Effects of Hyaluronan Lyase, Hyaluronidase, and Chondroitin ABC Lyase on Mammalian Vitreous Gel

Abstract

purpose. To determine the effects of enzymes on mammalian vitreous gel and to thus infer the structural roles of hyaluronan and chondroitin sulfate in the gel.

methods. The wet weights of bovine vitreous gels were compared before and after incubation with Streptomyces hyaluronan lyase, chondroitin ABC lyase, testicular hyaluronidase, or buffer alone. The extent of hyaluronan depolymerization was determined by chromatography and that of chondroitin sulfate depolymerization by western blot analysis.

results. After digestion with Streptomyces hyaluronan lyase (30 U/gel), the gel wet weight was the same as that of controls (incubated with buffer alone) despite 94% of the hyaluronan having been depolymerized; when digested with 100 U/gel, the gel wet weight decreased (to 57% of original wet weight versus 86% for controls, P = < 0.001) and hyaluronan was completely depolymerized. Chondroitin ABC lyase digestion (0.2 U/gel) resulted in a slight reduction in gel wet weight (90% versus 96%, P = < 0.001) and depolymerization of 88% of the hyaluronan; the presence of fully digested chondroitin sulfate chains was established. Digestions with 100 and 500 U/gel of testicular hyaluronidase resulted in a decrease (P = < 0.001, both cases) in gel wet weight (53% versus 82%, 100 U/gel; 57% versus 86%, 500 U/gel) with 75% and 97% hyaluronan depolymerization, respectively.

conclusions. Depolymerization of all vitreous hyaluronan and of chondroitin sulfate resulted in gel wet weight reduction but not gel destruction. Digestion with 30 U/gel of Streptomyces hyaluronan lyase revealed a small pool (6%) of relatively enzyme-resistant hyaluronan that specifically contributed toward maintaining gel wet weight.

The contribution made by glycosaminoglycans (GAGs) to the structure of the vitreous gel is poorly understood. The published data are largely based on qualitative assessments of the effect that enzymes that depolymerize GAG chains have on gel structure, and these data have provided conflicting results. ^{1 2 3 4 5 6} Many of the experiments were undertaken to assess the potential therapeutic value ⁷ of enzymes such as testicular hyaluronidase and chondroitin ABC lyase in liquefying the vitreous, ⁵ in promoting the separation of the vitreous gel from the retina, ^{5 8 9} or in aiding the clearance of blood from the vitreous cavity. ¹⁰  

The predominant GAG in mammalian vitreous gel is hyaluronan (HA), ¹¹ which is thought to form an extended meshwork throughout the vitreous gel. ^{12 13} In addition, the gel contains small quantities of sulfated GAGs including chondroitin sulfate. Chondroitin sulfate (CS) chains are attached to core proteins to form proteoglycans, and there are two known CS proteoglycans in vitreous-type IX collagen ¹⁴ and versican. ¹⁵  

During ageing and in various pathologic processes the vitreous gel liquefies and this is associated with the aggregation of collagen fibrils. ^{16 17} It has been proposed that both HA ^{3 13} and CS ¹³ space the vitreous collagen fibrils apart (i.e., they may act to prevent collagen fibril aggregation that in turn leads to vitreous liquefaction and gel collapse). It has also been suggested that CS may be involved in the adhesion between the vitreous and the inner limiting lamina of the retina. ⁹  

The aim of this study was to provide a quantitative analysis of the structural effects on the vitreous gel of removing HA and CS from the gel. Three enzymes were used viz. Streptomyces HA lyase, which specifically catalyzes the depolymerization (fragmentation into oligosaccharides) of HA, and chondroitin ABC lyase and testicular hyaluronidase, which both depolymerize HA and CS. ^{18 19}  

