residues of hydroxyproline per 1,000 residues, and an average residue molecular weight of 92. The calculated amount
304
PAZ AND GALLOP
of collagen was corrected for the degree of underhydroxylation. I)ROCEDURE
Forty T-75 flasks of WI-38 cells at the desired passage level were used. They were divided into two groups of 20 flasks each as soon as the cells reached confluency. One group received complete BME medium supplemented with 50 t~g per ml of L-ascorbic acid, the other received complete BME medium without the vitamin. Both groups of cells were kept for 1 week at coniiuency. The medium was changed and collected every other day. Twenty-four hours after the third change of medium, the cells were refed with 20 ml of fresh medium containing 2,3-L-[3H]proline (New England Nuclear, 39.7 Ci per mmole, 0.5 ~Ci per ml). After a labeling period of 24 hr, the medium was separated from the cells and pooled with unlabeled medium from the respective group of cultures. After extensive dialysis against distilled water, the medium was lyophilized and saved for further processing. The cells were washed three times with cold saline and were scraped off the surface with a rubber policeman. Following centrifugation, the cell pellet was used for DNA, protein, and collagen determinations. In the present study, the degree of hydroxylation of the collagen was measured during a 24-hr labeling period and was assumed to be representative of the entire week at confluency. The amount of collagen determined in the medium and cell layer represents collagen actually accumulated during 1 week at confluency (balance of synthesis and degradation). Collagen was extracted from the cells and medium with hot trichloroacetic acid (TCA) followed by collagenase. The TCA completely disrupts the cells and denatures the collagen, allowing a total and efficient digestion by collagenase. Following this procedure, a second digestion with collagenase was unnecessary.
Preparation of the fraction soluble in hot trichloroacetic acid. The cell pellet and the lyophilized medium were heated in a final concentration of 5% TCA for 15 min at 70~ The preparations were then centrifuged at 7000 • g for 15 to 20 min, and the supernatant fluids were separated from the residues. Supernatant fluids. To digest the collagen extracted by the hot TCA solution with collagenase (Worthington Biochemical Corporation, CLSPA-
2000), it was necessary to eliminate most of the TCA present in the soluble fraction. This was achieved by three extractions with approximately three volumes of ether, using a separatory funnel. The soluble fraction was then lyophilized and the residue dissolved in 5 to 10 ml of 0.5 m~ CaCl~ and the pH adjusted to 7.5 with 1 N NaOH. Collagenase was added at a weight ratio of substrate :enzyme of approximately 50:1. The mixture was incubated in a shaking bath at 37~ overnight. Collagenase was separated from the collagen peptides by precipitation at 4~ with 5% TCA followed by centrifugation at 7000 • g for 20 rain. The supernatant fluid was again extracted three times with ether to remove TCA and chromatographed for characteristic peptides in a single column amino acid analyzer using a gradient of sodium citrate buffers. Residues. The fractions insoluble in hot TCA were washed three times by centrifugation with water and suspended in 0.5 mM CaC12 and the pH adjusted to 7.5 with 1 N NaOH. Following addition of collagenase, the suspensions were incubated at 37~ overnight and centrifuged at 7000 • g for 20 min. The supernatant fluid contained the collagen, solubilized by the collagenase, as peptides, and the residue contained proteins other than collagen. The separation of the collagenase from the supernatant fluids was achieved by precipitation with cold 5% TCA followed by centrifugation. The TCA was then extracted with ether following the procedure described above.
Separation of the collagenase-released peptides by column chromatography. The main tripeptides, gly-pro-hyp, gly-pro-ala, and gly-pro-pro, obtained by digestion of normally hydroxylated and underhydroxylated collagen with collagenase are commercially available (CycloChemical and Schwarz/Mann), and their positions of elution from the column of the amino acid analyzer can be established. Using a gradient of sodium citrate buffers similar to that used for the separation of amino acids, gly-pro-hyp elutes between alanine and valine, and gly-pro-ala elutes together with gly-pro-pro between methionine and isoleucine. Inasmuch as the collagen present in the TCAsoluble fractions, and in the fractions solubilized by treatment with collagenase, is labeled, the elution profile from the column can be followed by counting the radioactivity in the fractions in a scintillation counter. Two major peaks are ob-
COLLAGEN AND AGING FIBROBLASTS rained in the elution system employed. The first major peak is gly-pro-hyp, and also includes small amounts of a larger peptide with one prolinc but no hydroxyproline (peak A). The second major peak, gly pro-ala, chromatographs with giy-propro (peak B). The ratio of total dpm in peak A to total dpm in peak B, can be determined from the elution profile. This ratio is defined as the hydroxylation index, (H), and has an experimental value of 3.52 for skin or fibroblast collagen (type I) when it is normally hydroxylated. Such a value is also predicted from the known sequences of the a l chain of type [ collagen and the known specificity of collagenase. The fact that the predicted ratio of the type I collagen from a l and the found ratio of (al)2a2 coincided, indicates that the a2 chains could also be split by collagenase into a similar distribution of these characteristic peptides. Normally, a collagenase digest of an a l collagen chain would contain approximately 22 equivalents of gly-pro-hyp, 4 equivalents of a larger peptide containing 1 proline and no hydroxyproline, and 14 residues of gly-pro-ala. In the absence of full hydroxylation of the prolyl residues there would be a This ratio, H, measures the average hydroxylation of the second prolyl residue in those glypro-pro triplets where collagenase can split the bonds preceding the glycine and following the second proline. This represents 22 of ,9,8 total glypro-pro + gly-pro-hyp triplets, and over-all, this accounts for about 20% of the total hydroxyproline in the collagen a l chain. Accordingly, H may not accurately reflect the average degree of hydroxylation of all the proline residues that can be hydroxylated. However, other peptide fractions, for example, gly-leu-hyp, which chromatographs immediately before peak B in our system, and gly-ala-hyp which chromatographs before gly-prohyp in fractions 20 to 40, and can be clearly separated from other peptides with a slightly modified gradient, show similar behavior to gly-pro-hyp in that their value decreases with underhydroxylation. The composite a l chain of rat skin and calf skin collagen, containing 1052 residues, has 112 prolyl positions which can be found partially or fully hydroxylated, corresponding to an analysis of 106 hydroxyproline residues for 1,000 residues. The experimental value of 96 hydroxyproline residues per 1,000 residues for an a l chain indicates that about 90% of all the susceptible prolines are normally hydroxylated. The collagen produced by young fibroblasts cultured in ascorbic acid-supplemented medium, shows an experimental value of 3.5 for II, which indicates 100% hydroxylation of the gly-pro-hyp tripeptides released by collagenuse.
