Anal. Chem. 1994,66,1134-1140
Separation of Complex Otigosaccharide Mixtures by Capillary Electrophoresis in the Open-Tubular Format Morgan Stefansson and Mllos Novotny’ Department of Chemistry, Indiana Universi?y,Bloomington, Indiana 4 7405
The fluorescent derivatives of complexoligosaccharidemixtures from different origins are separated in open capillaries. The number of negatively charged sulfonate groups in the tag molecule stronglyaffects separationefficiency and selectivity. Coating of the capillary surface with linear polyacrylamide has been essential to ensure fast and stable migration velocities. Efficiencies in excess of 1 million plates/m have been achieved, facilitating resolution of branched oligosaccharides. The effects of field strength and buffer composition on the apparent electrophoretic mobility of dextran and dextrin oligomersare discussed, with relation to borate complexation. Preliminary examples of applications to monitoring the action of hydrolytic and synthesizing enzymes are also described.
ferent borate concentrations and electric field strengths. The effect of borate complexation was compared for the series of dextrans and dextrins, i.e., oligomers with different sugar linkages. The analytical methodologydevelopedhere appears suitable to resolve certain branched species of oligomeric mixtures, as demonstrated with dextran and laminarin. The action of debranching and synthesizing enzymes on various saccharides can also be followed.
EXPERIMENTAL SE 3TION Chemicals. All polysaccharides and enzymes used in this study were received from Sigma (St. Louis, MO). Sigma also supplied all buffer chemicals, except for boric acid Oligosaccharides originating from various natural sources (Malinkrodt, Inc., Paris, KY). Acrylamide and ammonium often comprise complex mixtures due to their polydispersity persulfate were purchased from Bio-Rad Laboratories (Herand incidence of branched structures. In spite of their cules, CA). Sodium cyanoborohydride and the fluorogenic increasing importance in medical sciences, biotechnology, the reagents, 2-aminopyridine (AP) and 5-aminonaphthalene-2food industry, and other industrial uses, oligosaccharide sulfonate (ANA), were received from Aldrich (Milwaukee, mixtures have been poorly characterized. The structural WI), 8-Aminonaphthalene-1,3,64risulfonate (ANTS) was a product of Molecular Probes, Inc. (Eugene, OR). The complexity, in conjunction with a lack of proper separation endoglucanases, EG I and EG 111, were a gift from Dr. G. and detection methodology, is largely responsible for this situation. Pettersson (Department of Biochemistry, Uppsala University, Uppsala, Sweden). High-performance capillary electrophoresis (HPCE) with Column Preparation. Various lengths of fused silica laser-induced fluorescence detection has considerably imcapillaries (Polymicro Technologies, Phoenix, AZ) of 50 pm proved the prospects for future studies of glycoconjugates.14 i.d. (187 pm 0.d.) were used as separation columns. They While HPCE provides unprecedented component resolution were coated with a layer of linear polyacrylamide according for this class of compounds, fluorescent labeling becomes to a slightly modified version of the Hjert6n method:* First, important to Uvisualize” the otherwise spectroscopically uncharacteristic sugar molecules. Recently, we have s h o ~ n ~ . ~the new capillary was treated with 0.1 M NaOH for 1 h and rinsed with water and methanol. Next, (y-methacryloxyprothat gel-filled capillaries provide attractive separation media for the high resolution of complex mixtures of sugar oligomers. py1)trimethoxysilane (10 pL dissolved in dichloromethane containing 0.02 M acetic acid) was coupled to the silica wall We demonstrate here that, with suitable buffer composition and capillary wall treatment, comparable separations can be during a 60-min treatment performed under nitrogen pressure; the capillary was then rinsed with methanol and water. A 4% achieved in the open-tubular format. (w/w) acrylamide solution, containing 1 pL/mL N,N,N’,N’While rapid and highly efficient separations of oligomeric tetramethylethylenediamine (TEMED) and 1 mg/mL ammixtures of dextrans, amylose, and laminarin are demonmonium persulfate, was then passed through the capillary strated, brief optimization studies have also been performed under nitrogen pressure for 30 min. Finally, the capillary with respect to the buffer composition, migration modifiers, was rinsed with water and dried under a stream of nitrogen. and fluorescent tags. Since the borate complexation is now Sample Preparation and Enzymatic TreatmentProcedures. typically used as the means of providing charge on neutral The sugars were derivatized through the Schiff base formation saccharides,’ migration mechanisms were explored for difbetween the aromatic amine of a reagent and the aldehyde (1) Honda, S.; Makino, A.; Suzuki, S.; Kakehi, K. Anal. Biochem. 1990, 191, form of a sugar, followed by reduction of the Schiff base to 228-236. a stable product. The concentrations of the reagents were (2) Liu, J.; Shirota, 0.; Wieslcr, D.; Novotny, M. Proc. Natl. Acad. Sci. U.S.A. 1991,88, 2302-2306. 20-50 mM in 3% (w/w) acetic acid.9 The mixtures were (3) Liu, J.; Shirota, 0.;Novotny, M. Anal. Chem. 1991, 63, 413417. (4) Chicsa, C.; Horvath, C. J . Chromatogr. 1993, 645, 337-352. (5) Liu. J.; Shirota, 0.;Novotny, M. Anal. Chem. 1992, 64, 973-975. ( 6 ) Liu, J.; Dolnik, V.; Hsieh, Y.-Z.;Novotny, M. A w l . Chem. 1992,64, 13281336.