Methods

Enzyme Digestions of Vitreous Gels

The eyes from 2-year-old steers were transported from a local abattoir on ice, and the vitreous gels were carefully removed (within 5 hours) after making a coronal incision through the sclera in the region of the vitreous base. The wet weight of each gel was recorded before placing it in an equal volume of buffer. The buffer for Streptomyces HA lyase and testicular hyaluronidase digestions was 0.1 M sodium acetate, pH 6.0 (which is close to the optimal pH for maximal activity of these enzymes), and for the chondroitin ABC lyase digestions 0.1 M Tris–HCl, pH 7.4. Both of these buffers contained 0.15 M NaCl and protease inhibitors (2 mM phenylmethylsulfonyl fluoride, 2 mM EDTA, 5 mM benzamidine hydrochloride, and 10 mM N-ethylmaleimide). Streptomyces HA lyase (EC 4.2.2.1, Sigma, Poole, UK) digestions were performed using 30 or 100 units of enzyme per vitreous gel. Chondroitin ABC lyase (EC 4.2.2.4, affinity purified, Sigma) digestions were performed using 0.2 units of enzyme per gel. Hyaluronidase (EC 3.2.1.35, type IV-S from bovine testes, Sigma) digestions were performed using 100 or 500 units (USP definition) of enzyme per gel. All enzyme digestions were carried out at 37°C for 48 hours in a closed container. In total, for each experiment, 12 gels were digested with enzyme, and 12 control gels were placed in buffer without enzyme. After 48 hours the wet weights of the enzyme-digested gels and the controls were individually recorded by removing the gels from the buffer solution with a sieve (aperture size 1 mm²) and placing the gel on a balance within 15 to 30 seconds. The percentage of the original wet weight was calculated for each gel. For statistical analysis of these data, paired t-tests were performed comparing the percentage of the original wet weight of the gels placed in buffer alone (controls) with that of the gels placed in buffer containing enzyme. 

Analysis of HA

Bovine vitreous HA, which has a weight-averaged molecular weight of 170,000, elutes virtually entirely in the void volume of a Superose 12 HR (Pharmacia) gel filtration column. ¹⁵ An assay of the extent to which the HA depolymerizes during enzyme digestions of the vitreous gels was devised whereby the proportion of the total HA remaining in the void volume of this column after enzyme digestion was measured. From a set of 12 enzyme digestions the mean concentration of HA in the void volume of three randomly chosen enzyme-digested gels was determined; this was then expressed as a percentage of the mean concentration of HA in the void volume of three randomly chosen control (undigested) gels. 

To measure the HA concentration in the void volume of the Superose 12 HR column a previously described method for measuring vitreous HA concentration was modified. ¹⁵ Vitreous gels (whether enzyme-digested or controls) were boiled for 30 minutes, resulting in liquefaction of the gel apart from a small residual pellet. The pellet was removed and cysteine (20 mM, final concentration) added to the liquid before proteolytic digestion with 10μ g/ml papain at 60°C for 16 hours. Iodoacetic acid (4 mM) was then added to inhibit the action of the papain. Aliquots (1 ml) of this“ vitreous extract” were then applied to a 30 × 1 cm Superose 12 HR column using phosphate-buffered saline (pH 7.4) as the elution buffer. The column effluent was passed through an in-line Wyatt/Optilaboratory 903 interferometric refractometer (Optichem, Clywd, UK) to measure sample concentration. After papain digestion the only material that eluted in the void volume was HA. ¹⁵ This was confirmed for each assay by digesting the sample overnight with 2.5 U/ml of Streptomyces HA lyase immediately before chromatography; there was no detectable material in the void volume after this procedure (Fig. 1) . Hence, the material detected by the refractometer in the void volume of the column was HA, and the amount of HA was calculated after measurement of the integrated refractive index of the voided material (i.e., that which eluted before 11 ml from the column; Fig. 1 ). A previously derived dn/dc value of 0.155 ml/g ^{15 20} was used, and the data were analyzed using Astra for Windows 4.10 software (Optichem). These data were called the void volume HA concentration and were expressed in micrograms of HA per milliliter of vitreous gel. 