305
decrease in the amount of gly-pro-hyp, with an equivalent increase ill the amount of gly-pro-pro. Consequently, a change occurs in the ratio of peak A to peak B that is reflected by H the hydroxylation index. It is possible, for example, to determine that all experimental H of 0.7, as found ill the absence of ascorbic acid with cells from young cultures (see below), corresponds to a collagen with 49% of the normal prolyl residue hydroxylation. The following formula (a) can be derived for the percentage hydroxylation: 58(it)
-
4
% Hydroxylation = 100 44(H) + 44
(a)
where H is tile hydroxylation index determined experimentally from the ratio of dpm in peak A to that in peak B. This formula is derived as follows: there are 38 gly-pro-hyp triplets in an a l chain, of which 22 can be released as tripeptides by coIlagenase [the complete sequence of the a l chain is given in Hulmes et al. (20)]. The fractional hydroxylation of this peptide is X. In peak A there are also four residues of a peptide containing one equivalent of proline and no hydroxyproline. Then peak A has a total radioactivity proportional to (2 • 22 • X) + (1 • 4), where a factor of 2 is used because both proline and hydroxyproline are radioactive. In peak B there are both gly-pro-ala and gly-pro-pro. There are 31 gly-pro-ala triplets in an a l chain, of which 14 are released as tripeptides by collagenase. Thus, the radioactivity in peak B is proportional to (14 • 1) + (1 - X)(2 • 22) = 14 + 44 44X. Thus: II
peak A peak B
44X + 4 58 -- 44X
from which the percentage hydroxylation, 100X call be expressed in the formula (a) above. RESULTS
Figs. 1 to 3 show several of the elution profiles of labeled collagen after bacterial collagenase digestion obtained from the medium and from the cultures carried to different pas~ge levels. Cultures in pa~ages 26 and 41 originated from the same batch of young cultures, and were carried until they reached phase I I I at about passage 51. The passage classified as passage 41
306
PAZ AND GALLOP
"old" came from another initial cell batch, and the cells at this passage level already showed evidence of "sene~ence," with increased doubling time, accumulation of cellular debris, and ultimately, total degeneration of the culture. The heights of peak A and peak B clearly vary ae-
32oo
cording to the different levels of hydroxylation. In the elution patterns of the digested collagen ~ssociated with the cells (Fig. 3, a and b) the peak labeled "P" is much larger than the similar one present in the digested collagen of the medium (Fig. 1, a and b). This peak may result from di-
(a}
28 O0
2400
=E n"
d
A
2000
1600 I-t200
800
B
400
~L/~L~ 20
40
60
80
I00
120
140
160
180
Z00
22.0
?.40
TUBE NUMBER 3200
(b)
2800
2400
=E n"
B
200C
vd
A
1600 t--
E: I.-
1200
8OO
p
4OO
20
40
60
80
I00
120
140
160
180
200
220
240
TUBE NUMBER
FIG. 1. a, elution profile of the peptides after bacterial collagenase digestion of the collagen in the medium. Pooled medium from 1 week, passage 26, cells confluent in the presence of ascorbate, b, in the absence of ascorbate.
COLLAGEN AND AGING FIBROBLASTS
gestion of procollagen in the cells and is currently being investigated. The data shown in Tables 1 and 2 are related to the three groups of cells mentioned above. With "aging" by increasing passage levels (Table 1), a decrease in the amount
4,BOC
307
of DN A and total protein is observed, corresponding to a decrease in the total number of cells. The decrease of proliferative capacity in aging cultures is probably associated with an increased number of cells incapable of entering DNA
A
(a)
420C
5600 =E
0.: 3 0 0 0
24OO
B
IK
I-'-
1800
1200
600 |
20
40
60
80
I00
120
140
160
180
200
!
|
220
240
2~o
2~o
TUBE NUMBER
3600
5200
(b)
2800
2400 ....: ~E
d
200C
160C I'I--
1200
800
P
400
2~
20
~o
;o
i80 200 ,~o ,;o ,~o ~ ~-~/'t.~,,
TUBE NUMBER
Fro. 2. a, elution profile of the peptides after bacterial collagenase digestion of the collagen in the medium. Pooled medium from 1 week, passage 41, cells confluent in the presence of ascorbate, b, in the absence of ascorbate.
308
PAZ AND GALLOP
synthesis, as was shown by Cristofalo and Sharf (21) and confirmed in our laboratory. Much less protein and DNA were found in the passage 41 old cultures, but the level of collagen associated with the cells expressed as milligrams of collagen
per milligram of DNA did not differ significantly from that associated with the younger cultures. Twice as much collagen was associated with the young and old cultures when ascorbic acid was present at confluency. This collagen was found
450
400
(a)
55C
500 lE 132
250
d P 2:) i'-
200 1513 tOC
!
2O
20
801 1
'~' 1
I~O
I'~O
160
I;0
2~0
2'~3
2~1O -
TUBE NUMBER
450
4.00
(b)
A
S~O
50C
P n,
25c
l~E j
200
nr" I--
t50 iO0
50
2~
2o
~o
~
~r
,4~
~o
,~o
~o 2~o
21o ~
TUBE NUMBER
FIG. 3. a, elution profile of the peptides after bacterial collagenase digestion of the collagen associated with the ceils and cell layer from 1 week, passage 41 (old), cells confluent in the presence of ascorbate, b, in the absence of ascorbate.
COLLAGEN AND AGING FIBROBLASTS
309
TABLE 1 WI-38 CELLS. COLLAGEN ASSOCIATEDWITH THE CELLS AND CELL LAYER" Passage Levels 26
Ascorbie acid . . . . . . . . . . . . . . . . . Total protein, mg . . . . . . . . . . . . . Total DNA, mg . . . . . . . . . . . . . . Collagen soluble in TCA, m g . . . Hydroxylation index, H . . . . . Hydroxylation percentage... Collagen insoluble in T e A dissolved by collagenase, mg. Hydroxylation index, H . . . . . Hydroxylation percentage... Total collagen, mg . . . . . . . . . . . mg Collagen/mg DNA . . . . . . . . Collagen (eells)/total collagen produced, % . . . . . . . . . . . . . .
26
41
41 (old)
41 (old)
2.2 0.1 1.32 71
+ 3O 1.0 0.1 1.77 81
56 1.9 0~1 1.49 75
41
+ 114 2.6 0.2 2.92 96
123 2.8 0.2 0.79 53
+ 77 2.0 0.1 2.4 90
1.8 2.52 92 2.0 0.8
0.7 0.56 41 0.9 0.3
1.8 2.99 96 1.9 0.9
1.0 1.17 67 1.1 0.5
0.6 1.77 81 0.7 0.7
0.7 1.49 75 0.8 0.4
25
13
18
29
23
85
" Cultures kept 1 week at confluency. 20T75 flasks were used in each group. TABLE 2 WI-38 CELLS. COLLAGEN CONTENT OF MEDIUM FROM ONE WEEK AT CONFLUENCYa Passage Levels
Ascorbie acid . . . . . . . . . . . . . . . . . Collagen soluble in TCA, rag.. Hydroxylation index, H .... Hydroxylation percentage.. Collagen insoluble in TCA dissolved by collagenase, mg. Hydroxylation index, H . . . . . Hydroxylation percentage. Total Collagen . . . . . . . . . . . . . . . mg Collagen/mg DNA . . . . . . . . Collagen in medium/total collagen produced, % . . . . . . . Total collagen/medium and cells, mg . . . . . . . . . . . . . . . . . mg Collagen/mg DNA . . . . . . . .
26
26
+ 2.8 3.5 100 4.0 3.5 100 6.8 2.6 77 8.8 3.4
41
41
41 (old)
4t(o]d)
13.8 0.6 44
+ 3.4 3.36 99
6.2 0.82 54
+ 2.3 1.44 74
0.7 1.36 72
3.0 0.79 53 16.8 6.0
2.2 2.9 96 5.6 2.8
1.4 1.04 63 7.6 3.4
0.9 1.62 78 3.2 3.2
1.3 1.54 76 2.0 1.0
87
82
71
95 17.7 6.3
7.5 3.8
8.7 4.0
3.9 3.9
2.8 1.5
, Cultures kept 1 week at conflueney. 20T75 flasks were used in each group. to be very insoluble and only small amounts could be extracted with hot TCA. The remainder was finally solubilized by bacterial eollagenase. In the younger cultures, ascorbic acid has an over-all inhibitory effect on total collagen protein synthesis. Twice as much total collagen is synthesized by passage 26 cultures in the absence of ascorbate (Table 2). The collagen produced by young confluent cultures in the presence of ascorbate is fully hydroxylated, in dramatic contrast to that found in the absence of the
vitamin, when there is about 50% hydroxylation. This underhydroxylated collagen is soluble and is probably deficient in cross-linkage. Only 5% of the total collagen produced by passage 26 cultures in the absence of ascorbate is associated with the cells, the rest (95%) is present in soluble form in the medium. Ascorbic acid is clearly required by cells in young cultures to achieve complete collagen hydroxylation. These results were reproduced when another batch of young cultures were equally examined at passage 26.