1134
Analytical Chemktty, Vol. 66, No. 7, April 1, 1994
(7) Novotny, M.; Sudor, J. Electrophoresis 1993, 14, 373-389. (8) HjertCn, S. J. Chromatogr. 1985, 347, 191-198. (9) Jackson, P. Biochem. J . 1990, 270, 705-713. 0003-2700/94/0366-l134$04.5O/0
0 1994 Amerlcan Chemlcal Soclety
subsequently heated for 60 min at 90 OC, in the presence of 0.1 M sodium cyanoborohydride. The samples were stable when stored at -20 OC before analysis. The hydrodynamic technique of sample introduction was employed. Debranching of Dextran by Mycodextranase. A 5.0-mg sample of dextran with an average molecular weight of 18 300 was incubated at pH 4.5 for 3 h at 37 "C, with 25 units of mycodextranase (1,3-1,4-a-~-glucan-4-glucohydrolase; EC 3.2.1-61) from Penicillium funiculosum. The enzymatic activity was stopped by boiling the incubation mixture for 10 min. After derivatization with ANTS, the mixture was analyzed by direct injection into the capillary system. EnzymaticSynthesis of Dextrans. Dextran oligomers and polysaccharides were enzymatically synthesized by dextran sucrase (EC 2.4.1.5 from Leuconostoc mesenteroides, containing dextran primers with an average molecular weight of approximately 60 000-90 000) using 2.5 units of enzymatic activity in 0.2 mL of acetate (50 mM at pH 5.2) and 10% sucrose. The sample was incubated for 2 h at 30 OC after 1.O mM maltoheptaose, derivatized with ANTS (A7), was added as internal standard. A7 also served as a probe for the monitoring of dextrin branching activity. The enzymatic activity was stopped by boiling the mixture for 10 min. Enzymatic Cleavage of Laminarin. Laminarin (0.5 mg amounts) (from Laminaria digitata) was derivatized with ANTS and incubated up to 14 h at pH 5.2 (37 "C)with laminarinase (EC 3.2.1.6 from Penicillium species) or with different cellulase preparations. Aliquots were withdrawn from the mixtures at different time of incubation. The amount and type of enzymes used were as follows: (a) 1 unit of laminarinase, (b) 4 units of a commercial cellulase preparation (EC 3.2.1.4 from Trichoderma uiride), and (c) 32 pg of endoglucanase EG I or EG 111. Both EG preparations were from Trichoderma reesei.1° Apparatus. Our homemade instrumental setup for capillary electrophoresis/laser-inducedfluorescence has been described earlier.' A high-voltage power supply (Spellman High Voltage Electronics, Plainview, NY) capable of delivering 0 4 0 kV was employed. All separations were run in the cathodic mode due to the negatively charged analytes, and on-column fluorescence detection was accomplished with a Model 56X helium+admium laser (Omnichrome, Chino, CA) as the excitation source, operating at 325 nm, while fluorescence was measured at 375 nm for 2-aminopyridine, 475 nm for 5-aminonaphthalene-2-sulfonate, and 5 14 nm for 8-aminonaphthalene- 1,3,6-trisulfonate derivatives.