Analysis of Chondroitin Sulfate

Two monoclonal antibodies 2B6 and 3B3 (ICN Pharmaceuticals, CA) were used. These antibodies recognize residual oligosaccharide stubs attached to the core proteins of vitreous CS proteoglycans after exhaustive digestion with chondroitin ABC lyase ¹⁵ ; thus, reactivity with 2B6, 3B3 (CS stub antibodies), or both confirms the presence of fully digested CS chains. However, these antibodies do not recognize the oligosaccharide stub that remains after digestion with testicular hyaluronidase. 

Two chondroitin ABC lyase digested gels and two gels digested with 30 U/gel of Streptomyces HA lyase (as negative controls) were boiled and centrifuged. The resultant supernatants were collected, pooled for each enzyme digest, dialyzed against distilled water, and lyophilized. The lyophilized samples were then analyzed by western blot analysis after sodium dodecyl sulfate (SDS)–9% polyacrylamide gel electrophoresis (PAGE) under reducing conditions. The two CS stub antibodies were combined to probe the nitrocellulose membrane, and the membrane was also probed with a type IX collagen polyclonal antiserum (kindly provided by Dr. Shirley Ayad, Manchester, UK). ^{14 15} Secondary antibodies and the ECL detection system were used as previously described. ¹⁵  

Results

The gels that had been placed in buffer (with or without enzyme) for 48 hours lost wet weight, but did not disintegrate and in all cases a single gel structure was retained. By contrast, boiling vitreous gel (e.g., during the HA analysis) resulted in its complete destruction, leaving a liquid together with a residual pellet that had a wet weight of less than 1% of the original gel wet weight (and which was presumed to comprise collagen fibrils). Gels lost a small proportion of their wet weight in the buffer (without enzyme) for 48 hours (Table 1) ; the mean gel wet weight of these control gels decreased by 4% in the pH 7.4 buffer (96 ± 3% of original gel wet weight) and by 17% in the pH 6 buffer (83 ± 10% of original gel wet weight). Cutting gels in half before measuring the changes in wet weight over 48 hours in buffer did not significantly alter the results (i.e., the presence or absence of potentially intact vitreous cortex did not alter the stability of the vitreous gels; data not shown). As well as losing wet weight, the control gels also lost HA into the surrounding buffer; the mean void volume HA concentration was 375 ± 116 μg/ml after 48 hours in the pH 7.4 buffer and 430 ± 123 μg/ml after 48 hours in the pH 6 buffer compared with 497 ± 40 μg/ml for fresh gels (Fig. 1) . 

Vitreous gels digested with 30 U/gel of Streptomyces HA lyase for 48 hours showed no significant difference in gel wet weight compared with control gels (Table 1) . Each of the three vitreous extracts that were analyzed by gel filtration chromatography after digestion with 30 U/gel showed a small residual refractive index peak at the void volume (Fig. 2 A). Quantitative analysis showed that the mean void volume HA concentration was 27 ± 3 μg/ml after the enzyme digestion compared with 430 μg/ml for the control gels (i.e., 6% [27/430 × 100%] of the HA remained in the void volume after digestion with 30 units/gel of Streptomyces HA lyase; Table 1 ). 

When 100 U/gel of Streptomyces HA lyase was used, the mean gel wet weight decreased to 57% compared with 86% for the controls (Table 1) ; this was a statistically significant decrease (P = < 0.001). After digestion with this higher concentration of enzyme, chromatographic analysis of the vitreous extract showed that there was no residual peak of refractive index at the void volume (Fig. 2A) , and when the quantitative assay was performed, no HA was detected in the void volume. The assay was deemed to be sensitive down to a void volume HA concentration of 10 μg/ml (Table 1) . 