310
PAZ AND GALLOP
The collagen synthesis and collagen hydroxylation system of the older cultures are still ascorbate dependent, but to a lesser extent. Without ascorbate, somewhat more collagen is present in the medium, but the difference is not as dramatic as with younger cultures (Table 2). However, the total amount of collagen synthesized is not significantly dependent on the presence of ascorbate. Furthermore, collagen is hydroxylated without ascorbate to a larger extent than seen in a comparable situation with the younger eultttres (see Table 2). The ratio of total collagen synthesized per milligram of DNA is about the same with the senescent cultures of passage 41 old, as with the younger cultures when ascorbate is present. However, without ascorbate, senescent cultures appear to synthesize less collagen per milligram of DNA. In regard to the degree of hydroxylation, the senescent cultures (passage 41 old) showed little, if any, ascorbate dependence. In this situation, the collagen produced by the cells was equally hydroxylated to about 80% of the normal value, whether or not the vitamin was present. The results with older cultures seem to indicate that, with age, the dependence on ascorbic acid for hydroxylation decreases gradually towards almost complete independence in senescent cultures. Experiments are in progress in our laboratory with other cultures that show similar decrease in ascorbic acid dependence with age. Senescent cultures will be studied in more detail to reexamine the effect of ascorbate reported here. DISCUSSION There is considerable evidence for age-related variation in the hydroxylation of both ]ysine and pro]ine in collagen. Bone collagen o~ young chicks appears to show a continuous decrease in the over-all extent of hydroxylation up to 7 weeks of age (22). In the skin collagen of chicks, higher levels of hydroxylation were noted in the embryo as compared to the 3-week-old (23). Recently, in an intensive study of lysine hydroxylation from the 14-day embryo to the 18-month adult chicken, a marked decrease in the over-all extent of lysine hydroxylation with increasing age in both a l and c~2 chains was reported (24); the fall occurred mainly in a short period immediately after hatching. The data suggested that there was strict control of lysine hydroxylation
with perhaps separate enzymatic control of hydroxylation at sites important to cross-linking. In the studies reported here, WI-38 cells in culture show evidence for age-related variation in the hydroxylation of proline in the collagen they produce. Moreover, there is a marked effect of ascorbic acid on both the synthesis and hydroxylation of the collagen produced by "younger" cultures. WI-38 human diploid fibroblasts synthesize and extrude collagen into the medium both in the presence and in the absence of ascorbic acid in the stationary phase of growth. Young cultures (passage 26), in the absence of ascorbic acid, synthesize relatively large amounts of an underhydroxylated soluble form of collagen, which is found mainly in the medium. Somewhat similar results were reported with 3T6 fibroblasts (3), which synthesize an underhydroxylated form of collagen, both in the presence of aseorbie acid at logarithmic phase, and in the absence of ascorbie acid at the stationary phase of growth. The same laboratory (25) reported that ascorbate increased hydroxylation of proline accompanied by a marked increase in proline hydroxylase activity in late log-phase cultures; in the presence of aseorbate the enzyme activity was 3 to 10 times that found in its absence. Aseorbate may play an additional role in collagen modification by stimulating the conversion of an inactive precursor of prolyl hydroxytase to the active enzyme (2). The results in the present study indicate that, with young cultures, aseorbate may have an inhibitory effect on the total amount of collagen produced, in that about 50% less collagen is present in the medium of young confluent cultures in the presence of the vitamin. These observations are in contrast with those described by Peterkofsky (26) using BALB 3T3 and chick embryo fibroblast cultures studied in the logarithmic phase of growth. Total collagen synthesis by BALB 3T3 cultures was unaffected by the absence of aseorbate, but secretion into the medium was greatly diminished both in BALB 3T3 and chick embryo fibroblasts. Furthermore, the hydroxyproline content of the collagen was only 10 to 20% of the normal value. Ramaley and Rosenbloom (27), Margolis and Lukens (28), and Dehm and Prockop (29), using as hydroxylase inhibitor the iron chelator a , a ' dipyridyl, demonstrated the accumulation of underhydroxylated collagen inside the cells and
COLLAGEN AND AGING FIBROBLASTS the presence of low molecular weight collagenasedegradable material in the medium. On the other hand, in ascorbate-deficient cultures of 3T6 fibroblasts in late logarithmic cultures, Bates et al. (30) found neither significant reduction in the amount of collagen secreted during 24 hr, nor increased degradation of collagen to smaller fragments. Somewhat similar results were obtained with L-929 cells by Peterkofsky (1). The effects of ascorbic acid deficiency would be expected to differ from those obtained with a,a'dipyridyl, because the latter agent produces much more severe inhibition of hydroxylation than ascorbate deprivation, and may also be somewhat toxic to the cells. Our studies indicate that proline hydroxylation still occurs to the extent of 50% of the maximum in young ascorbic acid-deficient WI-38 cultures in the stationary phase. The collagen synthesized under those conditions may still contain the appropriate residues of hydroxyproline and/or hydroxylysine that might be required for normal extrusion. Evidently, there is a difference in the behavior of various cell types in culture, and it may not be valid to compare results obtained with various animal cell lines to those obtained with human diploid cell strains which maintain their diploid karyotype throughout their limited life-span. Moreover, many of the earlier studies employed cells which were in the logarithmic phase for short periods of time, so that difficulties of comparison with the system employed here are obvious. As found in this study, the presence in the medium of almost all (95%) of the underhydroxylated collagen produced by ascorbate-deficient young cultures suggests a failure of such cultures to produce insoluble collagen in the cell layer, as in the ascorbate-supplemented situation. This point is under further investigation. Older cultures showed less dependence on ascorbic acid. In the absence of ascorbate, more collagen was extruded into the medium, but the difference was not as dramatic as with the younger cultures. Moreover, although the older cultures in the presence of ascorbate produced fully hydroxylated collagen, without the vitamin, collagen could still be produced which is 60 to 70% hydroxylated. With the senescent cultures (passage 41 old) little ascorbate dependency was seen for proiine hydroxytation. This collagen of the senescent cultures never reached full hydroxyla-
311
tion even when the medium was supplemented with ascorbate. The value of about 80% of normal hydroxylation was obtained with and without ascorbate. Accordingly, the prolyl hydroxylase system and its various controls are clearly culture age dependent. The loss of dependency on ascorbic acid with age may result from defined or programmed changes in the enzyme(s). On the other hand, it may result from a gradual increase in age-dependent random imperfections leading to a hydroxylase system which ultimately is unable to bring about complete collagen prolyl residue hydroxylation even in the presence of ascorbic acid. Comparison of the enzymology and protein chemistry of the prolyl hydroxylation system as isolated from different aged cell cultures is warranted. REFERENCES 1. Peterkofsky, B. 1972. The effect of ascorbic acid on collagen polypeptides synthesis and proline hydroxylation during the growth of cultured fibroblasts. Arch. Biochem. Biophys. 152: 318-328. 2. Stassen, F. L. H., G. J. Cardinale, and S. Udenfriend. 1973. Activation of prolyl hydroxylase in L-929 fibroblasts by ascorbic acid. Proc. Natl. Acad. Sci. U.S.A. 70: 10901093. 3. Bates, C. J., C. J. Prynne, and C. I. Levene. 1972. The synthesis of under-hydroxylated collagen by 3T6 mouse fibroblasts in culture. Biochim. Biophys. Acta 263: 397-405. 4. Green, H., and B. Goldberg. 1963. Kinetics of collagen synthesis by established mammalian cell lines. Nature 200: 1097-1098. 5. Langness, U., and S. Udenfriend. 1974. Collagen biosynthesis in nonfibroblastic cell lines. Proc. Natl. Acad. Sci. U.S.A. 71: 50-56. 6. Hayflick, L., and P. S. Moorhead. 1961. The serial cultivation of human diploid cell strains. Exp. Cell Res. 25: 585-621. 7. Holliday, R., and G. M. Tarrant. 1972. Altered enzymes in aging human fibroblasts. Nature 238: 26-30. 8. Lewis, C. M., and G. M. Tarrant. 1972. Error theory and aging in human diploid fibroblasts. Nature 239: 316-318. 9. Orgel, L. E. 1963. The maintenance of the accuracy of protein synthesis and its relevance to aging. Proc. Natl. Acad. Sci. U.S.A. 49: 517-521. 10. Hayflick, L. 1970. Aging under glass. Exp. Gerontol. 5: 291-303. 11. Martin, G. M., C. A. Sprague, and C. J. Epstein. 1970. Replicative life-span of cultured human cells: effects of donor's age, tissue and genotype. Lab. Invest. 23: 86-92.