RESULTS AND DISCUSSION Polysaccharides are polydisperse compounds, often with a wide distribution in molecular weight as well as a substantial variation in their primary structure.12 Their detailed analysis has traditionally been a difficult task because of the lack of component resolution by means of ~hromatography.'~In addition, the absence of spectroscopically distinct moieties in their molecules makes detection difficult. Modern HPCE (10) Flgerstam, L. G.; Pettersson, L. G. FEBS Lett. 1980, 119, 97-100. (1 1) Liu, J.; Hsieh, Y.-2.;Wiesler, D.;Novotny, M. Anal. Chem. 1991.63.408-
412. (1 2 ) Yalpani, M. Polysaccharides: Synthesis, Modifications and Structure/ Property Relations; Elsevier: Amsterdam, 1988. (13) Koizumi, K.; Fukuda, M.; Hizukuri, S. J . Chromatogr. 1991,585, 233-238.
provides the necessary resolution capability3either if charged moieties reside naturally in the polysaccharide molecules or if charge can be imposed on the neutral polysaccharide so that the solutes can migrate in the electric field. A remedy for both the charge and the detection problems is a covalent attachment of a charged and spectroscopically active group through sample derivatizationa2Fluorogenic reagents suitable for the laser-inducedfluorescencedetection appear particularly suitable for the analysis of complex carbohydrate samples.' Electrophoretic mobility can also be induced by a dynamic formation of charged complexes between the sugar units and oxy acids present in the electrolyte, a situation best exemplified by the well-known complexation with boric acid.7 The importance of charge-to-mass ratio is obvious in the analysis of high molecular weight compounds, as the separation of polysaccharides with subtle differences in structure (carbohydrate sequence, types of glycosidic bonds, number and position of various branched chains, etc.) puts extremely high demands on the selectivity and peak capacity of the method used. The analytical utilization of certain debranching enzymes12 may be a useful adjunct in addressing the obvious sample complexity. Choiceof Fluorescent Tag. For fluorometric detection using the 325-nm line of a helium+admium laser as the excitation source, we have evaluated two common aminonaphthalene fluorogenic reagents, 5-aminonaphthalene-2-sulfonateand 8-aminonaphthalene-1,3,6-trisulfonate. 2-Aminopyridyl derivatives' were also included for comparison. AP, ANA, and ANTS derivatives possess the negative charges of zero, one, and three sulfonic groups, respectively. The influence of the three tagging reagents on the apparent electrophoretic mobility and component resolution is exemplified by Figure 1, showing the separation of a dextran sample originating from L. mesenteroides (average molecular weight of 18 300). Interestingly, the effect of charges on a fluorescent tag is considerably more pronounced than we had expected, given the fact that there should be one fluorescent tag per sugar molecule, including large oligosaccharides, which in turn should be multiply charged through borate complexation. However, since dextrans (poly(a-( 1-6)-glucose), with most hydroxy groups present in trans orientation) seem to complex poorly, the effect of sulfonic groups on electrophoretic mobility appears to be strong. Additionally, other buffer components may compete in complexation. At any rate, the average migration times for the individual ANTS oligomers is roughly one-third of those recorded for AP derivatives, with ANA derivatives being intermediate. The polydispersity of the MW 18 300 dextran, shown in Figure lC, is quite distinct. The use of ANTS derivatization clearly exhibits the advantages of this derivative in terms of narrower peaks, greater resolution, shorter analysis time, and peakdetection. The satellite peaks in the vicinity of the major oligomers are most likely due to the branched forms of these oligomers, with one or several 1,4, 1,3, and/or 1,2 glycosidic bonds.I4 After the sample was treated with mycodextranase (known to hydrolyze 1,4 and 1,3 linkages12)), no differences were observed in the recorded electropherograms. ( L ome(14) Castillo, E.; Iturbc, F.;Lopez-Munguia, A.; Pelcnc, V.; Paul, F.; Monsan, P. Ann. N.Y.Acad. Sei. 1993, 672, 425430.
Analytical Chemism, Vol. 66,No. 7, April 1, 1994
1135
A
I
I B
C
V
30
90
I 1
10
PO
m!"
Flgure 1. Influence of the fluorescent tag on the separation of a dextran standard with an average molecular weight of 18 300. The reagents used were (A) 2aminopyrMlne, (8)5amlnonaphthalene-2sulfonate, and (C) 8amlnonaphthalenal,3,6-trlsulfonate. Condltlons: -500 Vlcm (10 MA) uslng 0.1 M borate-trls at pH 8.65 as the electrolyte. The effective length of the separation capillary was 35 cm.