Chondroitin ABC lyase digestion resulted in a slight decrease in mean gel wet weight (90%) compared with that in the controls (96%); this decrease was statistically significant (P = < 0.001). The refractive index profile of the vitreous extracts after gel filtration chromatography was in all cases different from that for Streptomyces HA lyase digestion in that there was a more gradual rise in refractive index that persisted into the included part of the elution profile (Fig. 2B) . The mean void volume HA concentration after chondroitin ABC lyase digestions was 46 ± 3 μg/ml (Fig. 2B ; i.e., 12% [46/375 x 100%] of the HA remained in the void volume after chondroitin ABC lyase digestion; Table 1 ). 

SDS–PAGE under reducing conditions and western blot analysis were performed using the CS stub antibodies (2B6 and 3B3 pooled together) on the material released from the chondroitin ABC lyase–digested gels by boiling (Fig. 3 , lane 2) and as a negative control, on material released from Streptomyces HA lyase–digested gels (Fig. 3 , lane 1). The CS stub antibodies revealed two components in the material from the chondroitin ABC lyase–digested gels (Fig. 3 , lane 2); these two components were not present to a significant degree in the material from the Streptomyces HA lyase–digested gels (Fig. 3 , lane 1). The slower-migrating component (high molecular weight component) probably represented versican. ¹⁵ The faster-migrating component was the α2(IX) chain of type IX collagen. This was demonstrated by comparing lane 2 with a western blot of the same material using the polyclonal antiserum that recognizes the α-chains of type IX collagen (lane 3). Of the three distinct α-chains of type IX collagen [i.e., α1(IX), α2(IX), and α3(IX)], only theα 2(IX) chain carries a CS chain; and after chondroitin ABC lyase digestion, this comigrates with the α1(IX) chain on SDS–PAGE under reducing conditions. ^{14 15} The α2(IX) chain that was recognized by the type IX collagen antiserum migrated to the same position as the lower of the two bands recognized by the CS stub antibodies; therefore, the CS stub antibodies recognized the oligosaccharide stub left on the α2(IX) chain after chondroitin ABC lyase digestion. In summary, the western blot data showed that some if not all the CS chains on the CS proteoglycans of vitreous had been completely digested by chondroitin ABC lyase. 

Digestions with 100 and 500 U/gel of testicular hyaluronidase produced a significant reduction in gel wet weight (P = < 0.001); the mean gel wet weights of the enzyme-digested gels were 53% and 57%, respectively, of the original wet weight, whereas those of the controls were 82% and 86%, respectively (Table 1) . The vitreous extracts analyzed by gel filtration chromatography after digestion with 100 U/gel all showed a significant peak in refractive index at the void volume, but this was not present after digestion with 500 U/gel of enzyme (Fig. 2C) . The mean void volume HA concentration after digestion with 100 U/gel was 108 ± 49 μg/ml, whereas that after digestion with 500 U/gel was 14 ± 5 μg/ml (Fig. 2C) . That is to say, after digestion with 100 and 500 U/gel of testicular hyaluronidase, 25% [108/430 × 100%] and 3% [14/430 × 100%], respectively, of the HA remained in the void volume (Table 1) . 

Discussion

The purpose of this study was to determine the effects of enzymes on mammalian vitreous gel and to thus infer the structural roles of HA and CS in the gel. In some in vitro studies it was observed that testicular hyaluronidase digestion does not destroy vitreous gel structure. ^{1 2} Furthermore, Ueno et al. ⁴ showed that the gel structure was not destroyed but rather measured a reduction in gel wet weight. By contrast, Mayne et al. ³ showed that digestion with testicular hyaluronidase, as well as with Streptococcal HA lyase and chondroitin ABC lyase, produced collapse of mammalian vitreous gel. In vivo experiments in which testicular hyaluronidase was injected intravitreally into rabbit eyes demonstrated partial vitreous liquefaction. ^{5 6} These studies, with the exception of Ueno et al., ⁴ did not quantify gel collapse or liquefaction, and none of them measured the extent to which vitreous HA had been depolymerized by the various enzymes. In this study we have directly correlated loss of gel wet weight with a measure of the extent to which the vitreous HA was depolymerized (Table 1) . 

