Binding of hyaluronic acid to mammalian fibrinogens

Biochimica et Biophysica Acta, 1034 (1990) 39-45

39

Elsevier BBAGEN 23281

Binding of hyaluronic acid to mammalian fibrinogens Stephen J. Frost and Paul H. Weigel Department of Human Biological Chemistry and Genetics, The University of Texas Medical Branch, Galveston, TX (U.S.A.)

(Received4 August 1989) (Revised manuscript received27 November 1989)

Key words: Hyaluronicacid; Fibrinogen;Extracellularmatrix; Wound healing; (Human)

We have postulated that the interaction of hyaluronic acid (HA), an extracellnlar matrix glyeosaminoglycan, with fibrin is important during the early stages of wound healing and inflammation (J. Theor. Biol. 11~.219;, 1986), and have demonstrated the specific binding of 12sI-iabeled HA to human fibrinogen (J. Biol. Chem. 261:12 586; 1986). To determine whether HA binding is limited to human fibrinogen, we tested the ability of fibrinogens from various mammalian species to bind t2sI-HA using a dot-blot assay. Increasing amounts of fibrinogen were adsorbed to nitrocellulose, and incubated with 12sI-HA in the presence or absence of a 100-fold excess of nonradiolabeled HA to assess specific binding. In three independent experiments, the amount of 1251-HA bound/mg fihrinogen was determined from the slope derived by linear regression analysis of specifically bound t2SI-HA versus protein concentra, tion. A Student's t-test was performed to determine whether the slopes were statistically greater than zero. HA binding was considered statistically significant when P < 0.05 was obtained by this analysis. Rabbit and dog fibrinogens significantly bound HA in all three trials. Baboon fibrinogen demonstrated significant HA binding in two of three trials. Pig, sheep and goat fibrinogens bound HA significantly in only one of three trials, whereas horse, rat and cow fibrinogens did not bind HA significantly at all. We conclude that fibrinogen from mammalian species other than human can specifically bind HA. The ability of fibrinogen to bind HA appears to correlate with an evolutionary divergence that separated human, baboon, dog, rabbit and rat from cow, pig, horse, goat and sheep.

Introduction HA, a nonsulfated glycosaminoglycan, is a ubiquitous component of the mammalian extraceUular matrix [1]. HA aids in the organization of the extraceUular matrix by interacting with receptors on cells such as fibroblasts [2,3], and by binding to specific proteins in the matrix like hyaluronectin [4], proteoglycans and hnk proteins [5]. HA is imphcated in events associated with wound healing such as angiogenesis [6], and cell migration into the wound [7]. Human fibrinogen, the precursor of fibrin which forms a blood clot at the wound site, binds HA oligosaccharides ( M r = 35000) specifically with a K d of 10-7 M [8]. This interaction was predicted as an important part of a model for the early events in inflammation and wound healing [9]. It was postulated that fibrin in the clot would bind newly synthesized HA

Abbreviations: TBS, Tris-bufferedsaline; HA, hyaluronicacid.

Correspondence: P.H. Weigel, Department of Human Biological Chemistry and Genetics, The University of Texas Medical Branch, Galveston, TX 77550, U.S.A.

produced at the wound to create a new fibrin-HA matrix. The possible consequences of this transitory matrix could include making the clot more porous, increasing the water retention of the clot and stimulating inflammatory cell functions. This change in structure and composition of the early wound matrix would allow cells to infiltrate the clot and further modify the changing wound tissue [9]. If this hypothesis is true, then fibrinogen from species other than human might also bind HA. Therefore, we developed a rapid dot blot assay to determine whether HA binds to commercially available fibrinogens from nine different mammalian species.