312
PAZ AND GALLOP
12. Goldstein, S., J. W. Littlefield, and J. S. Sveldner. 1969. Diabetes mellitus and aging: diminished plating efficiency of cultured human fibroblasts. Proc. Natl. Acad. Sci. U. S. A. 64: 155-160. 13. Cardinale, G. J., R. E. Rhoads, and S. Udenfriend. 1971. Simultaneous incorporation of 1sO into succinate and hydroxyproline catalyzed by collagen proline hydroxylase. Biochem. Biophys. Res. Commun. 43: 537543. 14. Kivirikko, K. I., K. Shudo, S. Sakakibara, and D. J. Prockop. 1972. Studies on protocollagen lysine hydroxylase. Hydroxylation of synthetic peptides and the stoichiometric decarboxylation of a-ketoglutarate. Biochemistry 11: 122-129. 15. Lowry, O. H., N. J. Rosebrough, A. L. Farr, and R. J. Randall. 1951. Protein measurement with the Folin phenol reagent. J. Biol. Chem. 193 : 265-275. 16. Zamenhof, S., and E. Chargaff. 1957. Preparation and assay of deoxyribonucleic acid from animal tissue. In: S. P. Colowick and N. O. Kaplan (Eds.), Methods in Enzymology, Vol. III, Academic Press, New York, p. 702. 17. Giles, K. W., and A. Myers. 1965. An improved diphenylamine method for the estimation of deoxyribonucleic acid. Nature 206: 93. 18. Bornstein, P. 1967. Comparative sequence studies of rat skin and tendon collagen. I. Evidence for incomplete hydroxylation of individual prolyl residues in the normal proteins. Biochemistry 6: 3082-3093. 19. Soberano, M. E., and G. Schoellman. 1972. Specificity of bacterial collagenase: studies with peptides newly synthesized using the solid-phase method. Biochim. Biophys. Acta 271 : 133-144. 20. Hulmes, D. J. S., A. Miller, D. A. D. Parry, K. Piez, and J. Woodhead-Galloway. 1973. Analysis of the primary structure of collagen for the origins cf molecular packing. J. Mol. Biol. 79: 137-143.
21. Cristofalo, V. J., and B. B. Sharf. 1973. Cellular senescence and DNA synthesis. Exp. Cell Res. 76: 419--427. 22. Miller, E. J., G. R. Martin, K. A. Piez, and M. J. Powers. 1967. Characterization of chick bone collagen and compositional changes associated with maturation. J. Biol. Chem. 242: 5481-5489. 23. Barnes, M. J., B. J. Constable, L. F. Morton, and E. Kodicek. 1971. Hydroxylysine in the N-terminal telopeptides of skin collagen from chick embryo and newborn rat. Biochem. J. 125: 925-928. 24. Barnes, M. J., B. J. Constable, L. F. Morton, and P. M. Royce. 1974. Age-related variations in hydroxylation of lysine and proline in collagen. Biochem. J. 139: 461-468. 25. Levene, C. I., J. J. Aleo, C. J. Prynne, and C. J. Bates. 1974. The activation of protocollagen proline hydroxylase by acetic acid in cultured 3T6 fibroblasts. Bioehim. Biophys. Acta 338: 29-36. 26. Peterkofsky, B. 1972. Regulation of collagen secreted by ascorbic acid in 3T3 and chick embryo fibroblasts. Biochem. Biophys. Res. Commun. 49: 1343-1350. 27. Ramaley, P. B., and J. Rosenbloom. 1971. Inhibition of proline and lysine hydroxylation prevents normal extrusion of collagen by 3T6 fibroblasts in culture. FEBS Lett. 15 : 59-64. 28. Margolis, R. L., and L. N. Lukens. 1971. Hydroxylation in the secretion of collagen by mouse fibroblasts in culture. Arch. Biochem. Biophys. 147: 612-618. 29. Dehm, P., and D. J. Prockop, 1971. Synthesis and extrusion of collagen by freshly isolated cells from chick embryo tendon. Biochim. Biophys. Acta 240: 358-369. 30. Bates, C. J., A. J. Bailey, C. J. Prynne, and C. I. Levene. 1972. The effect of ascorbie acid on the synthesis of collagen precursor secreted by 3T6 mouse fibroblasts in culture. Biochim. Biophys. Acta 278: 372-390.
We wish to thank Dr. L. Hayflick and Ms. Eva A. Pfendt for supplying the WI-38 cells and for helpful suggestions. We appreciate the aid of Dr. Katherine Chert and the excellent technical assistance of Mr. Douglas Tapper. The work was supported by National Institutes of Health grants AM15671 and HD07376.
COLLAGEN SYNTHESIZED AND MODIFIED BY AGING FIBROBLASTS IN CULTURE MERCEDES A. PAZ1 AND PAUL M. GALLOP The Department of Biological Chemistry and Orthopaedic Surgery, Harvard Medical School and Harvard School of Dental Medicine, and The Children's Hospital Medical Center, 800 Longwood Avenue, Boston, Massachusetts 02115
SUMMARY Collagen is produced by WI-38 diploid human fibroblast cultures throughout their life cycle. It is examined by a sensitive method based on the analysis of specific peptides obtained after digestion with bacterial collagenase. The production and hydroxylation of the collagen is strongly dependent upon the age (population doublings) of the culture and the presence of ascorbic acid. Young cultures (passage 26) produce large amounts of collagen in the absence of ascorbic acid, and this collagen is about 50% hydroxylated compared to that produced by young cultures in the presence of ascorbic acid. Ascorbic acid reduces to about one-half the amount of collagen produced by these young cultures. The young confluent cultures also depend strongly on ascorbic acid for hydroxylation of proline. The dependence declines rapidly with the age of the culture. The collagen produced by young cultures supplied with ascorbic acid is very similar to the type I collagen produced by normal individuals and has about the same degree of hydroxylation of its prolyl residues. The amount of collagen produced by "older" cultures is unaffected by ascorbic acid, but the degree of hydroxylation is normal only if ascorbic acid is present, and is decreased to about 60 to 70% in the absence of the vitamin. "Senescent" cultures showed little, if any, dependency on ascorbic acid, and the collagen produced, with and without the vitamin, is about 80% hydroxylated. The prolyl hydroxylation system of the WI-38 cells and the various controls on the SySr tern are age-dependent. Key words: collagen; fibroblast; hydroxylation; culture; ascorbic acid; age-dependence. Fibroblasts and other cell types such as ehondroblasts, smooth muscle cells, and epithelial cells can synthesize collagen. There is evidence that transformation of certain fibroblastic cells results in established cell lines that continue to produce and hydroxylate collagen (1-5). For this study, we have chosen a human diploid fibroblast strain WI-38 derived from embryonic lung tissue. This system displays the reproducible limited cell doubling phenomenon described by Hayiiick and Moorhead (6). It has been suggested that WI-38 fibroblasts may serve as an experimental model of aging (7, 8). This cell strain and related strains have been employed
for a wide variety of investigations which attempt to elucidate the mechanism of "programmed" death in culture. Several investigators have been considering the possibility that this system may represent an experimental model to test the Orgel hypothesis (9), in which aging is considered to be a consequence of accumulated errors in the protein synthetic apparatus. These errors are presumed to occur in the transcriptional and/or translational processes; some errors could accumulate and thereby promote additional increases in error frequency with each cell doubling. Such errors and their consequences may be expressed in a more or less random manner, or in some specific sensitive loci in the genetic material of the cells. In the latter case, 1 To whom reprint requests should be sent at the Department of Orthopaedic Surgery, Children's they could be considered analogous to somatic Hospital, 300 Longwood Ave., Boston, Mass. 02115. mutations. 302