The in vitro experiments described herein show that depolymerization of all the HA in the vitreous gel (i.e., by digestion with 100 U/gel of Streptomyces HA lyase) resulted in a moderate reduction in wet weight (57% versus 86% in controls, Table 1 ) but not complete collapse of the gel structure. Depolymerization of CS and 88% of the HA (i.e., by digestion with chondroitin ABC lyase) did not collapse the gel but rather produced a slight reduction in wet weight when compared with control gels (90% versus 96%). These data challenge the hypothesis of Scott ¹³ that an infinite meshwork of GAGs resulting from HA:HA, CS:CS, and HA:CS interactions is of pivotal importance in stabilizing the vitreous gel. Mayne et al. ³ have proposed a model whereby an infinite network of HA separates the collagen fibrils; although this may be correct, our data suggest that other factors are also involved in preventing the aggregation of collagen fibrils and subsequent gel collapse. 

Digestion with 30 U/gel of Streptomyces HA lyase did not result in any demonstrable loss in wet weight of the vitreous gels (when compared with control gels) despite 94% of the HA being removed from the void volume; at this concentration of enzyme all the HA would have been removed from the void volume had it been in free solution. ¹⁵ These data suggest that a small proportion of the total HA (approximately 6%) is both relatively resistant to this enzyme and has a crucial role in maintaining the wet weight of the vitreous gel. This type of fractionated response to Streptomyces HA lyase has been described previously in the context of studying HA release from the cell surface of virally transformed chondrocytes. ²¹ These investigators showed that at the lower of two concentrations of enzyme, a proportion of the HA was not released, and they ascribed this to the HA being protected from the enzyme by other macromolecules, being bound to macromolecules that inhibited the enzyme, or being physically trapped. In our experiments the first two alternatives were possibilities; the HA was not trapped in our experiments because it was all in free solution (owing to the papain digest) before analysis. Two types of HA network have been observed in mammalian vitreous by rotary shadowing electron microscopy: one composed of fine HA filaments and the other containing bundles of laterally aggregated HA filaments that appeared to be stabilized by globular proteins. ¹² A possible explanation for our observations is that these lateral aggregates of HA are relatively resistant to the enzyme Streptomyces HA lyase and that they specifically contribute toward the spacing apart of the collagen fibrils, thus preventing loss of gel wet weight. 

Testicular hyaluronidase digestion resulted in a loss of gel wet weight, but the mechanism was not fully elucidated. At 100 U/gel concentration this enzyme produced a moderate loss in wet weight when compared with that of controls (53% versus 82%), which is similar to the result obtained by Ueno et al. ⁴ Nevertheless, only 75% of the HA had been depolymerized (Table 1) . By contrast, 30 U/gel of Streptomyces HA lyase produced no loss of gel wet weight, but 94% of the HA was depolymerized. It is unlikely that the moderate loss of gel wet weight produced by testicular hyaluronidase (53% versus 82%) was due to its action on CS because the chondroitin ABC lyase digestions produced only a slight loss in gel wet weight (90% versus 96%). One possibility is that the protease inhibitors used in this study did not fully inhibit nonspecific proteolytic activity associated with the enzyme preparation. Intravitreal injections of testicular hyaluronidase have been advocated for therapeutically inducing vitreous gel liquefaction, ⁵ for promoting separation of the vitreous gel from the retina, ^{5 8} or for aiding the clearance of blood from the vitreous cavity. ¹⁰ It has generally been assumed that the effects of this enzyme are due to depolymerization of HA, but the present study demonstrates that such assumptions are not necessarily valid. 

 Supported by The Wellcome Trust. PNB holds a Wellcome Trust Clinician Scientist Fellowship.