Methods Materials. HA from human umbilical cord, Nonidet P-40, and fibrinogens from all species used in this study were purchased from Sigma Chemical Co, St. Louis, MO. Human fibrinogen was also obtained from US Biochemical Corp., Cleveland, OH. Multiple lots of all fibrinogens were obtained except for rabbit and goat. Nitrocellulose (0.45 gm) and dot-blot manifold Model No. SRC-76/0 were obtained from Schleicher and

0304-4165/90/$03.50 © 1990 Elsevier SciencePubhshers B.V. (BiomedicalDivision)

40 Schuell, Keene, NH. The Red Rocker platform shaker was from Hoefer Scientific Instruments, San Francisco, CA. Lux tissue culture four-well multiplates were obtained from Miles Scientific, Naperville, IL. Rotator, model number 69, was obtained from Technilab Instruments, Pequannock, NJ. Bradford reagent was obtained from Pierce, Rockford, IL. 125I-Fibrinogen was prepared as described by Fraker and Speck [10]. Preparation of 125I-HA. HA was purified and the unique alkylamine derivative of HA and its BoltonHunter adduct were prepared and iodinated as described [11] with the following modifications. The starting HA for synthesis of the aminohexyl-HA was M r = 60 000 and a 2-fold molar excess of sodium periodate to reducing ends was used during the synthesis. Unreacted diaminohexane was removed from HA-amine by the following procedure. The pH of the reaction mixture was adjusted to 2.5 with acetic acid, and 1 / 1 0 volume of 5 M NaC1 was added. The HA-amine was then precipitated by adding 3 vol. of ethanol [11]. After 30 rain at - 2 0 o C, the precipitate was pelleted by centrifugation at 14 000 × g for 30 rain at - 2 0 ° C. The pellet was dissolved in distilled water, the pH was adjusted to 11 with N a O H and the HA-amine was again precipitated with ethanol in the presence of 0.125 M NaC1 and pelleted by centrifugation. This procedure of an acidic ethanol precipitation followed by a basic ethanol precipitation was repeated three times. Buffers. TBS contained 154 m M NaC1, 10 m M TrisHC1 (pH 7.4). Buffer 1 contained 143 mM NaC1, 6.8 mM KC1 and 10 mM Hepes (pH 7.4). Dot-blot assay. The assay is based on the 'Western' blot procedure of Burnette [12]. The nitrocellulose and Whatman 3 MM chromatography paper were soaked in TBS, placed in a dot-blot manifold with the nitrocellulose above the chromatography paper, and subjected to a mild vacuum. Different volumes of a 250 # g / m l solution of fibrinogen in Buffer I were added to individual wells. Usually, not more than 45 #g of fibrinogen was added to each well. 500 #1 of TBS were allowed to filter through the nitrocellulose and the vacuum was removed. The nitrocellulose was taken from the manifold, incubated for 1 h with 5% BSA in TBS at 25 o C, and then rotated on a Technilab Instruments rotator in a sealed plastic bag at 2 5 ° C for 2 h with 6 / ~ g / m l of 125I-HA in TBS containing 0.1% BSA and 2 m M CaC12. An identically treated piece of nitrocellulose was incubated with a 100-fold excess of nonradiolabeled HA. After incubation, the nitrocellulose sheets were transfeared to 7 × 22 cm plastic trays and washed four times in 10 rain with TBS containing 0.01% Nonidet P-40. Under these conditions, only a minimal amount of fibrinogen was lost from the nitrocellulose ( < 10%) and the non-specific binding of 125I-HA to > 30 /~g human fibrinogen was < 33% of the total binding. The nitrocellulose was air dried and the blots of individual wells

were excised with a cork borer, placed into 12 × 75 mm gamma tubes and 1251 was determined. Fibrinogen clottability. In order to assess the activity and approximate degree of purity of the various fibrinogen preparations, we determined the protein incorporated into thrombin-induced clots in the presence of Ca 2+ [13]. These values ranged from 82% to 98%. Our values for clottable protein were equal to or greater than those reported by the commercial source. General. ~zsI Radioactivity was determined using a Packard Multiprias 2 gamma spectrometer. Protein was determined by the method of Bradford [14] using BSA as the standard.

