Anal. Chem. 1998, 70, 568-573
High-Resolution Studies of Hyaluronic Acid Mixtures through Capillary Gel Electrophoresis Mingfang Hong, Jan Sudor, Morgan Stefansson,† and Milos V. Novotny*
Department of Chemistry, Indiana University, Bloomington, Indiana 47405
Hyaluronic acid is a negatively charged polysaccharide with a high degree of polydispersity that makes the separation of its oligomers extremely difficult. Through the use of columns filled with a highly viscous polyacrylamide matrix, the unit resolution of hyaluronate oligomers could be achieved, up to at least 80 kDa of mass, through capillary electrophoresis. As analytical application examples, the fractions of enzymatically or ultrasonically degraded hyaluronates were monitored through this method. Because of the very high resolving power, peaks additional to the regular oligomers can be observed that are assumed to be conformers of this regular, unbranched biopolymer. Hyaluronic acid (HA) is a linear polysaccharide, featuring a repeating, negatively charged disaccharide structure of D-glucopyranoside uronic acid bound through a β-1,4-linkage to 2-N-acetyl2-deoxy-D-glucopyranose:
HO
CH2OH O HNAc
HO O
O HO2C
OH O
Depending on its source and biological function, the average molecular mass of this polydisperse polymer may vary, but HA entities reaching up to several million daltons are not uncommon.1 Various important physiological functions have been attributed to HA. Although this biopolymer is not a “true proteoglycan” itself, HA performs very specific functions in assisting proteoglycan aggregation in numerous connective tissues. Beyond its function in soft connective tissues, HA has important roles in cellular regulation,2-5 and, possibly, cell protection.6 The specificity of binding certain proteins to HA appears suggestive of some advanced structural details of this molecule, in various solutions, beyond its monotonous primary structure. The evidence for its solution secondary structure comes primarily from circular dichroism studies7,8 and NMR spectrometry.9,10
Various properties of the HA preparations isolated from biological materials are increasingly finding industrial and medical applications11 in ophthalmology, clinical diagnosis, drug delivery, and cosmetic preparations. Yet, the mixtures of HA oligomers remain ill-defined conglomerates from the analytical point of view, and additionally, relatively little is known about HA’s propensity to form advanced structures. Here, we report that capillary gel electrophoresis (CGE), with its superb resolution, is a very effective tool for analyzing various biologically and technologically important fractions of HA. The resolving power of CGE for the individual oligomers is first demonstrated with a sample of intact HA and, subsequently, in monitoring various fractions of the enzymatically cleaved and ultrasonically degraded HA preparations. The average molecular mass and polydispersity of these preparations were first checked through the combination of size-exclusion chromatography with low-angle light-scattering detection. Subsequently, the fluorescently labeled oligomers were separated by CGE using laserinduced fluorescence (LIF) detection. To ensure a high degree of resolution with such complex mixtures, i.e., the minimum column overload,12 highly sensitive detection techniques (such as LIF) are clearly mandatory. Minor peaks that are clearly resolved in addition to the main oligomers, under different experimental conditions, are likely due to different conformational forms, suggesting a potential value for CGE to study the advanced structural situations. EXPERIMENTAL SECTION Apparatus. All electrophoretic measurements were performed with a homemade capillary electrophoresis/LIF system which had been described previously.13 A high-voltage dc power supply (Spellman High Voltage Electronics Corp., Plainview, NY), capable of delivering 0-40 kV, was used. The on-line fluorescence measurements employed a Model 543-AP argon ion laser (Om-
† Current address: Department of Analytical Chemistry, Box 531 Uppsala University, S-75121 Uppsala, Sweden. (1) Laurent, T. C. In Chemistry and Molecular Biology of the Intercellular Matrix; Balasz, E. A., Ed., Academic Press: New York, 1970; pp 703-732. (2) Alho, A. M.; Underhill, C. B. J. Cell Biol. 1989, 108, 1557-1565. (3) Laurent, T. C.; Fraser, J. R. FASEB J. 1992, 6, 2397-2404. (4) West, D. C.; Hampson, I. N.; Arnold, F.; Kumar, S. Science 1985, 228, 13241326. (5) Heldin, P.; Laurent, T. C.; Heldin, C. H. Biochem. J. 1989, 258, 919-922. (6) Underhill, C. B.; Toole, B. P. J. Cell. Physiol. 1982, 110, 123-128.
