CHEMISTRY
Vol. 255, No. 19, Issue of October 10, pp, 9427-9433, 1980
Printed in U.S.A.
(Received for publication, August 3, 1979, and in revised form, January 14, 1980)
We havemeasuredthe diffision coefficientand Several models of the substructure of collagen fibrils based
weight average molecular weight of type I collagen on theoretical and experimental considerations have been
fibril fragments obtained by acid extraction of rat tail proposed. The microfibril, the smallest structural unit, con-
tendons and neutral extraction of lathyritic chick skin, tains from 4 to 8 collagen molecules in cross-section (7-10).
using laser light scattering techniques. The molecular Other evidence suggests that lateral order within a fibril is not
weightandtranslationaldiffusioncoefficientwere long range (11) and may be associated with the asymmetry of
found to be 8.05 f 0.40 X 10’ and 0.450 -C 0.04 X lo-’ the molecule.
cm2/s,respectively, for preparations which contain ag- In vivo collagen aggregation may beginbefore secretion
*
gregates in 0.01 M HCl, and 2.82 0.20 X 10’ and 0.780 since intracellular vacuoles containing aggregates with lengths
f 0.04 x lo-’ cm2/s for collagen single molecules. of 2 or more collagen molecules (12,13) have been observed.
Using these data, as well as the theoretical difhsion Similar aggregates have been observed both intracellularly
coefficient for a single collagen molecule, models for and extracellularly in a wide variety of tissues (14,15).In vitro
various staggering modes and aggregate mixtures studies were of type I collagen fibrillogenesisindicate that initiation
developed in an attempt to understand the structure of
the fibril components present in solution. It was foundcollagen fibril formation occurs by the conversion of single
of
that several models with 1 to 4 D (D = 67 nm)staggers molecules into 4 D staggered dimers and trimers, suggesting
fit the experimental data for the aggregated prepara- that the 4 D stagger may be thermodynamically preferred
tions. Analysis of the end-to-end distance of negatively over other staggering modes (16). Further aggregation into D
stainedsegmentlongspacingcrystallitesprepared staggered aggregates is a secondary event (16).
from solutionscontainingfibrilfragmentsandthe The purpose of this investigation has been to characterize
bandingpatternofpositivelystainedsegmentlong the physicochemical and ultrastructural properties of collagen
spacing crystallites suggest that collagen solutionssolutions con- derived from solubilized type I collagen fibrils and
tainlinearaggregateswith 4 D andpossibly 4.4 D to assess the organization of such fragments in respect to their
staggers in agreement with light scattering Ther- data. possible mode of packing in the native fibril.
mal denaturation studies demonstrate that the agpar-
ent melting point of cross-linked linear aggregates, 33.5 MATERIALS AND METHODS’
2 0.5”C, is identical with that of single molecules atpH Rat Tail Tendon Collagen-Acid-soluble rat tail tendon collagen
2.0. We conclude that a linear filament with predomi- was prepared (see Fig. 1) by dissolving tendons from young rats in
nant 4 D stagger is abasicunit of type I collagen 0.01 M HCI a t 4°C for 4 h followed by centrifugation for 30 min a t
30,000 X g. The supernatant was sequentially filtered through 0.8-,
fibrillar structure. 0.65, and 0.45-micron Millipore filters. SDS’ electrophoresis and
amino acid analyses of this preparation showed that it contained
predominantly type I collagen a chains and p and y components. The
relative proportions of a,p, and y components were determined by
The structure of collagen in solution has been studied chromatographing heat-denatured samples in 1.0 M CaCL on an
extensively both at the molecular (1-6) and supramolecular agarose A-5m column (2.0 X 100 cm) and computing the peak areas
levels (2-6). Early studies focused on the size and shape of of the effluent monitored at 230 n~ in a recording spectrophotometer.
collagen single molecules (1,2),while some of the later studies Peak identification was achieved by estimating the molecular weight
attempted to focus on supramolecular forms present in solu- using laser light scattering techniques.