COLLAGEN AND AGING FIBROBLASTS Modified proteins that are the end products of such events might be expected to show increasing imperfections with cell doublings until a lethal situation is attained at the so-called "error catastrophe stage." This stage is expected at phase I I I in the life of the diploid cell culture. As phase I I I is approached, less perfect, albeit functional, proteins may be found which might be more heat labile, and might also demonstrate a broadened or modified substrate specificity or an altered response to regulatory agents. Heatlabile glucose-6-phosphate dehydrogenase and catalytically inactive lactic dehydrogenase accumulate in an aging diploid cell strain derived from embryonic lung (7, 8), similar to the WL38 strain. This may be a first experimental confirmation of the Orgel hypothesis, providing valuable, but still controversial, evidence supporting the concepts of Hayflick (10) and others (11, 12), which relate limited doubling potential to cellular aging. We have selected as our test system aging WI-38 fibroblasts. After increasing numbers of doublings, we examined the continued potential of these cells to produce enzymes that can correctly modify precursor collagen at various post-translational steps. Collagen is a large structural protein produced by fibroblasts and other cells. Precursor collagen is subject to several specific post-translational modifications that occur both within the cell and after extrusion of the protein from the cell. We have focused on the specific collagen-hydroxylating systems, namely the two mixed function oxidase systems, prolyl hydroxylase and lysyl hydroxylase. Both of these hydroxylase systems require ferrous iron as co-oxidant and molecular oxygen as oxidant, and a-ketoglutarate as an obligatory co-substrate and co-reductant (13, 14). Aseorbate appears here, as in many other mixed function oxidase systems, to be an effective mediator, but its mode of action is not completely understood. In addition to its possible mediator role in the hydroxylation reaction, ascorbate may be an activator of an inactive form or precursor of prolyl hydroxylase (2). In this investigation, collagen synthesis, collagen prolyl residue hydroxylation, and the role of ascorbic acid are examined in "young," "older," and "senescent" cultures. A new and sensitive procedure is introduced, which employs bacterial collagenase for digestion of collagen, and mapping of the peptides with examination
303
of the ratio of two important and specific collagen tripeptide fractions. This can accurately define the level of hydroxylation, and allow one to measure relatively small changes in the levels of collagen hydroxylation. We have examined as a function of age the effect of ascorbate on collagen in the culture medium and collagen which remains within and around confluent cells with regard to synthesis and prolyl hydroxylation. Later experiments will be concerned with hydroxylysine formation and distribution, collagen carbohydration, and cross-linkage maturation in the collagen produced by the aging WI-38 cultures. MATERIALS AND METHODS
The WI-38 diploid fibroblast cell strain em: ployed was initially derived from female embryonic human lung. Starter cultures were obtained from Professor L. Hayflick of Stanford University. These cultures were usually received at the 16th or 17th passage level and were grown in 75-cm2 plastic tissue culture flasks (T-75). Cells were transferred at weekly intervals at a 1:4 split ratio. General procedures were similar to those used by Hayflick and Moorhead (6). The growth medium consisted of BME (basal medium Eagle, Earle's salts; GIBCO), supplemented with 10% fetal calf serum (GIBCO) and 50 ttg per ml of aureomycin (Lederle, product 4691=96 intravenous). Cultures were grown at 37~ in an atmosphere of 5% C02 and 95% air, and were monitored routinely for mycoplasmal contamination by Dr. L. Hayflick. Cells were trypsinized and counted, using a hemocyt0meter, after the cultures had reached conftuency. Total protein and DNA contents were determine d9 just after the cultures had reached confluency and after 1 week at confluency. The procedures of Lowry et al. (15) and a modification of the biuret reaction (16) were used for protein determinations. The diphenylamine reaction (17) was used to determine DNA content. Collagen concentration was determined by amino acid analysis of the collagenase-digested material after hydrolysis with 6 N HC1 at 110~ for 22 hr. The amount of collagen present was calculated from the hydroxyproline value, assuming for a normal hydroxylated collagen an average of 100 residues of hydroxyproline per 1,000 residues, and an average residue molecular weight of 92. The calculated amount
304
PAZ AND GALLOP
of collagen was corrected for the degree of underhydroxylation. I)ROCEDURE
Forty T-75 flasks of WI-38 cells at the desired passage level were used. They were divided into two groups of 20 flasks each as soon as the cells reached confluency. One group received complete BME medium supplemented with 50 t~g per ml of L-ascorbic acid, the other received complete BME medium without the vitamin. Both groups of cells were kept for 1 week at coniiuency. The medium was changed and collected every other day. Twenty-four hours after the third change of medium, the cells were refed with 20 ml of fresh medium containing 2,3-L-[3H]proline (New England Nuclear, 39.7 Ci per mmole, 0.5 ~Ci per ml). After a labeling period of 24 hr, the medium was separated from the cells and pooled with unlabeled medium from the respective group of cultures. After extensive dialysis against distilled water, the medium was lyophilized and saved for further processing. The cells were washed three times with cold saline and were scraped off the surface with a rubber policeman. Following centrifugation, the cell pellet was used for DNA, protein, and collagen determinations. In the present study, the degree of hydroxylation of the collagen was measured during a 24-hr labeling period and was assumed to be representative of the entire week at confluency. The amount of collagen determined in the medium and cell layer represents collagen actually accumulated during 1 week at confluency (balance of synthesis and degradation). Collagen was extracted from the cells and medium with hot trichloroacetic acid (TCA) followed by collagenase. The TCA completely disrupts the cells and denatures the collagen, allowing a total and efficient digestion by collagenase. Following this procedure, a second digestion with collagenase was unnecessary.
Preparation of the fraction soluble in hot trichloroacetic acid. The cell pellet and the lyophilized medium were heated in a final concentration of 5% TCA for 15 min at 70~ The preparations were then centrifuged at 7000 • g for 15 to 20 min, and the supernatant fluids were separated from the residues. Supernatant fluids. To digest the collagen extracted by the hot TCA solution with collagenase (Worthington Biochemical Corporation, CLSPA-
2000), it was necessary to eliminate most of the TCA present in the soluble fraction. This was achieved by three extractions with approximately three volumes of ether, using a separatory funnel. The soluble fraction was then lyophilized and the residue dissolved in 5 to 10 ml of 0.5 m~ CaCl~ and the pH adjusted to 7.5 with 1 N NaOH. Collagenase was added at a weight ratio of substrate :enzyme of approximately 50:1. The mixture was incubated in a shaking bath at 37~ overnight. Collagenase was separated from the collagen peptides by precipitation at 4~ with 5% TCA followed by centrifugation at 7000 • g for 20 rain. The supernatant fluid was again extracted three times with ether to remove TCA and chromatographed for characteristic peptides in a single column amino acid analyzer using a gradient of sodium citrate buffers. Residues. The fractions insoluble in hot TCA were washed three times by centrifugation with water and suspended in 0.5 mM CaC12 and the pH adjusted to 7.5 with 1 N NaOH. Following addition of collagenase, the suspensions were incubated at 37~ overnight and centrifuged at 7000 • g for 20 min. The supernatant fluid contained the collagen, solubilized by the collagenase, as peptides, and the residue contained proteins other than collagen. The separation of the collagenase from the supernatant fluids was achieved by precipitation with cold 5% TCA followed by centrifugation. The TCA was then extracted with ether following the procedure described above.