senteroides is often used for commercial production of dextran and is known to form 1,2 glycosidic branches.) Interestingly, it made little difference to the enzymatic reaction when the substrates were fluorescently tagged prior to hydrolysis. Obviously, the tag resides at the reducing end of an oligosaccharide molecule and is remote from most cleavage sites. To follow the enzymatic synthesis of dextran oligomers, the enzymatic preparation of dextran sucrase (containing a small quantity of primers) was incubated in 10% sucrose, and small samples were withdrawn at different time intervals for CE analysis. Separation of a representative reaction mixture, formed after 90-min incubation, is presented in Figure 2. While the amount of formed dextrans increased with time, the ratios of different profile constituents remained relatively constant. The major products were detected as two broad "humps", migrating at 60-1 00 min, well after the small oligosaccharides at the beginning of the electropherogram. The enzymes seem to have favored these large structures as acceptors in the transfer of glucoseunits. Most certainly, they originated from the low-concentration primers (undetectable at the beginning of enzymatic synthesis) added to the enzyme preparation. The broadness of the two humps is due to high dispersity in molecular weight of the polysaccharides formed, demonstrating the limits of the open-tubular methodology described herein. Under similar experimental conditions, additional polysaccharides (various water-soluble modified celluloses, amylopectin, glycogen, etc.) with MW 100 000 and higher, yielded similar unresolved humps. Our efforts are currently 1130 Analytical Chemlstry, Vol. 66,No. 7, April 1, 1994
0
10 min
60
100 mln
Flgure 2. Enzymatic synthesls of dextran oligosaccharides formed after a 90-mln lncubatlonof dextran sucrase with sucrose (10 % w/w). The numbers Indicatedegreeof polymerization. Conditions as In Flgure 1.
being directed toward resolving larger polysaccharides in entangled polymer solutions and variable electric fields.ls Separation Conditions. The importance of using walltreated separation capillaries with minimum electroosmosis, in relation to work reported here, must be stressed. In an uncoated fused silica capillary, the mobility due to electroosmotic flow is usually much higher than the electrophoretic mobility of most sugar conjugates. With an increasing size of oligomeric units, solute migration approaches the electroosmotic velocity asymptotically, diminishing severely the separation system's resolving capacity. Moreover, we find peak efficiencies considerably lower in uncoated ~apillaries.~ A decrease in electrophoretic mobility of the dextran oligomers was observed when 10% (v/v) methanol or tetrahydrofuran was added to the electrolyte, while acetonitrile caused an increase in mobility. However, no changes in the separation selectivity were observed, indicating that the changes in mobilities were probably due to the viscosity alterations. When 10 mM sodium dodecyl sulfate (SDS),a well-known ionic and micelle-forming additive, was incorporated into the buffer, an increase in migration occurred. This may be due to secondary equilibria between SDS and the dextran oligomers. Amylose was another polydisperse, oligomeric mixture that resolved readily in the buffer systems investigated herein. An example of a high-speed (3-min) separation is illustrated with a sample of corn amylose in Figure 3. The amylose sample was further used as a model mixture to study the effects of voltage and buffer composition on the electrophoretic mobility. The apparent mobilities for amylose oligomers at different field-strength values are shown in Figure 4 (a log-log plot). The electrolyte used was 0.1 M borate-tris, at pH 8.65. As ~~
(15) Sudor, J.; Novotny, M . Proc. Natl. Acad. Sci. U.S.A. 1993, 90,9451-9455.
pe 5.oe-4
ANTS
3.oe-4
0
DPlO
8
DP20 DP30
0
3
DP40
1ae-4 100
200
300
500
400
600
700
F i e l d S t r 8 ~(v/cm) Flgure 5. Demonstration of thermal effects. The solld lines are fltted according to a secondorder polynomial equation (see text for explanation).
3 min
0
Flgure 3. High-speed separation of corn amylose. Conditions: -750 V/cm (15 PA) using 0.1 M borate-tris at pH 8.65 as the eiectroiyte. The effective length was 15 cm.
-3.5
l :o
p
0
E
200Vlcm
0
300 V l c m 400 V l c m
I 0
-3.9
1
0
5W V l c m BW V l c m 700 V l c m
.. . . 0.6
0.8
1.0
1.2
1.4
1.6
1.8
log DP Figure 6. Effects of borate concentration on the apparent electrophoretic mobility of amyloses. The fleld strength was -300 V/cm.