tion (3-6). From these efforts, we know that extracts of type Lathyritic Chick Skin Collagen-Extracts of lathyritic chick skin
with 0.4 ionic strength potassium phosphate buffer, pH 7.6, were
I collagen containing tissues are mixtures of single molecules purified by repeated NaCl precipitation at both neutral and acidic
and high molecular weight aggregates. Since these fragments pH. The purified collagenwas desalted by dialysis at 4°C after
are derived from native collagen fibrils,structural analyses of resolubilization in 0.01 M HCI. Evaluation of chemical purity was done
these components may generate more information which can by amino acid and hexosamine analysis on HC1-hydrolyzed aliquots.
beused to develop a better understanding of localfibril
structure. ’ Portions of this paper (including part of “Materials andMethods”
and Tables 1 through 7) are presented in miniprint at the end of this
* This work was supported by Grant PCM-7818903 from the Na- paper. Miniprint is easily read with the aid of a standard magnifying
tional Science Foundation, Grants HL-18714 and GM 20007 from the glass. Full size photocopies are available from the Journal of Biolog-
National Institutes of Health, National Service Award No. 5 T32 HD ical Chemistry, 9650 Rockville Pike, Bethesda, Md. 20014. Request
07092-01, and by the Shriners Burns Institute. The costs of publication Document No.79M-1568, cite author(s), and include a check or
of this article were defrayed in part by the payment of page charges. money order for $1.65 per set of photocopies. Full size photocopies
This article must therefore be hereby marked “advertisement” in are also included in the microfilm edition of the Journal that is
accordance with 18 U.S.C. Section 1734 solely to indicate this fact. available from Waverly Press.
+ To whom correspondence should be addressed, at the Shriners The abbreviations used are: SDS, sodium dodecyl sulfate; SLS,
Burns Institute, 51 Blossom St., Boston, Mass. 02114. segment long spacing crystallites.
9427
Type 1Collagen inSolution
Preparation of Segment Long Spacing Crystallites (SLS)"sLS Tendon 1 g
crystallites of rat tail tendon collagen were prepared by dialysis of
collagen solutions in HC1, pH 2.0, a t 4°C uersus 0.01 M ATP in HC1,
pH 2.0, overnight. A drop of the dialysate was placed on carbon/
Formvar-coated copper grids.
Transmission Electron Microscopy-SLS of acid-soluble rat tail
tendon collagen were stained negatively with phosphotungstic acid
(PTA), pH 7.6, and positively with saturated aqueous uranyl acetate
I HC1, pH 2.0
4"C, 4 h
Solubyized tendon
Centrifugation -
pellet
discard
and viewed with a Philips 300 transmission electron microscope. The
end-to-end length of monomeric collagen SLS was determined from
negatively stained samples. The locations of the NH2 and COOH
1 30 min @ 30,000 g
Supernatant
termini of negatively stained SLS were determined by measuring a
molecular length from each free end. In most cases, the overlap of Collagen solution
negatively stained polymeric SLS was obvious by the positions of the Overall recovery of 0.609 g
NH2 and COOH termini of 2 adjacent molecules, as is seen in Fig. 6.4.
The molecular overlap was also determined by measuring the
distance between Bands 3 and 4 and 51 and 52 (17) of positively
stained polymeric SLS. Measurements on 10 different preparations
of monomeric SLS indicated that Bands 3 and 4 and 51 and 52 are
Recovery
"91%
1 Filter through
0.8-, 0.65-, and 0.45-p fdters
High molecular weight fraction
both about 7% from the ends of the molecule, while the distance (A?r = 8.05 f 0.40 X lo5,D20.w= 0.45 f 0.04 x 10" cm2/s)
between these bands is about 86%of the molecular length. Overall recovery
RESULTS 0.554 g
The relative composition of the high molecular weight acid-
soluble rat tail tendon as defined by chromatography on A-
5m was as follows: a chains, 33.6%;p components, 42.5%; and
y components, 23.9%;a recovery of 88%was obtained. Amino
acid analysis was typical of type I collagen; specifically,there (ar
Recovery
=M%
I
Filter through
0.22-p filter
Single molecule fraction
*
= 2.82 0.20 X lo5, DX).,,, = 0.80 f 0.04 X 10" cm2/s)
'OK
P \ :
$ 099C \"\x &
i 0
',
0
E ~- #\.: \
A
B
' 098Cl
0-
-Q"'""t&*"*
0 .