Separation of the collagenase-released peptides by column chromatography. The main tripeptides, gly-pro-hyp, gly-pro-ala, and gly-pro-pro, obtained by digestion of normally hydroxylated and underhydroxylated collagen with collagenase are commercially available (CycloChemical and Schwarz/Mann), and their positions of elution from the column of the amino acid analyzer can be established. Using a gradient of sodium citrate buffers similar to that used for the separation of amino acids, gly-pro-hyp elutes between alanine and valine, and gly-pro-ala elutes together with gly-pro-pro between methionine and isoleucine. Inasmuch as the collagen present in the TCAsoluble fractions, and in the fractions solubilized by treatment with collagenase, is labeled, the elution profile from the column can be followed by counting the radioactivity in the fractions in a scintillation counter. Two major peaks are ob-
COLLAGEN AND AGING FIBROBLASTS rained in the elution system employed. The first major peak is gly-pro-hyp, and also includes small amounts of a larger peptide with one prolinc but no hydroxyproline (peak A). The second major peak, gly pro-ala, chromatographs with giy-propro (peak B). The ratio of total dpm in peak A to total dpm in peak B, can be determined from the elution profile. This ratio is defined as the hydroxylation index, (H), and has an experimental value of 3.52 for skin or fibroblast collagen (type I) when it is normally hydroxylated. Such a value is also predicted from the known sequences of the a l chain of type [ collagen and the known specificity of collagenase. The fact that the predicted ratio of the type I collagen from a l and the found ratio of (al)2a2 coincided, indicates that the a2 chains could also be split by collagenase into a similar distribution of these characteristic peptides. Normally, a collagenase digest of an a l collagen chain would contain approximately 22 equivalents of gly-pro-hyp, 4 equivalents of a larger peptide containing 1 proline and no hydroxyproline, and 14 residues of gly-pro-ala. In the absence of full hydroxylation of the prolyl residues there would be a This ratio, H, measures the average hydroxylation of the second prolyl residue in those glypro-pro triplets where collagenase can split the bonds preceding the glycine and following the second proline. This represents 22 of ,9,8 total glypro-pro + gly-pro-hyp triplets, and over-all, this accounts for about 20% of the total hydroxyproline in the collagen a l chain. Accordingly, H may not accurately reflect the average degree of hydroxylation of all the proline residues that can be hydroxylated. However, other peptide fractions, for example, gly-leu-hyp, which chromatographs immediately before peak B in our system, and gly-ala-hyp which chromatographs before gly-prohyp in fractions 20 to 40, and can be clearly separated from other peptides with a slightly modified gradient, show similar behavior to gly-pro-hyp in that their value decreases with underhydroxylation. The composite a l chain of rat skin and calf skin collagen, containing 1052 residues, has 112 prolyl positions which can be found partially or fully hydroxylated, corresponding to an analysis of 106 hydroxyproline residues for 1,000 residues. The experimental value of 96 hydroxyproline residues per 1,000 residues for an a l chain indicates that about 90% of all the susceptible prolines are normally hydroxylated. The collagen produced by young fibroblasts cultured in ascorbic acid-supplemented medium, shows an experimental value of 3.5 for II, which indicates 100% hydroxylation of the gly-pro-hyp tripeptides released by collagenuse.
305
decrease in the amount of gly-pro-hyp, with an equivalent increase ill the amount of gly-pro-pro. Consequently, a change occurs in the ratio of peak A to peak B that is reflected by H the hydroxylation index. It is possible, for example, to determine that all experimental H of 0.7, as found ill the absence of ascorbic acid with cells from young cultures (see below), corresponds to a collagen with 49% of the normal prolyl residue hydroxylation. The following formula (a) can be derived for the percentage hydroxylation: 58(it)
-
4
% Hydroxylation = 100 44(H) + 44
(a)
where H is tile hydroxylation index determined experimentally from the ratio of dpm in peak A to that in peak B. This formula is derived as follows: there are 38 gly-pro-hyp triplets in an a l chain, of which 22 can be released as tripeptides by coIlagenase [the complete sequence of the a l chain is given in Hulmes et al. (20)]. The fractional hydroxylation of this peptide is X. In peak A there are also four residues of a peptide containing one equivalent of proline and no hydroxyproline. Then peak A has a total radioactivity proportional to (2 • 22 • X) + (1 • 4), where a factor of 2 is used because both proline and hydroxyproline are radioactive. In peak B there are both gly-pro-ala and gly-pro-pro. There are 31 gly-pro-ala triplets in an a l chain, of which 14 are released as tripeptides by collagenase. Thus, the radioactivity in peak B is proportional to (14 • 1) + (1 - X)(2 • 22) = 14 + 44 44X. Thus: II
peak A peak B
44X + 4 58 -- 44X
from which the percentage hydroxylation, 100X call be expressed in the formula (a) above. RESULTS
Figs. 1 to 3 show several of the elution profiles of labeled collagen after bacterial collagenase digestion obtained from the medium and from the cultures carried to different pas~ge levels. Cultures in pa~ages 26 and 41 originated from the same batch of young cultures, and were carried until they reached phase I I I at about passage 51. The passage classified as passage 41
306
PAZ AND GALLOP
"old" came from another initial cell batch, and the cells at this passage level already showed evidence of "sene~ence," with increased doubling time, accumulation of cellular debris, and ultimately, total degeneration of the culture. The heights of peak A and peak B clearly vary ae-
32oo
cording to the different levels of hydroxylation. In the elution patterns of the digested collagen ~ssociated with the cells (Fig. 3, a and b) the peak labeled "P" is much larger than the similar one present in the digested collagen of the medium (Fig. 1, a and b). This peak may result from di-
(a}
28 O0
2400
=E n"
d
A
2000
1600 I-t200
800
B
400
~L/~L~ 20
40
60
80
I00
120
140
160
180
Z00
22.0
?.40
TUBE NUMBER 3200
(b)
2800
2400
=E n"
B
200C
vd
A
1600 t--
E: I.-
1200
8OO
p
4OO
20
40
60
80
I00
120
140
160
180
200
220
240
TUBE NUMBER
FIG. 1. a, elution profile of the peptides after bacterial collagenase digestion of the collagen in the medium. Pooled medium from 1 week, passage 26, cells confluent in the presence of ascorbate, b, in the absence of ascorbate.
COLLAGEN AND AGING FIBROBLASTS
gestion of procollagen in the cells and is currently being investigated. The data shown in Tables 1 and 2 are related to the three groups of cells mentioned above. With "aging" by increasing passage levels (Table 1), a decrease in the amount
4,BOC
307
of DN A and total protein is observed, corresponding to a decrease in the total number of cells. The decrease of proliferative capacity in aging cultures is probably associated with an increased number of cells incapable of entering DNA
A
(a)
420C
5600 =E
0.: 3 0 0 0
24OO
B
IK
I-'-
1800
1200
600 |
20
40
60
80
I00
120
140
160
180
200
!
|
220
240
2~o
2~o
TUBE NUMBER
3600
5200
(b)
2800
2400 ....: ~E
d
200C
160C I'I--
1200
800
P
400
2~
20
~o
;o
i80 200 ,~o ,;o ,~o ~ ~-~/'t.~,,
TUBE NUMBER
Fro. 2. a, elution profile of the peptides after bacterial collagenase digestion of the collagen in the medium. Pooled medium from 1 week, passage 41, cells confluent in the presence of ascorbate, b, in the absence of ascorbate.
308
PAZ AND GALLOP
synthesis, as was shown by Cristofalo and Sharf (21) and confirmed in our laboratory. Much less protein and DNA were found in the passage 41 old cultures, but the level of collagen associated with the cells expressed as milligrams of collagen
per milligram of DNA did not differ significantly from that associated with the younger cultures. Twice as much collagen was associated with the young and old cultures when ascorbic acid was present at confluency. This collagen was found
450
400
(a)
55C
500 lE 132
250
d P 2:) i'-
200 1513 tOC
!
2O
20
801 1
'~' 1
I~O
I'~O
160
I;0
2~0
2'~3
2~1O -
TUBE NUMBER
450
4.00
(b)
A
S~O
50C
P n,
25c
l~E j
200
nr" I--
t50 iO0
50
2~
2o
~o
~
~r
,4~
~o
,~o
~o 2~o
21o ~
TUBE NUMBER
FIG. 3. a, elution profile of the peptides after bacterial collagenase digestion of the collagen associated with the ceils and cell layer from 1 week, passage 41 (old), cells confluent in the presence of ascorbate, b, in the absence of ascorbate.
COLLAGEN AND AGING FIBROBLASTS
309
TABLE 1 WI-38 CELLS. COLLAGEN ASSOCIATEDWITH THE CELLS AND CELL LAYER" Passage Levels 26
Ascorbie acid . . . . . . . . . . . . . . . . . Total protein, mg . . . . . . . . . . . . . Total DNA, mg . . . . . . . . . . . . . . Collagen soluble in TCA, m g . . . Hydroxylation index, H . . . . . Hydroxylation percentage... Collagen insoluble in T e A dissolved by collagenase, mg. Hydroxylation index, H . . . . . Hydroxylation percentage... Total collagen, mg . . . . . . . . . . . mg Collagen/mg DNA . . . . . . . . Collagen (eells)/total collagen produced, % . . . . . . . . . . . . . .