Results In order to compare the HA binding capability of fibrinogens from various species, we needed a rapid assay to quantitate 125I-HA binding. We have previously bound lzSI-labelled human fibrinogen to HASepharose and then eluted the fibrinogen specifically with HA [8]. However, this is time consuming and labor intensive when examining many samples. Therefore, we developed a dot-blot assay to compare HA binding to a number of different fibrinogens quickly. The dot-blot assay allows for easy separation of 125I-HA bound to fibrinogen from the free 125I-HA and for easy quantitation of specifically bound HA. Briefly, after adsorbing the soluble proteins to nitrocellulose, the nitrocellulose is incubated with radioactive HA and then washed with a detergent solution to remove the unbound lzSI-HA. The individual blots are cut from the nitrocellulose and the 125I is determined. The amount of 125I-HA specifically bound per mg fibrinogen is calculated by plotting 125I-HA specific binding versus the amount of fibrinogen added to nitrocellulose. We performed several controls to assess the linearity of fibrinogen binding to nitrocellulose and whether the bound 125I-fibrinogen was removed from nitrocellulose by detergent washes. The amount of human 1251fibrinogen bound to nitrocellulose increased linearly up to 40 #g of protein (Fig. 1). Within this range, 83% of the added protein was adsorbed. We also examined the effect of increasing concentrations of the nonionic detergent, Nonidet P-40, on retention of human 125Ifibrinogen bound to nitrocellulose. In the presence of 0.2% (v/v) Nonidet P-40, a concentration normally used in 'Western Blot' procedures [12], > 90% of the 125I-fibrinogen bound to the nitrocellulose was removed in a 1 h incubation (Fig. 2). Even 0.02% Nonidet P-40 removed > 70% of the prebound 125I-fibrinogen (Fig. 2). TBS containing 0.01% Nonidet P-40 did not significantly remove any fibrinogen ( P < 0.05). Nonetheless, this detergent concentration was high enough to minimize the nonspecific binding of 125I-HA. Therefore, the

41

i

o

I

I

I

I

N

!

u

I

m

I

I

~

15

•"

10

30

5

~ ,-d, 20 "5 E -- Z _

10 0

i

I

0

a z

15 30 45 60 125 I-FIBRINOGEN ADDED TO NITROCELLULOSE(pg)

Fig. 1. Human 125I-fibrinogen binding to nitrocellulose. Variable amounts of a 0.25 mg/ml solution of human 125I-fibrinogen in Buffer 1 were applied onto the nitrocellulose and washed with 500/L1 of TBS. The individual blots were cut and the bound 125I was determined. Each point represents the mean of triplicates and the error bars represent the sample standard deviation. The correlation coefficient of the line determined by least-squares linear regression analysis from 0 to 45 Fg was 0.995 with a slope of 8.3/~g fibrinogen bound/10 /~g fibrinogen added.

t

• I

I

1 I

I

i

--

I

I

I

I

BSA

o m

_.q 15 0 0 Z 0 n..

3

10 5

G

0 I

I

Goat IgG

15

1C

nitrocellulose washes in the subsequent assays were done with 0.01% Nonidet P-40 for less than 1 h. Since fibrinogen quantitatively binds to the nitrocellulose and remains adsorbed during the subsequent washes and manipulations, we then examined the 1251HA binding to immobilized human fibrinogen using the dot blot assay (Fig. 3). The 125I-HA binding to human fibrinogen was linear up to at least 45 /~g protein. Furthermore, the binding was judged to be specific because it competed > 66% with a 100-fold excess of nonradiolabeled HA when human fibrinogen was at a concentration of > 30 /~g. BSA and IgG, control pro-

0

1

I

1

I

/

10

20

30

40

5o

PROTEIN ADDED TO NITROCELLULOSE ( p g )

Fig. 3. 125I-HA binding to various proteins. Human fibrinogen, BSA and goat IgG were adsorbed to nitrocellulose and a dot-blot assay was performed as described in Methods. Total binding of 125I-HA is shown by open circles and nonspecific binding is shown by closed circles. Each point represents the mean of triplicate determinations and the error bars represent the sample standard deviation. The background binding in the absence of adsorbed protein has been subtracted. The lines were derived by least-squares linear regression analyses.