CONCENTRATION f m g l m f
FIG. 4. Translational difisioncoefficient back calculated to
standard conditions, Dz0.,, versus concentration for high mo-
lecular weight (0)and single molecule fractions (0)of pH 2.0
acid-solubilized rat tail tendon collagen at 4°C. All measure-
ments were made at 4' on solutions with Kayleigh factors in agree-
ment with those reported in the legend to Fig. 3 using a Chromatix
KMX-6 light scattering device modified for photon correlation.
'
from lathyritic and nonlathyritic rat skin are 7.40 X lo5 and
8.10 to 18.0 X lo5,respectively, and by Fletcher (5) of the two-
DISSOLUTION
n+
GELATION
component nature of acid solutions.
\ HE AT
In summary, our findings suggest that type I fibrils are at
least partially composed of 4 D staggered filaments as dia-
\--
---.-
-
"
t"
grammed in Fig. 7 and that the 4 D stagger is probably more
stable than other types of interactions in acid or neutral
solutions at 4°C. The ability of collagen to reassemble into
FIG. 7. Schematic representation of dissolution and gelation native fibrils probably derives from sequence information
of type I collagen from a collagen fibril. The relationship between contained in the nonhelical ends and in the imino acid-poor
monomers and 4 D staggered dimers, trimers, and tetramers in the regions about 0.4 D from each end. How or if this information
standard two-dimensional presentation of the collagen fibril is high-
lighted. The 0.4 D overlap or 4 D stagger of molecules reflects a facilitates fibril assembly in uzuo is uncertain; however, in uiuo
particularly stable form of aggregation in solution for both lathyritic morphological observations that tendon fibroblasts and cor-
and nonlathyritic collagens. Dissolution of the fibrilrenders such neal epithelial cells (12,13,15)contain intracellular aggregates
aggregates soluble as indicated by the fragments in the lower portion. 2 and 3 molecules in length may reflect this preference.
Gelation of such solutions apparently involves the same 4 D staggered
intermediates (16). REFERENCES
1. Boedtker, H., and Doty, P. (1956) J . Am. Chem. SOC.78, 4267-
4280
is close to the melting point of collagen at pH 2.0 determined 2. Davison, P. F., and Drake, M. P. (1966) Biochemistry 5,313-321
by viscometry (26, 27) and optical methods (26). After dena- 3. Obrink, B. (1972) Eur. J. Biochem. 25,563-572
turation, the single molecule solution had a weight average 4. Yuan, L., and Veis, A. (1973) Biopolymers 12, 1437-1444
molecular weight of 1.30 X lo5, suggesting that the weight 5. Fletcher, G . C. (1976) Biopolymers 15, 2201-2217
fraction of a chains must be at least 0.6, assuming a molecular 6. Thomas, J. C., and Fletcher, G. C. (1979) Biopolymers 18, 1333-
weight of 0.95 X lo5 and 1.80 X lo" for 01 chains and ,8 1352
7. Smith, J. W. (1968) Nature (Lond.) 219, 157-158
components, respectively. In comparison, the molecular 8. Veis, A., Anesey, J., and Mussell, S. (1967) Nature (Lond.) 215,
weight after denaturation of the high molecular weight frac- 931-934
tion containing linear aggregates was found to be 3.4 x lo5, 9. Hosemann, R., Dressig, W., and Nemetschek, TH. (1974) J. Mol.