(7) Buffington, L. A.; Pysh, E. S.; Chakrabarti, B.; Balasz, E. A. J. Am. Chem. Soc. 1977, 99, 1730-1734. (8) Cowman, M. K.; Balasz, E. A.; Bergmann, C. W.; Meyer, K. Biochemistry 1981, 20, 1379-1385. (9) Holmbeck, S. M. A.; Petillo, P. A.; Lerner, L. E. Biochemistry 1994, 33, 14246-14255. (10) Toffanin, R.; Kvam, B. J.; Flaibani, A.; Atzori, M.; Biviano, F.; Paoletti, S. Carbohydr. Res. 1993, 245, 113-128. (11) Drobnik, J. Adv. Drug Deliv. Rev. 1991, 7, 295-308. (12) Jorgenson, J. W. In New Directions in Electrophoretic Methods; Jorgenson, J. W., Phillips, M., Eds.; ACS Symposium Series 335; American Chemical Society: Washington, 1987; pp 182-198. (13) Liu, J.; Hsieh, Y.-Z.; Wiesler, D.; Novotny, M. V. Anal. Chem. 1991, 63, 408-412.
568 Analytical Chemistry, Vol. 70, No. 3, February 1, 1998
S0003-2700(97)00768-3 CCC: $15.00
© 1998 American Chemical Society Published on Web 02/01/1998
nichrome, Chino, CA) as the light source (5-mW power at 488 nm). The incident laser beam was aligned to its optimum position on the capillary flow cell, at a right angle, by adjusting the positioner holding capillary. Fluorescence emission at 514 nm was collected through a microscopic lens. The emitted light was measured through a PMT (Hamamatsu Photonics K. K., Shizuoka Prefecture, Japan), while the signal was further amplified by a lock-in amplifier (in phase with a mechanical chopper). For size-exclusion chromatography (SEC), an LC-5A highpressure pump (Shimadzu Co., Kyoto, Japan) was used to deliver the mobile phase. Before the mobile phase entered the columns, it was passed through a Shodex degasser and a pulse dampener (both from Showa Denko K. K.) and two specialty filters (Anodisc 25, 0.1 µm from Alltech Associates (Deerfield, IL 60015) and Anodisc 25, 0.02 µm from Whatman International (Maidstone, England)). The three columns connected in series were Bio-Sil SEC-125, SEC-250, and SEC-400-5 from Bio-Rad Laboratories (Hercules, CA). The two serial detectors employed here for the molecular mass determinations were a differential refractometer (model 410 from Waters, Framingham, MA) and a Minidawn lowangle light-scattering detector (Wyatt Technology Co., Santa Barbara, CA). The overall system and its use in molecular mass determinations have been detailed elsewhere.14 Sonication of a hyaluronic acid sample was carried out with a Branson 1200 ultrasonic cleaner (150 W, Branson Ultrasonics Co., Danbury, CT). Materials and Chemicals. Various lengths of fused-silica capillaries (Polymicro Technologies, Phoenix, AZ) of 50-µm i.d. (352-µm o.d.) were used as the separation columns in all CGE experiments. Two hyaluronic acid samples (sodium salts), one from bovine trachea and the other from Streptococcus zooepidemicus, the disaccharide standard, two enzymes, hyaluronidase (EC 3.2.1.35) and hyaluronate lyase (EC 4.2.2.1), TEMED buffer, ammonium persulfate, sodium cyanoborohydride, and [γ-(methacryloxy)propyl]trimethoxysilane were all received from Sigma Chemical Co. (St. Louis, MO), while acrylamide and Tris buffer were the products of Bio-Rad Laboratories. Polyacrylamide (MW 700 000-1 000 000, 10% aqueous solution) was from Polysciences, Inc. (Warrington, PA), phosphoric acid from Fisher Scientific (Fairlawn, NJ), and sodium phosphate (monobasic) from Malinkrodt, Inc. (Paris, KY). Aminodextran (MW 10 000) was a product of Molecular Probes, Inc. (Eugene, OR). Finally, 1-aminopyrene3,6,8-trisulfonic acid, trisodium salt (APTS) was received from Lambda Fluoreszenztechnologie (Graz, Austria). Capillary Coating and Sample Derivatization. The inner surface of separation capillaries was coated with a linear polyacrylamide according to a slightly modified Hjerte´n’s method.15 Specifically, a new capillary was treated with 1 M NaOH for 1 h and rinsed with water and methanol for 15 min. γ-(Methacryloxy)propyl]trimethoxysilane (30 µL dissolved in 1 mL of CH2Cl2 containing 0.02 M acetic acid) was then reacted with the silica wall for 1 h (under nitrogen pressure). The capillary was then rinsed with methanol and water, followed with 4% (w/w) acrylamide aqueous solution, containing 2 µL/mL N,N,N′,N′-tetramethylethylenediamine (TEMED) and 2 mg/mL ammonium persulfate, which was then passed through the capillary under nitrogen pressure for an additional hour. Finally, the capillary was rinsed (14) Chmelik, J.; Chmelikova, J.; Novotny, M. V. J. Chromatogr., in press. (15) Hjerte´n, S. J. Chromatogr. 1985, 347, 191-198.