O

tr

~3e too

0 0

75 ~

0 25 ~ .002

.005

.01

.02

.05

10

.20

NONIDET P-40 CONCENTRATION (% ~ v/v)

Fig. 2. Effect of Nonidet P-40 concentration on retention of human 125,I-fibnnogen bound to mtrocellulose. 30/~g of human 125I-fibdnogen in Buffer i was adsorbed to nitrocellulose as described in Fig. 1. Strips of nitrocellulose containing four blots were transferred to 24 nun × 67 mm tissue culture four-well multiplates and incubated with increasing Nonidet P-40 concentrations in TBS for 1 h at 25 ° C on a Red Rocker shaker platform. The detergent solutions were aspirated, the individual blots were cut out and the radioactivity was determined. The amount of protein bound was calculated from the specific activity of the 125I-fibrinogen solution. Each point represents the mean from four blots. The error bars represent the sample standard deviation. The dashed line represents the fibrinogen bound to nitrocellulose after a i h incubation with TBS alone. •

.

teins, showed no significant protein-dependent 125I-HA binding in a parallel assay (Fig. 3). We conclude that the dot-blot assay is suitable for detecting specific tESIHA binding to proteins. When we examined other immobilized mammalian fibrinogens, we found variabihty among species in their capacity to bind 125I-HA (Fig. 4). Some species demonstrated consistent significant 12SI-HA binding, some consistently showed no significant binding and some species showed inconsistent binding. The results from three independent experiments are summarized in Table I, with the species hsted in descending order of evolutionary similarity to human [15]. In each experiment freshly dissolved fibrinogen solutions and different preparations of 1z5I-HA were used. Two different lots of fibrinogen from each species were used in these experiments with the exception of those from rabbit and goat which were from a single lot. In experiments 1 and 2 the same lot of fibrinogen was used. For some species (i.e.,

42 "

I

I

I

I

I

GOAT

1o

I

~-

0

I

I

I

I

I

COW

!

i

i

!

I

I

I

I

I

I

I

I

I

I

40

50

BABOON

10

0 z_ m

~

•

HORSE

I

I

I

I RAT

10

o 5

.J <

0

>."r _1

I

I

I

I

I

PIG

o u.

RABBIT

10

0 I

I

I

I

I

I

I

SHEEP

DOG

10

I

|

I

I

I

10

20

30

40

50

0

I

I

I

10

20

30

I

_

FIBRINOGEN (pg)

Fig. 4. Specific I25I-HA binding to fibrinogens from various species. Fibrinogens from different mammalian species as indicated in the panels were adsorbed onto nitrocellulose, and a dot blot assay was performed as described in Methods. The fibrinogen content of the commercial preparations was estimated from the product of (protein concentration) multiplied by (the fraction of protein that was clottable). In each case two independent experiments are shown by the different symbols, Each point represents the average of triplicate determinations and the error bars represent the sample standard deviation. The amount of specifically bound 125I-HA/mg of fibrinogen was determined from the slope of the line derived by least-squares linear regression analysis. At least nine individual points (three values for 125I-HA specific binding for at least three different fibrinogen concentrations) were used to determine the slope of each line.

goat, sheep and pig), there was considerable variability between experiments due to the low amount of specific binding exhibited. Additionally, these species had greater standard deviations at each individual point. In general, the species that consistently bound HA (i.e., baboon, rabbit and dog) in this assay also demonstrated reproducible results. The overall trend is that fibrinogens from the species more closely related to humans (Fig. 5) specifically bind ~25I-HA, whereas more divergent species do not. Fibrinogens from baboon, dog and rabbit consistently bound significant amounts of ~25I-HA. Pig, sheep and goat fibrinogens significantly bound 125I-HA only once in three trials. Horse, rat and cow fibrinogens did not

show significant specific binding to ~25I-HA in any experiment. Discussion

The binding of endogenous or newly synthesized HA to fibrin(ogen) may be important in wound healing [9]. Human fibrinogen specifically binds to HA [8] and HA alters the ability of purified human fibrinogen to form thrombin-induced clots in vitro [13]. The presence of small HA oligosaccharides ( M r ~ 32000) decreases the lag time for clotting, increases the rate of fibrin polymerization and increases the effective diameter of the fibrin polymer bundles. The decrease in clotting time