which is greater than the molecular weight of a y component Biol. 83,275-280
and therefore must reflect the fact that at least some of the 10. Piez, K. A. (1975) in Extracellular Matrix Influences on Gene
linear aggregates are covalently cross-linked together. This Expression (Slavkin, H. C., and Greulich, R. C., eds) pp. 231-
fact is significant since, in the absence of cross-linking, it could 236, Academic Press, New York
11. Hukins, D. W. L., and Woodhead-Galloway,J. (1977) Mol. Cryst.
be argued that linear aggregates are a quasi-stable solution Liq. Cryst. 41, 33-39
state and do not accurately represent fibril fragments. 12. Trelstad. R. L. (1971) J. Cell Biol. 48.689-694
In addition tothe experiments on acid-soluble rat tail 13. Trelstad; R. L., and Hayashi, K. (1979) Deu. Biol. 71, 228-242
tendon collagen, we have studied a purified neutral extract of 14. Weinstock, M. (1977) J. Ultrastruct. Res. 61, 218-229
lathyritic chick skin collagen in 0.01 M HCl. This preparation 15. Bruns, R. R., Hulmes, D. J. S., Therrieu, S. F., and Gross, J.
was salt-precipitated several times so that induced aggregation (1979) Proc. Natl. Acad.Sci. U. S. A . 76, 313-317
16. Silver, F. H., Langley, K. H., and Trelstad, R. L. (1979) Biopoly-
during purification was a distinct possibility. Table 1 and Fig. mers 18, 2523-2535
3 compare the weight average molecular weight, average 17. Bnms, R. R., and Gross, J. (1973) Biochemistry 12,808-815
diffusion coefficient, and the experimental normalized corre- 18. Tanford, C. (1961) Physical Chemistry of Macromolecules, Chap.
lation function to thevalues of these parametersobtained for 5 and 6, John Wiley & Sons, New York
the acid-soluble rat tail tendon solution at 4°C. These data 19. Schwartz, D., and Veis, A. (1978) Connect. Tissue Res.6,185-190
indicate that the lathyritic skin collagen solution and the rat 20. Doty, P., and Edsall, J. T. (1951) Adu. Protein Chem.6, 50-55
21. Ford, N. C., Jr. (1972) Chem. Scr. 2,193-206
tail collagen in acid appeartobequite similar and that 22. Berne, B. J., and Pecora, R. (1976) Dynamic Light Scattering,
purification or the lathyritic state does not influence the state Chap. 8, John Wiley & Sons, New York
of aggregation. These results suggest that the type I native 23. Cohen, R. J., and Benedek, G. B. (1976) J. Mol. Biol. 108, 151-
fibril from at least two different vertebrate tissues probably 178
contains 4 D staggered filaments at least 3 or 4 molecules in 24. Rauterberg, J., von Bassewitz, D. B. (1975) Hoppe-Seyler's 2.
length which appear tobe identical with filaments which form Physiol. Chem. 356,95-100
25. Veis, A., and Yuan, L. (1974) Biopolymers 14,895-900
during the initial stages of collagen heat gelation (16). The 26. Hayashi, T., and Nagai, Y. (1973) J. Biochem (Tokyo) 73, 999-
specificity for the 4 D staggered interaction probably involves 1006
the imino acid-poor regions located about 0.4 D from each end 27. Dick, Y. P., and Nordwig, A.(1966)Arch. Biochem. Biophys. 117,
of the moleculewhich from stereochemical considerations 466-468
may be more flexible than theneighboring parts of the colla- 28. Comper, W. D., and Veis, A. (1977) Biopolymers 16,2133-2142
gen helix. The importance of the nonhelical ends on fibril 29. Hayashi, T., and Nagai, Y. (1973) J. Biochem (Tokyo) 74, 253-
262
formation in vitro (28-30) probably reflects the necessity for 30. Helseth, D. L., Jr., Lechner, J. H., and Veis, A. (1979) Biopolymers
the proper alignment of the telopeptides and imino acid-poor 18,3005-3014
regions (30, 31). 31. Silver, F. H., and Trelstad, R. L. (1979) J. Theor. Biol. 81, 515-
The presence of aggregates in acid solutions agrees with 526
9432 Type I Collagen in Solution
component
OD
0.858
0.718
1
2
2a = 300.