with water and dried under a stream of nitrogen. The sample derivatization16 of HA was accomplished through mixing a 2-mg sample with 30 µL of 20 mM APTS aqueous solution (containing 3% (v/v) acetic acid) in a glass vial. Subsequently, 1 µL of 1 M NaCNBH3 aqueous solution was added to the above mixture, and the glass vial was heated to 90 °C in a metal block for 1 h. After cooling, the solution was diluted with distilled water to 500 µL and stored at -20 °C. Enzymatic and Ultrasonic Degradation of HA. Hyaluronic acid (4 mg) was derivatized with APTS, as described above. Subsequently, the sample was preincubated for 30 min at 37 °C. After the addition of hyaluronidase (1.4 mg, 290 units), the total sample volume was adjusted to 500 µL with distilled water. The aliquots of 30-µL volumes of the degraded sample were taken out at kinetically appropriate time intervals. They were first subjected to 100 °C in a water bath for 10 min, to deactivate the enzyme. After the mixture cooled to room temperature, it was centrifuged at 10 000 rpm for 10 min. The supernatant was injected into SEC or CGE columns for the analysis. Hyaluronic acid (3.4 mg) derivatized with APTS was diluted with distilled water to the volume of 340 µL before being placed in the ultrasonic generator. At kinetically appropriate time intervals, aliquots of 30 µL were taken and introduced into the SEC or CGE units for analysis. To avoid possible degradation of HA samples at room temperature, the sample aliquots were analyzed as soon as possible. Fresh samples were prepared daily. RESULTS AND DISCUSSION During recent years, capillary zone electrophoresis of glycoconjugates has been demonstrated with different analytical systems, becoming the subject of several reviews.17-21 The method’s high resolving power is particularly attractive in dealing with complex carbohydrate mixtures, even though the oftenneeded fluorescence labeling for LIF detection represents a certain procedural complexity. We have previously shown22 that the choice of a fluorescent label at the oligosaccharides’ reducing end is strategically important in extending the molecular mass range of migrating oligosaccharides as well as in maximizing their resolution. For the analysis of oligosaccharides in the open tubular format, highly charged reagents such as aminonaphthalene-1,3,6-trisulfonate (ANTS)22,23 or APTS16 appear particularly appropriate. As shown with the model dextran mixtures,14,22 there is an upper limit at which the high-mass oligomers can be effectively resolved and detected. In order to extend the CE-based methodologies to larger polysaccharides, a practical and common approach is to use polymeric sieving media, inducing size-dependent migration of (16) Chen, F. T. A.; Evangelista, R. A. Anal. Biochem. 1995, 230, 273-280. (17) Novotny, M. V.; Sudor, J. Electrophoresis 1993, 14, 373-389. (18) El Rassi, Z.; Nashabeh, W. High-Performance Capillary Electrophoresis of Carbohydrates and Glycoconjugates. In Carbohydrate Analysis: High Performance Liquid Chromatography and Capillary Electrophoresis; El Rassi, Z., Ed.; Elsevier Science: Amsterdam, 1994; pp 447-514. (19) Novotny, M. V. Methods Enzymol. 1996, 271, 319-349. (20) Novotny, M. Capillary Electrophoresis of Carbohydrates. In High-Performance Capillary Electrophoresis; Khaledi, M. G., Ed.; Wiley: New York, in press. (21) Chiesa, C.; O’Neill, R. A.; Horvath, C.; Oefner, P. J. In Capillary Electrophoresis in Analytical Biotechnology; Righetti, P. G., Ed.; CRC Press: Boca Raton, FL, 1996; pp 270-430. (22) Stefansson, M.; Novotny, M. V. Anal. Chem. 1994, 66, 1134-1140. (23) Chiesa, C.; Horvath, C. J. Chromatogr. 1993, 645, 337-352..