 Submitted for publication November 24, 1998; revised March 9, 1999; accepted April 21, 1999.

 Proprietary interest category: N.

 Corresponding author: Paul Bishop, School of Biological Sciences, Room 2.205, Stopford Building, University of Manchester, Oxford Road, Manchester, M13 9PT, UK. E-mail: [e-mail]

 

Figure 1.

 

Refractive index profiles (exemplar) of vitreous extracts after Superose 12 HR gel filtration chromatography. A vitreous extract from fresh (undigested) vitreous gel (A); the vitreous extract was obtained by boiling the gel, proteolytically digesting with papain, and adding iodoacetic acid before chromatography. The void volume HA concentration was calculated from the integrated refractive index of material that eluted before 11 ml (shaded area); 11 ml was chosen because it was determined from the shape of the elution profile that all the material in the void volume had eluted from the column at this volume. The same vitreous extract as (A) but with subsequent overnight digestion at 37°C with 2.5 U/ml Streptomyces HA lyase before chromatography (B); as a result of this additional digestion the refractive index peak was completely abolished from the void volume, thus demonstrating that only HA contributed to this peak. Therefore, measurement of the integrated refractive index of material in the void volume allowed direct calculation of the concentration of HA. V_o indicates the void volume; this contains macromolecules that are sufficiently large to be excluded from the column beads and consequently elute first from the column.

Table 1.

 

Vitreous Gel Wet Weights with and without Enzyme Digestion and Analysis of the HA Present in the Void Volume of a Superose 12 HR Column after Enzyme Digestion

Enzyme and Concentration	% of Original Gel Wet Weight, *			Void Volume HA after Digestion
		Controls	Digests	Concn ; μg/ml, * , ‡	% Left in Void Volume, §
Streptomyces HA lyase, U/gel
30	83 ± 10, †	84 ± 9, †	27 ± 3	6
100	86 ± 8	57 ± 14	<10	0
Chondroitin ABC lyase, U/gel
0.2	96 ± 3	90 ± 3	46 ± 3	12
Testicular hyaluronidase, U/gel
100	82 ± 11	53 ± 8	108 ± 49	25
500	86 ± 7	57 ± 13	14 ± 5	3

 Conc

ⁿ  , concentration in.

*  Data are expressed as mean ± SD. For each experiment 12 vitreous gels were subjected to enzyme, and 12 controls were placed in buffer alone. Paired t-tests were performed to compare the wet weights of enzyme digested and control gels. In all cases, except where indicated

†  , enzyme digestion resulted in a significant loss in gel wet weight (P < 0.001).

‡  The concentration of HA in the void volume of a Superose 12 HR gel filtration column was measured for enzyme-digested gels. A mean value of (3) assays of randomly chosen (from 12) gels was derived and expressed as the concentration of void volume HA per milliliter of vitreous gel.

§  The HA in the void volume of the column after enzyme digestion is expressed as a percentage of the HA in the void volume of the control gels.

Figure 2.

 

Refractive index profiles (exemplar) of vitreous extracts after Superose 12 HR gel filtration chromatography. Intact vitreous gels were digested (before extraction) with (A) 30 and 100 U of Streptomyces HA lyase, (B) 0.2 U of chondroitin ABC lyase, and (C) 100 and 500 U of testicular hyaluronidase at 37°C for 48 hours. V_o, void volume of the column; u/g, units of enzyme per gel.

Figure 3.

 

SDS–9%PAGE under reducing conditions and western blot analysis of material released by boiling of Streptomyces HA lyase–digested gels (lane 1) and chondroitin ABC lyase–digested gels (lanes 2 and 3). The blots were performed with the combined CS stub antibodies (lanes 1 and 2) and a polyclonal antiserum that detects the α-chains of type IX collagen (lane 3). After reduction, the α1(IX) and chondroitin ABC lyase-digested α2(IX) chains comigrate.

The authors thank Lorraine Schmidt for her technical assistance.