2a = 3 0 0 . 2b --
Zb = l.5nm
3.0nn
--
OD 0.758 3 2a = 3 0 0 . 2h 3.0nm
OD 0.717 4 La 300, 2h = 4.Onm
on
on
0.717
0.717
5
6
2a
2a - 300,
300,
Zb
2b
= 4 Onm
= 4.Onm
In
10
1D
0.642
0.559
0.471
2
3
4
2a
2a
21
--
~ 368,
436,
504,
Zh
2h
2h
= 3.0nm
--
= 3.Onrn
4.0nm
10 0.425 5 l a = 972, 2b 4.0nm
2D 0.559 2 2a = 436, 2h = :.Om
2D 0.446 3 2a = 572, 2b = 3.0nm
-- -
2D 0.316 4 Za = 708. 2b = 4.0nm
2D 0.308 5 2a 844, 2b = 4.Onm
30 0.555 2 2a 504, ?h 1.5nm
3n 0.415 3 2a = 708, 2h = l.5nm
3n 0.337 4 Za = 9 1 2 , 2b = 1 5nm
30 0.281 5 2a = 1116, 2h = I S n m
40 0.498 2 La = 572. 2h 1.Sna~
..
Calculated u r l n g e q u a r l o n 4 bee Methods
~
Table 3
rrlners x
cd/sec
1.667 0.685
1.439 0.728
1.250 0.772
1.111 0.815
0.8 2.14 0.b19
0.6 1.667 0.079
0.4 1.364 0.734
0.2 1.154 0.798
0.8 2.5 11.548
0.0 1.818 0 .6 2 6
0.4 1.750 0.703
0.2 1.240 0.7 8 0
0.8 2.78 0.~12
0.6 1.923 u.svn
0.4 1.470 0.685
0.2 1.279 0.771
0.2 2.14 0.625
0.4 2.31 0.009
0.6 2.50 0.192
0.8 2.73 0.576
where .2
,I 0.8 3.85 a 408
0.4 O.b 3.13 0.112
0.6 0.4 2 b3 0 555
0.8 0.2 2.27 0.599
Type I Collagen in Solution 9433
Table 4
Table 6
Table 7
4D Staggered F l b r l l F r a g m e n f Models
Tahle I Weighrlng Factors
M o d e l s 3~ D S t a g g e r
Olner Trlmer retramer D~~ x IO'? .sl * c ( ~ A ~ )
Weighting F a c t o r s C.j/SeC a, I
0.5 0.5 0.428 2.5 1.~20
rTlmer ti20 Y 0.1 0 4 0 5 0.4Z7 2.9 0.946
C d l S e C
0.2 0.3 0.5
0.2 .
11 8 1.667
0.475 2.8 1.1180
0.4 O.b 1.429 0.1 0.4 0.5
0 0 0 4b5 2.4 1 064
0.4 1.250
0 8 0.2 1.111 0.2 0.3
0.2 0.5 0.502 2 3 1.210
0.8 2.14
!I 4 . 0.0 1.667 0.2 0.3 0.25 0 25
<I 6 . 0.4 1.364
U.489 2.55 1.100
u.n 0.2 1.154 0.3 0.2 0.25 0.25 O . 5 l 1b. 3 5 0 2.45
0.: 0.8 2.50
0.4 0.6 1.818 0.1 0.4 0.25
(I .h 0.25 0.446 2,bS 1.090
0.4 1.429
U.8 0.2 1.176
(I. 2 0.8 2.78
0 4 0.6 1.923
!J h 0.4 1.470
0.2 0.8 1.279
0 2 0.8 2.73
0.4 0.6 2.50
D,b 0.4 2.31
0 .8 0.2 2.14
0.2 3.33
0.4 2.86
0.6 2.50
0 8 2.22
0 2 3.85
0.4 3.13
0.0 2.63
0.n 2.27