Analytical Chemistry, Vol. 70, No. 3, February 1, 1998
569
Figure 1. Electropherogram of HA in an entangled polymer solution. Conditions: -403 V/cm (10 µA) using 25 mM citric acid and 12.5 mM Tris buffer as the electrolyte (pH 3.0); 5% LPAA. The effective length of the separation capillary was 50 cm. The inset in the upper corner corresponds to a detail of the electropherogram. The numbers indicate the degree of polymerization.
the derivatized polysaccharides. The sieving media appear particularly beneficial to the separation of large, uniformly charged (“free-draining”) polyelectrolytes (such as a fully ionized HA) because such polymers tend to migrate with size-independent velocities in free solutions.24 Using highly concentrated (cross-linked) gels in CE, we were able to separate relatively small HA oligomers from an enzymatic digest at a previous occasion.25 However, because of the limited reproducibility and fragility of these “chemical gels”, the separation capillaries had a limited analytical value. The present investigation involves the use of entangled polyacrylamide matrixes under optimized conditions. A substantial improvement over the use of cross-linked gels,25 a pullulan-type polymer,26 and free-solution conditions27,28 has, consequently, been achieved. (24) Manning, G. S. J. Phys. Chem. 1981, 85, 1506-1515. (25) Liu, J.; Dolnik, V.; Hsieh, Y.-Z.; Novotny, M. Anal. Chem. 1992, 64, 13281336.
570
Analytical Chemistry, Vol. 70, No. 3, February 1, 1998
The extraordinarily high resolving power of the analytical system is demonstrated with a sample of intact HA in Figure 1, in which each major peak represents an increment of one disaccharide unit. (A plausible explanation for the minor (secondary) peaks will be given below.) Counting from the HA disaccharide (the only available standard), we could visualize electromigration of up to 190 disaccharide units. Since the molecular mass of an HA disaccharide is 401.3, the migration pattern ends at approximately 80 kDa of mass. Admittedly, this is short of the molecular mass of 258.6 kDa that we had determined for this intact HA sample through light-scattering measurements (see below), however, far in the excess of any previous attempt by this (26) Hayase, S.; Oda, Y.; Honda, S.; Kakehi, K. J. Chromatogr., A 1997, 768, 295-305. (27) Grimshaw, J.; Kane, A.; Trocha-Grimshaw, J.; Douglas, A.; Chakravarthy, V.; Archer, D. Electrophoresis 1994, 15, 936-940. (28) Park, Y.; Cho, S.; Linhardt, R. J. Biochim. Biophys. Acta 1997, 1337, 216226.
Figure 2. SEC profiles of the enzymatic degradation of HA. Mobile phase: 0.05 M NaH2PO4, 0.05 M Na2HPO4, 0.15 M NaCl, and 0.01 M NaN3, pH 6.8; flow rate 1.0 mL/min. Key: (1) intact HA; (2-5) HA hydrolysates after enzymatic degradation for 1-, 2-, 5-, and 24-h periods, respectively.
laboratory or others. There is no intention, through Figure 1, to advocate the use of CGE for analyzing the intact biopolymer, but merely to show the current scope and limitations. We have recently demonstrated14 with the model dextrans that the CE analytical results obtained from samples with a broad molecular mass distribution should be interpreted with caution, and the same applies to the present case with HA. When working with the dextrans, we observed that the CE/LIF profiles were biased in comparison to light-scattering data. Whether the lack of response with the higher polysaccharides on CE/LIF is due to spectral or kinetic reasons remains unclear at present. For both the bioanalytical and physicochemical studies of HA, it appears sensible to degrade the biopolymer into smaller fractions. The enzymatic procedures for this purpose may involve either the commonly used hyaluronidases (hydrolases) of animal origin or eliminases from bacterial sources.29 Another general method for degrading large biopolymers involves ultrasonication.30 Both types of degradation were employed in this study. In order to prepare a fraction with desirable oligomeric composition through the enzymatic procedure, the reaction is terminated at a suitable kinetic time point and a fraction is recovered after enzyme denaturation. The extent of enzymatic degradation was first assessed here through SEC with lightscattering detection, and the resulting molecular mass distribution plot is given in Figure 2. The CGE profiles corresponding to the recovered fractions are shown in Figure 3. In order to slow down electromigration of the smaller HA fragments generated upon the action of hyaluronidase, we used, in this example, a 10 000 MW aminodextran (a complexation agent) as an effective migration moderator. (The mechanism of its action on the electromigration of negatively charged polysaccharides has been discussed in a (29) Linhardt, R. J. In Current Protocols in Molecular Biology: Analysis of Glycoconjugates; Varki, A., Ed.; Wiley Interscience: Boston, 1994; pp 17.13.17-17.13.32. (30) Chabrecek, P.; Soltes, L.; Orvisky, E. J. Appl. Polym. Sci. 1991, 48, 233241.