26
41
41 (old)
41 (old)
2.2 0.1 1.32 71
+ 3O 1.0 0.1 1.77 81
56 1.9 0~1 1.49 75
41
+ 114 2.6 0.2 2.92 96
123 2.8 0.2 0.79 53
+ 77 2.0 0.1 2.4 90
1.8 2.52 92 2.0 0.8
0.7 0.56 41 0.9 0.3
1.8 2.99 96 1.9 0.9
1.0 1.17 67 1.1 0.5
0.6 1.77 81 0.7 0.7
0.7 1.49 75 0.8 0.4
25
13
18
29
23
85
" Cultures kept 1 week at confluency. 20T75 flasks were used in each group. TABLE 2 WI-38 CELLS. COLLAGEN CONTENT OF MEDIUM FROM ONE WEEK AT CONFLUENCYa Passage Levels
Ascorbie acid . . . . . . . . . . . . . . . . . Collagen soluble in TCA, rag.. Hydroxylation index, H .... Hydroxylation percentage.. Collagen insoluble in TCA dissolved by collagenase, mg. Hydroxylation index, H . . . . . Hydroxylation percentage. Total Collagen . . . . . . . . . . . . . . . mg Collagen/mg DNA . . . . . . . . Collagen in medium/total collagen produced, % . . . . . . . Total collagen/medium and cells, mg . . . . . . . . . . . . . . . . . mg Collagen/mg DNA . . . . . . . .
26
26
+ 2.8 3.5 100 4.0 3.5 100 6.8 2.6 77 8.8 3.4
41
41
41 (old)
4t(o]d)
13.8 0.6 44
+ 3.4 3.36 99
6.2 0.82 54
+ 2.3 1.44 74
0.7 1.36 72
3.0 0.79 53 16.8 6.0
2.2 2.9 96 5.6 2.8
1.4 1.04 63 7.6 3.4
0.9 1.62 78 3.2 3.2
1.3 1.54 76 2.0 1.0
87
82
71
95 17.7 6.3
7.5 3.8
8.7 4.0
3.9 3.9
2.8 1.5
, Cultures kept 1 week at conflueney. 20T75 flasks were used in each group. to be very insoluble and only small amounts could be extracted with hot TCA. The remainder was finally solubilized by bacterial eollagenase. In the younger cultures, ascorbic acid has an over-all inhibitory effect on total collagen protein synthesis. Twice as much total collagen is synthesized by passage 26 cultures in the absence of ascorbate (Table 2). The collagen produced by young confluent cultures in the presence of ascorbate is fully hydroxylated, in dramatic contrast to that found in the absence of the
vitamin, when there is about 50% hydroxylation. This underhydroxylated collagen is soluble and is probably deficient in cross-linkage. Only 5% of the total collagen produced by passage 26 cultures in the absence of ascorbate is associated with the cells, the rest (95%) is present in soluble form in the medium. Ascorbic acid is clearly required by cells in young cultures to achieve complete collagen hydroxylation. These results were reproduced when another batch of young cultures were equally examined at passage 26.
310
PAZ AND GALLOP
The collagen synthesis and collagen hydroxylation system of the older cultures are still ascorbate dependent, but to a lesser extent. Without ascorbate, somewhat more collagen is present in the medium, but the difference is not as dramatic as with younger cultures (Table 2). However, the total amount of collagen synthesized is not significantly dependent on the presence of ascorbate. Furthermore, collagen is hydroxylated without ascorbate to a larger extent than seen in a comparable situation with the younger eultttres (see Table 2). The ratio of total collagen synthesized per milligram of DNA is about the same with the senescent cultures of passage 41 old, as with the younger cultures when ascorbate is present. However, without ascorbate, senescent cultures appear to synthesize less collagen per milligram of DNA. In regard to the degree of hydroxylation, the senescent cultures (passage 41 old) showed little, if any, ascorbate dependence. In this situation, the collagen produced by the cells was equally hydroxylated to about 80% of the normal value, whether or not the vitamin was present. The results with older cultures seem to indicate that, with age, the dependence on ascorbic acid for hydroxylation decreases gradually towards almost complete independence in senescent cultures. Experiments are in progress in our laboratory with other cultures that show similar decrease in ascorbic acid dependence with age. Senescent cultures will be studied in more detail to reexamine the effect of ascorbate reported here. DISCUSSION There is considerable evidence for age-related variation in the hydroxylation of both ]ysine and pro]ine in collagen. Bone collagen o~ young chicks appears to show a continuous decrease in the over-all extent of hydroxylation up to 7 weeks of age (22). In the skin collagen of chicks, higher levels of hydroxylation were noted in the embryo as compared to the 3-week-old (23). Recently, in an intensive study of lysine hydroxylation from the 14-day embryo to the 18-month adult chicken, a marked decrease in the over-all extent of lysine hydroxylation with increasing age in both a l and c~2 chains was reported (24); the fall occurred mainly in a short period immediately after hatching. The data suggested that there was strict control of lysine hydroxylation
with perhaps separate enzymatic control of hydroxylation at sites important to cross-linking. In the studies reported here, WI-38 cells in culture show evidence for age-related variation in the hydroxylation of proline in the collagen they produce. Moreover, there is a marked effect of ascorbic acid on both the synthesis and hydroxylation of the collagen produced by "younger" cultures. WI-38 human diploid fibroblasts synthesize and extrude collagen into the medium both in the presence and in the absence of ascorbic acid in the stationary phase of growth. Young cultures (passage 26), in the absence of ascorbic acid, synthesize relatively large amounts of an underhydroxylated soluble form of collagen, which is found mainly in the medium. Somewhat similar results were reported with 3T6 fibroblasts (3), which synthesize an underhydroxylated form of collagen, both in the presence of aseorbie acid at logarithmic phase, and in the absence of ascorbie acid at the stationary phase of growth. The same laboratory (25) reported that ascorbate increased hydroxylation of proline accompanied by a marked increase in proline hydroxylase activity in late log-phase cultures; in the presence of aseorbate the enzyme activity was 3 to 10 times that found in its absence. Aseorbate may play an additional role in collagen modification by stimulating the conversion of an inactive precursor of prolyl hydroxytase to the active enzyme (2). The results in the present study indicate that, with young cultures, aseorbate may have an inhibitory effect on the total amount of collagen produced, in that about 50% less collagen is present in the medium of young confluent cultures in the presence of the vitamin. These observations are in contrast with those described by Peterkofsky (26) using BALB 3T3 and chick embryo fibroblast cultures studied in the logarithmic phase of growth. Total collagen synthesis by BALB 3T3 cultures was unaffected by the absence of aseorbate, but secretion into the medium was greatly diminished both in BALB 3T3 and chick embryo fibroblasts. Furthermore, the hydroxyproline content of the collagen was only 10 to 20% of the normal value. Ramaley and Rosenbloom (27), Margolis and Lukens (28), and Dehm and Prockop (29), using as hydroxylase inhibitor the iron chelator a , a ' dipyridyl, demonstrated the accumulation of underhydroxylated collagen inside the cells and