43 TABLE I

125I-HA specific binding to fibrinogens from various mammals The specific 125I-HA binding to mammalian fibrinogens was quantitated as described in Fig. 4. The specific binding at a particular amount of fibrinogen was determined by subtracting the average nonspecific binding (n = 3) from each individual value of the total binding. The amount of 125I-HA bound per mg fibrinogen for each sample tested was obtained from the slope of the line derived from the least-squares linear regression analysis of the protein titration. When a positive slope for the specific binding of 125I-HA vs. mg of fibrinogen was obtained, the slope was analyzed by a Student's t-test to determine whether it was significantly greater than zero; t = slope + (the sample standard deviation of the slope). If P < 0.05, the slope was considered significantly different from zero and the corresponding values in the table are labeled with an asterisk. A zero means that the slope of the line was _< 0. Two different lots of fibrinogen from all species were tested with the exception of rabbit and goat which were from a single lot. In experiments 1 and 2 one lot was used and in Experiment 3 a second lot. Negative controls for these experiments were BSA, goat IgG and lysozyme, which exhibited no statistically significant binding to lzsI-HA (Fig. 3). Species

Baboon Rabbit Rat Dog Horse Pig Cow Goat Sheep

Specific 125I-HA binding (fmol/mg fibrinogen) Expt. 1

Expt. 2

Expt. 3

89 144 35 150 93 0 0 201 143

116 154 0 151 131 93 0 78 0

191 * 113 * 0 93 * 155 46 0 0 0

* * *

*

* * * *

caused by HA was also observed by Pandolfi and Hedner [16]. This interaction could be important in remodeling of the fibrin clot when HA is synthesized at the wound site. This HA-fibrin(ogen) interaction may be one of the reasons why it is important that HA levels in the blood of healthy individuals are low, even though there is a large HA turnover in the body every day [17]. Elevated blood HA levels could interfere with normal Human(++) Baboon(*+)

~Rabbit

(*+÷) Rat

Sheep(+) Goat (+) Cow XX~///

Fig. 5. Mammalian evolutionary tree and fibrinogen binding ability. The diagram is based on six different proteins, including fibrinopeptides A and B, used by Penney and Hendy to construct an evolutionary tree [15]. One plus sign is shown for each of three separate experiments in which the fibrinogen tested significantly bound HA.

hemostasis and could be deleterious for the organism, for example by causing increased thrombosis. The effect of HA on fibrin clot formation may be a possible explanation for the disseminated intravascular coagulation (DIC) syndrome that can occur in patients who have experienced burns or other severe trauma, surgery, infectious diseases, liver disease or injury, or childbirth [18]. In these situations patients are likely to have greatly elevated levels of circulating HA [19] that could bind to fibrinogen and stimulate clot formation as observed in vitro. In addition, the structure of a fibrin clot formed in the presence of HA may be abnormal and its subsequent function and interaction with cells in vivo could be different. Despite the low concentration of HA in blood, it is known that the HA content of the early wound increases and then decreases [20,21]. We have proposed [9] that the transient increase in HA content in the early wound helps to set the biological clock for the woundhealing process. HA is important both as a structural and a regulatory molecule and influences processes such as clot swelling, cell infiltration, phagocytosis and angiogenesis. The first and most critical of the predictions derived from the proposed wound healing model was that HA and fibrin(ogen) should interact with one another either by direct binding of HA to fibrin(ogen) or by a molecule that crosslinks them [9]. This prediction has been partially validated, since human fibrinogen specifically binds HA [8,13]. The goal of the present study was to screen a broad range of mammalian species for their ability to bind 125I-HA directly. In addition to developing a method for the rapid screening of specific HA binding to proteins, we find that fibrinogens from some species can bind HA. These species include rabbit, dog and baboon. An interesting result is that, with the exception of rat, mammalian species closer to humans on the evolutionary 'tree' bind 125I-HA better in the dot-blot assay than the more divergent species (Fig. 5). The ability of fibrinogen to bind HA appears to correlate with an evolutionary divergence that separated human, baboon, dog, rabbit and rat from cow, pig, horse, goat and sheep [15]. This divergence occurred approx. 80 million years ago [22]. Therefore, it is likely that the HA binding site on fibrinogen has evolved within the last 80 million years. Mammalian and perhaps even nonmammalian species, which do not demonstrate a direct specific binding of fibrinogen to HA, may possess a molecule that can crosslink HA to fibrin(ogen) during wound healing. We compared the full length amino acid sequences of fibrinogen chains from different species using the Bionet data bank and the ALIGN program. While all the human fibrinogen chain sequences are known, only the rat Aa, bovine Bfl chain precursor, and the rat and