Figure 3. CE profiles for the enzymatic degradation of HA. Conditions: -416 V/cm (12 µA) using 25 mM citric acid and 12.5 mM Tris buffer as the electrolyte, pH 3.0, 4 M urea, 0.03% aminodextran as an additive, and 3% LPAA. The effective separation capillary length was 45 cm. electropherogram 1 intact HA; electropherograms 2-5, HA fragments after enzymatic degradation for 1-, 2-, 5-, and 24-h periods, respectively.
different report.31) While this step may have impaired the component resolution (compared to Figure 1), the main purpose of this application was to demonstrate a complete profile after action of the enzyme. Ultrasonic treatment has been shown previously30 to degrade large HA entities into smaller fragments. When evaluating the effect of sonication on a distribution of HA oligomers (Figure 4), we find a small, but easily measurable shift to lower molecular masses and a decrease in polydispersity. These results are in general agreement with the data by others30,32 who utilized a larger HA preparation than shown here. In one attempt,30 the HA (31) Stefansson, M.; Sudor, J.; Hong, M.; Chmelikova, J.; Chmelik, J.; Novotny, M. V. Anal. Chem. 1997, 69, 3846-3850. (32) Chabrecek, P.; Soltes, L.; Ka´llay, Z.; Novak, I. Chromatographia 1990, 30, 201-204.
Analytical Chemistry, Vol. 70, No. 3, February 1, 1998
571
Figure 4. SEC profiles for the ultrasonic degradation of HA. The separation conditions were the same as in Figure 2: (1) intact HA; (2, 3) HA fragments after ultrasonic degradation of 3- and 9-h periods, respectively.
Figure 5. CE profiles for the ultrasonic degradation of HA. The same conditions as in Figure 1, except the effective length of the capillary was 45 cm, using -416 V/cm: electropherogram 1, intact HA; electropherograms 2 and 3, HA fragments after ultrasonic treatment for 3- and 9-h periods, respectively.
macromolecules were ultrasonically degradated to approximately 200 kDa, which is roughly the average molecular mass of our original sample. A further degradation is obviously inefficient, although the relatively small differences are quite easily measurable through the light-scattering detection method (Figure 4). A subsequent monitoring of the ultrasonically treated fractions through CGE indicates a qualitatively similar trend in the molecular mass shift (Figure 5). However, more importantly, a detailed inspection of the satellite peaks under high resolution (Figure 6) reveals an unexpected observation: due to the ultrasonic treatment, yet another set of peaks has become increasingly visible. With HA being widely known as a monotonous polymer, we tentatively attribute this peak-splitting phenomenon to the formation of different HA conformers, in this case, due to physical agitation or, possibly, the temperature effects of ultrasonication, 572 Analytical Chemistry, Vol. 70, No. 3, February 1, 1998
Figure 6. Details of Figure 5. The numbers 20 and 40 indicate the degree of polymerization. (1-3) correspond to the same description as in Figure 5.
although we do not have any evidence to rule out the breakages at different sites of a chain, leading to fragments with different charges (e.g., ABABA vs BABAB). The origin of the satellite peaks, under different experimental conditions, was further investigated. HA is generally known to be a linear polymer, unlike galacturonic acid, which features certain branched structures.33 A regular appearance of the satellite peaks makes possible impurities in the HA preparations an unlikely source of this peak splitting as well. In addition, the regular pattern was also observed with the HA preparations derived from two different sources, mammalian and bacterial (results not shown). To implicate or rule out an advanced structure as the reason for peak-splitting behavior, we evaluated the high-resolution electropherograms under different pH conditions and compared their appearance in the presence and absence of 8 M urea. HA has one carboxylic group per disaccharide unit. The corresponding pKa value is about 3.5, so that at very low pH, the carboxy groups should lose their charge and the polymer chains could assume a compact conformation as the dominant form. At higher pH values, the electrostatic repulsion forces (due to the negatively charged groups) might force the HA chains increasingly into less compact forms. The observations of electropherograms, under different pH conditions, verified, at least partially, our assumptions. At very low pH (