COLLAGEN AND AGING FIBROBLASTS the presence of low molecular weight collagenasedegradable material in the medium. On the other hand, in ascorbate-deficient cultures of 3T6 fibroblasts in late logarithmic cultures, Bates et al. (30) found neither significant reduction in the amount of collagen secreted during 24 hr, nor increased degradation of collagen to smaller fragments. Somewhat similar results were obtained with L-929 cells by Peterkofsky (1). The effects of ascorbic acid deficiency would be expected to differ from those obtained with a,a'dipyridyl, because the latter agent produces much more severe inhibition of hydroxylation than ascorbate deprivation, and may also be somewhat toxic to the cells. Our studies indicate that proline hydroxylation still occurs to the extent of 50% of the maximum in young ascorbic acid-deficient WI-38 cultures in the stationary phase. The collagen synthesized under those conditions may still contain the appropriate residues of hydroxyproline and/or hydroxylysine that might be required for normal extrusion. Evidently, there is a difference in the behavior of various cell types in culture, and it may not be valid to compare results obtained with various animal cell lines to those obtained with human diploid cell strains which maintain their diploid karyotype throughout their limited life-span. Moreover, many of the earlier studies employed cells which were in the logarithmic phase for short periods of time, so that difficulties of comparison with the system employed here are obvious. As found in this study, the presence in the medium of almost all (95%) of the underhydroxylated collagen produced by ascorbate-deficient young cultures suggests a failure of such cultures to produce insoluble collagen in the cell layer, as in the ascorbate-supplemented situation. This point is under further investigation. Older cultures showed less dependence on ascorbic acid. In the absence of ascorbate, more collagen was extruded into the medium, but the difference was not as dramatic as with the younger cultures. Moreover, although the older cultures in the presence of ascorbate produced fully hydroxylated collagen, without the vitamin, collagen could still be produced which is 60 to 70% hydroxylated. With the senescent cultures (passage 41 old) little ascorbate dependency was seen for proiine hydroxytation. This collagen of the senescent cultures never reached full hydroxyla-
311
tion even when the medium was supplemented with ascorbate. The value of about 80% of normal hydroxylation was obtained with and without ascorbate. Accordingly, the prolyl hydroxylase system and its various controls are clearly culture age dependent. The loss of dependency on ascorbic acid with age may result from defined or programmed changes in the enzyme(s). On the other hand, it may result from a gradual increase in age-dependent random imperfections leading to a hydroxylase system which ultimately is unable to bring about complete collagen prolyl residue hydroxylation even in the presence of ascorbic acid. Comparison of the enzymology and protein chemistry of the prolyl hydroxylation system as isolated from different aged cell cultures is warranted. REFERENCES 1. Peterkofsky, B. 1972. The effect of ascorbic acid on collagen polypeptides synthesis and proline hydroxylation during the growth of cultured fibroblasts. Arch. Biochem. Biophys. 152: 318-328. 2. Stassen, F. L. H., G. J. Cardinale, and S. Udenfriend. 1973. Activation of prolyl hydroxylase in L-929 fibroblasts by ascorbic acid. Proc. Natl. Acad. Sci. U.S.A. 70: 10901093. 3. Bates, C. J., C. J. Prynne, and C. I. Levene. 1972. The synthesis of under-hydroxylated collagen by 3T6 mouse fibroblasts in culture. Biochim. Biophys. Acta 263: 397-405. 4. Green, H., and B. Goldberg. 1963. Kinetics of collagen synthesis by established mammalian cell lines. Nature 200: 1097-1098. 5. Langness, U., and S. Udenfriend. 1974. Collagen biosynthesis in nonfibroblastic cell lines. Proc. Natl. Acad. Sci. U.S.A. 71: 50-56. 6. Hayflick, L., and P. S. Moorhead. 1961. The serial cultivation of human diploid cell strains. Exp. Cell Res. 25: 585-621. 7. Holliday, R., and G. M. Tarrant. 1972. Altered enzymes in aging human fibroblasts. Nature 238: 26-30. 8. Lewis, C. M., and G. M. Tarrant. 1972. Error theory and aging in human diploid fibroblasts. Nature 239: 316-318. 9. Orgel, L. E. 1963. The maintenance of the accuracy of protein synthesis and its relevance to aging. Proc. Natl. Acad. Sci. U.S.A. 49: 517-521. 10. Hayflick, L. 1970. Aging under glass. Exp. Gerontol. 5: 291-303. 11. Martin, G. M., C. A. Sprague, and C. J. Epstein. 1970. Replicative life-span of cultured human cells: effects of donor's age, tissue and genotype. Lab. Invest. 23: 86-92.
312
PAZ AND GALLOP
12. Goldstein, S., J. W. Littlefield, and J. S. Sveldner. 1969. Diabetes mellitus and aging: diminished plating efficiency of cultured human fibroblasts. Proc. Natl. Acad. Sci. U. S. A. 64: 155-160. 13. Cardinale, G. J., R. E. Rhoads, and S. Udenfriend. 1971. Simultaneous incorporation of 1sO into succinate and hydroxyproline catalyzed by collagen proline hydroxylase. Biochem. Biophys. Res. Commun. 43: 537543. 14. Kivirikko, K. I., K. Shudo, S. Sakakibara, and D. J. Prockop. 1972. Studies on protocollagen lysine hydroxylase. Hydroxylation of synthetic peptides and the stoichiometric decarboxylation of a-ketoglutarate. Biochemistry 11: 122-129. 15. Lowry, O. H., N. J. Rosebrough, A. L. Farr, and R. J. Randall. 1951. Protein measurement with the Folin phenol reagent. J. Biol. Chem. 193 : 265-275. 16. Zamenhof, S., and E. Chargaff. 1957. Preparation and assay of deoxyribonucleic acid from animal tissue. In: S. P. Colowick and N. O. Kaplan (Eds.), Methods in Enzymology, Vol. III, Academic Press, New York, p. 702. 17. Giles, K. W., and A. Myers. 1965. An improved diphenylamine method for the estimation of deoxyribonucleic acid. Nature 206: 93. 18. Bornstein, P. 1967. Comparative sequence studies of rat skin and tendon collagen. I. Evidence for incomplete hydroxylation of individual prolyl residues in the normal proteins. Biochemistry 6: 3082-3093. 19. Soberano, M. E., and G. Schoellman. 1972. Specificity of bacterial collagenase: studies with peptides newly synthesized using the solid-phase method. Biochim. Biophys. Acta 271 : 133-144. 20. Hulmes, D. J. S., A. Miller, D. A. D. Parry, K. Piez, and J. Woodhead-Galloway. 1973. Analysis of the primary structure of collagen for the origins cf molecular packing. J. Mol. Biol. 79: 137-143.
21. Cristofalo, V. J., and B. B. Sharf. 1973. Cellular senescence and DNA synthesis. Exp. Cell Res. 76: 419--427. 22. Miller, E. J., G. R. Martin, K. A. Piez, and M. J. Powers. 1967. Characterization of chick bone collagen and compositional changes associated with maturation. J. Biol. Chem. 242: 5481-5489. 23. Barnes, M. J., B. J. Constable, L. F. Morton, and E. Kodicek. 1971. Hydroxylysine in the N-terminal telopeptides of skin collagen from chick embryo and newborn rat. Biochem. J. 125: 925-928. 24. Barnes, M. J., B. J. Constable, L. F. Morton, and P. M. Royce. 1974. Age-related variations in hydroxylation of lysine and proline in collagen. Biochem. J. 139: 461-468. 25. Levene, C. I., J. J. Aleo, C. J. Prynne, and C. J. Bates. 1974. The activation of protocollagen proline hydroxylase by acetic acid in cultured 3T6 fibroblasts. Bioehim. Biophys. Acta 338: 29-36. 26. Peterkofsky, B. 1972. Regulation of collagen secreted by ascorbic acid in 3T3 and chick embryo fibroblasts. Biochem. Biophys. Res. Commun. 49: 1343-1350. 27. Ramaley, P. B., and J. Rosenbloom. 1971. Inhibition of proline and lysine hydroxylation prevents normal extrusion of collagen by 3T6 fibroblasts in culture. FEBS Lett. 15 : 59-64. 28. Margolis, R. L., and L. N. Lukens. 1971. Hydroxylation in the secretion of collagen by mouse fibroblasts in culture. Arch. Biochem. Biophys. 147: 612-618. 29. Dehm, P., and D. J. Prockop, 1971. Synthesis and extrusion of collagen by freshly isolated cells from chick embryo tendon. Biochim. Biophys. Acta 240: 358-369. 30. Bates, C. J., A. J. Bailey, C. J. Prynne, and C. I. Levene. 1972. The effect of ascorbie acid on the synthesis of collagen precursor secreted by 3T6 mouse fibroblasts in culture. Biochim. Biophys. Acta 278: 372-390.
We wish to thank Dr. L. Hayflick and Ms. Eva A. Pfendt for supplying the WI-38 cells and for helpful suggestions. We appreciate the aid of Dr. Katherine Chert and the excellent technical assistance of Mr. Douglas Tapper. The work was supported by National Institutes of Health grants AM15671 and HD07376.