44 Human rat

KDLLPSRDRQHLPLIKMKPVPDLVPGNFKSQLQKVPPEWKALTDMPQMRMELERPGGNEI E AK Y A S S EG E R KDGL

250 251

human rat

TRGGSTSYGTGSETESPRNPSSAGSWNSGSSGPGSTGNRNPGSSGTGGTATWKPGSSGPG REEIYQGDSR DSATR ............... KIR- MTP H SGY R S

310 295

human rat

STGSWNSGSSGTGSTGNQNPGSPRPGSTGTWNPGSSERGSAGHWTSESSVSGSTGQWHSE D N G TT SDD TWGA S S . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

370

human rat

SGS FRPD S PG SGNARPNNPDWGTFEEVSGNVS PGTRREYHTEKLVTSKGDKELRTGKEKV

430

.........

373

LK S

E S FG SS

A

K

G

LI

N

322

Fig. 6. Comparison of amino acid sequences of human and rat Aa chain precursors of fibrinogen. Dashed lines indicate deleted residues when the two sequences are afigned using the ALIGN program in the BIONET system. Only the differences in the rat sequence are shown using the one letter code for amino acids. The asterisks indicate clusters of two or more hydroxyl-containing amino predicted to mediate binding to HA.

bovine 3' chain precursor sequences have been determined. N o fibrinogens from other species that had significant H A binding are completely sequenced. However, we could compare fibrinogens from a species that binds H A very well (human) to two species that do not bind H A at all (rat and bovine). The h u m a n and bovine B/3 [23] and T chains [24,25] are almost identical ( > 75% homology). The greatest variations occur in the Nterminal region of the B/3 chain, which contains the signal sequence and fibrinopeptide B sequence, and the C-terminal region of the bovine Y chain, which has a twelve amino acid insert [25]. Rat and h u m a n 3' chain precursors are also very similar ( = 80% homology). The only chain that has a dramatically different sequence c o m p a r e d to h u m a n is the rat A a chain [26]. Rat (r) and h u m a n (h) A a precursor sequences [26-28] align almost perfectly until rLys-248 (Fig. 6). The rat chain then has a divergent sequence (19 out of 20 residues) and a deletion of 15 amino acids corresponding to hPro-267 to Ser-281. The next stretch of 55 amino acids in the rat A a precursor chain is 53% identical to the h u m a n chain. There is then a major deletion in the rat chain of a 42 amino acid sequence corresponding to hThr-338 to Pro379. We hypothesize that the region of the h u m a n A a chain from hGly-254 to hGly-380 has appropriate structural properties to mediate the binding of fibrin(ogen) to polysaccharides such as HA. This region is rich in hydroxyl-containing residues, threonine and serine, that could be used for hydrogen binding to a polysaccharide. Also, the tertiary structure in this region of the A a chain is predicted to be a r a n d o m coil [28]. This feature would allow access of the linear protein polymer to the linear H A molecule so the two could interact and form a complex. Doolittle et al. [28] have noted that there is a concensus sequence of 13 amino acids (NPGS._SSGPGSTG_TW) repeated ten times in this

region. Five of these residues are hydroxyl amino acids and the concensus sequence contains two hydroxyl residues. This would be important for the ability of fibrinogen to bind to HA, since the polysaccharide is also c o m p o s e d of repeating clusters of hydroxyl groups. The h u m a n sequence contains 14 clusters of hydroxyl amino acids in this region of the A a chain, while the rat sequence only has four clusters [26]. Therefore, we predict that this d o m a i n in the A a chain of the h u m a n fibrinogen molecule will bind to HA.