_.I
I
0.8
1 .o
1.4
1.8
log DP Flgure 4. log-log plot of the apparent electrophoretic mobility for amyloses at different fieldstrengths versusthe degree of polymerlzatlon (DP). The eiectroiyte was 0.1 M borate-tris, at pH 8.65.
the degree of polymerization increases, the corresponding mobilities exponentially decrease due to a decrease in chargeto-mass ratio. The complexation with borate is not sufficient to keep up with the contribution of a highly charged ANTS moiety. The curves shown in Figure 4 suggest that the relative electrophoretic migration behavior of amyloses is basically size-independent, indicating that all oligomers have similar conformation and complexing properties when borate is used. The electrophoretic mobility is a constant, which should be independent of the applied potential (at least, for relatively small and rigid molecules), unless thermal effects and changes in viscosity result due to insufficient dissipation of the Joule heat. In the experiments carried out in this work, electric current was observed to increase irregularly at a field strength of >300 V/cm, indicating that some thermal effects might be present. It has been shown previous1yl6that the electrophoretic mobility increases with the square root of field strength in such cases. The excellent fits of the second-order polynomial (r2 L 0.99) equations to our data clearly indicate the effect ~
(16) Grushh, E.; McCormick, R. M.; Kirkland, J. J. Anal. Chem. 1989,61,241246.
of temperature on mobilities (Figure 5 ) . Moreover, the relative increase in migration was found to be somewhat dependent on the size of the oligosaccharide molecule, even when the mobilities of amylose oligosaccharides were normalized with respect to the mobility of ANTS. The reason for this behavior is not immediately obvious, but can tentatively be explained by a decreased borate complexation at increased temperatures. Effects of Borate Concentration. The log-log plots between electrophoretic mobility and the degree of polymerization (DP) for amylose oligomers at different tris-borate concentrations (pH 8.65) are graphically presented in Figure 6. The mobility decrease observed with increasing tris-borate concentrations was most likely caused by a combination of secondary equilibrium changes due to the following events: (a) Ion-pairing occurs between the charged form of tris(hydroxymethy1)aminomethane (tris) and the strongly negative ANTS end of the analyte molecules. Presumably, this effect will increase with the local charge density of a solute. In a normalized plot (not shown), where the oligosaccharide mobilities were related to the mobility of ANTS at 0.1 M tris-borate concentration, the relative decrease in mobility was less pronounced for the larger oligosaccharides; this indicates mobilities being caused primarily by borate complexation. (b) There is an extensive complexation between tris and borate, leading to a decrease of the borate’s ”analytical” Analytical Chemise, Vol. 66, No. 7, April 1, 1994
1157
log le
m dennn 0 25 M boraIe4r~ denran 0 25 M bomb-Ins m denrin 0 25 M borate-morph
*
-3.7
8 0
dextran 0 25 M brats-morph dennn 0 t € 6 M TAPS-Ins dextran 0 I €6 M TAPS.ln6
1.35
NOMIGRATION W E X
1.25
-4.0
1.15
0 1 M turate.tris
-4.3
0.9
(I
1.2
1.5
0 25 M bratelm
0 25 M bora1e.mrph
Flgure 8. Normalized mobility data for selected dextrin oligomers.
log DP Flgure 7. Influenceof buffer compositionon the electrophoretic mobiity of dextran and dextrin oligomers. The field strength was -300 V k m .
concentration, and thus a decrease in charge on the solute molecules. At increasing borate concentrations (>0.1 M), and using hydroxide as the base, a gradual decrease in pH is obtained due to formation of polymeric borate complexes with lower acid dissociation constants. However, this was not the case with the tris-borate system, indicative of competing secondary equilibria and, hence, competition for borate complexation. (c) Physical binding of tris, tris-borate, or polymeric borate complexes to the analyte molecules could decrease the solutes’ charge-to-mass ratio. Such an effect is unlikely to be molecular weight dependent. Nevertheless, interaction between amines and polyols has been extensively studied and used, for example, in the separation of sugars and polyols by partition chromatography. (d) Increases in the ionic strength and viscosity of the buffer medium also affect electrophoretic mobility. Once again, such effects should not be molecular weight discriminating. Yet, corresponding thermal effects could decrease borate complexation. The values of electrophoretic mobility for different amyloses plotted against the inverse square root of the activity (calculated according to the Davies equation for the “medium to high” ionic strengths18 were concave (results not shown), indicating that some effects other than those associated with activity were present. Effect of Buffer Composition. In order to study secondary equilibria associated with tris-borate, a second 0.25 M borate system, titrated to the same pH and ionic strength, but containing morpholine as the base (instead of tris), was utilized. Additionally, a borate-free electrolyte with N- [tris(hydroxymethyl)methyl]-3-aminopropanesulfonate (taps) and tris as buffer components was used. Both dextran and dextrin oligomers were employed as the solute ”probes” of a given buffer system. As seen in Figure 7, the following order of decreasing mobilities was experienced by the same oligomers: borate-morpholine > borate-tris > taps-tris. This clearly indicates interaction between tris and other components of the separation medium. Both dextran and dextrin oligomers exhibited similar electrophoretic mobilities in the borate-morpholine system at the lower values of DP, but as the molecular weight
’’
(17) Rabcl, F. M.;Caputo, A. G.;Butts, E.T. J . Chromatogr. 1976,126. 731. (18) Anulyfical Applicotions of Complex Equilibriu; InczBdy, J., Ed; Ellis Horwood Limited, J. Wiley & Sons, Inc: New York, 1967.