Acknowledgements We thank Janet Oka for preparing the figures and Betty Jackson for help in preparing the manuscript. This work was supported by National Institutes of Health grant G M 35978.

References 1 2 3 4 5 6 7 8 9 10 11 12 13

Laurent, T.C. (1987) Acta Otolaryngol. (Stockh.) Suppl. 442, 7-24. Underhill, C.B. and Toole, B.P. (1979) J. Cell Biol. 82, 475-484. Turley, E.A. and Torrance, J. (1984) Exp. Cell Res. 161, 17-28. Delpech, B., Bertrand, P., Hermelin, B., Delpech, A., Girard, N., Halkin, E. and Chauzy, C. (1986) Front. Matrix Biol. 11, 78-89. Hascall, V.C. (1977) J. Supramol. Struct. 7, 101-120. West, D.C., Hampson, I.N., Arnold, F. and Kumar, S. (1985) Science 228, 1324-1326. Hakansson, L. and Venge, P. (1987) J. Immunol. 138, 4347-4352. LeBoeuf, R.D., Raja, R.H., Fuller, G.M. and Weigel, P.H. (1986) J. Biol. Chem. 261, 12586-12592. Weigel, P.H., Fuller, G.M. and LeBoeuf, R.D. (1986) J. Theor. Biol. 119, 219-234. Fraker, P.J. and Speck, J.C., Jr. (1978) Biochem. Biophys. Res. Commun. 80, 849-857. Raja, R.H., LeBoeuf, R.D., Stone, G.W. and Weigel, P.H. (1984) Anal. Biochem. 139, 168-177. Burnette, W.N. (1981) Anal. Biochem. 112, 195-203. LeBoeuf, R.D., Gregg, R.R., Weigel, P.H. and Fuller, G.M. (1987) Biochem. 26, 6052-6057.

45 14 15 16 17 18 19

20 21

Bradford, M.M. (1976) Anal. Biochem. 72, 248-254. Penny, D. and Hendy, M. (1986) Mol. Biol. Evol. 3, 403-417. Pandolfi, M. and Hedner, U. (1984) Ophthalmology 91, 864-866. Laurent, U.B.G. and Laurent, T.C. (1981) Biochem. Int. 2, 195199. Minna, J.D., Robboy, S.J. and Colman, R.W. (1974) Disseminated Intravascular Coagulation in Man. Thomas, Springfield, IL. Whelan, J. and Evered, D. (eds) (1989) The Biology of Hyaluronan, Ciba Foundation Symposium 143, John Wiley & Sons, Chichester. Bentley, J.P. (1967) Ann. Surg. 165, 186-191. Toole, B.P. (1976) in Neuronal Recognition (Barondes, S.H., ed.), pp. 275-325, Plenum Press, New York.

22 Doolittle, R.F. (1983) Ann. N.Y. Acad. Sci. 408, 13-27. 23 Watt, K.W.K., Takagi, T. and Doolittle, R.F. (1979) Biochemistry 18, 68-76. 24 Chung, D.W., Chan, W-Y and Davie, E.W. (1983) Biochemistry 22, 3250-3256. 25 Brown, W.M., Dziegielewska, K.M., Foreman, R.C. and Saunders, N.R. (1989) Nucleic Acids Res. 17, 6397. 26 Crabtree, G.R., Comeau, C.M., Fowlkes, D.M., Fornace, A.J., Jr, Malley, J.D. and Kant, J.A. (1985) J. Mol. Biol. 185, 1-19. 27 Kant, J.A., Lord, S.T. and Crabtree, G.R. (1983) Proc. Natl. Acad. Sci. USA 80, 3953-3957. 28 Doolittle, R.F., Watt, K.W.K., Cottrell, B.A., Strong, D.D. and Riley, M. (1979) Nature 280, 464-468.