I130
Analytical Chemistry, Vol. 66, No. 7, April 1, 1994
increased, lower than expected mobilities were recorded for the dextrans. This behavior was even more pronounced in the taps-tris system, including a change in migration order between the dextran and dextrin series. The nonlinear course of the log-log relationship for dextran in the taps-tris system suggests that the apparent molecular weight of oligomers, as expressed by their mobilities, increases more rapidly than the degree of polymerization. This may indicate size-dependent binding of the tris moiety to the solutes. However, differences in molecular conformation could also affect mobilities: dextran is a random coil, whereas dextrin is known to form a helical structure in the presence of a variety of organic molecules.19 Conformation is known to alter the frictional properties of the solutes migrating in a separation medium. Both dextrins and dextrans migrated slower in the boratetris system, as compared to the borate-morpholine buffer. The curve’s slope was steeper for the larger dextrins, demonstrating that charge formation due to borate complexation was less effective. The steeper slope means higher separation selectivity: however, it comes at the cost of longer analysis time. The slopes for dextrans were about the same for both buffer systems. The mobility data for dextrin oligomers with DP between 10 and 40, from the different systems, were normalized with respect to the mobility of the same sample in the taps-tris system. These numbers were further normalized with respect to the increase in migration of ANTS (in different systems) and presented as normalized migration indexes (NMI) in Figure 8. They are relative measures of change in migration for different oligomers in all systems investigated, showing also a relative change in mobility, compared to ANTS for the same systems. It is the fluorescent tag that has the permanent sulfonate charges, and any value of N M I deviating from unity thus reflects sample-specific changes in migration due to variance in borate complexation, ion pairing, adsorption phenomena, etc. In conclusion, the addition of borate to the electrolytes affected all solutes positively, and NMI increased with DP. Higher concentrations of borate, furthermore, affected the larger oligosaccharides to a greater extent, resulting in a sizedependent “selectivity”. This selectivity seemed to be independent of the base used for pH adjustment. Tentatively, the differences in apparent mobilities for various systems were due to shielding of the negative charge by cations present in ~~
~
(19)Simpson, T.D.;Dintzis, F. R.; Taylor, N. W . Biopolymers 1972, I ! , 25912600.
A
SO
0
Smin
0
min
B
Flgure Q. Separation of laminarin (L. digltata). Condltlons: 50 mM MES and 25 mM trls (pH 5.95) as the electrolyte. The fleld strength was -500 V/cm (18 FA) and the effective length of the separation capillary was 85 cm.
the system (ion pairing). This seems reasonable, as the stability constants for such complexes can differ between different cations. Application to Laminarin, a Highly Branched Polysaccharide. As a complexing agent, borate supplies the means for adjusting selectivity because different carbohydrate moieties, as well as linear or branched structures, will form complexes to a different degree, depending on the magnitude of their stability constants. The chargeimposed on the polysaccharides through borate complexation will be concentration-dependent, and variation in the borate concentration is necessary to optimize the selectivity for each different sample type. For the dextran and dextrin samples shown above, the optimum concentrationswere found to be 0.1 and 0.2 M, respectively. Laminarin is a polysaccharide found in many seeweeds20 and features approximately 30 glucose units linked as /3-1,3 and &1,6. It is currently investigated in biomedical applications due to its beneficial effect on diseases with bacterial,21 viral,22funga1,23and parasitic 0rigin.2~It has also been shown to modify immune suppression and the course of experimental neoplastic disea~e.2~ In the separation of laminarin, borate buffers increased migration of larger oligomers, but decreased resolution of oligomers with the same DP. Evidently, mobility differences due to structural variation within carbohydrate chains can be cancelled upon the additionof borate. Optimum separation conditions were achieved with an electrolyte containing 50 mM 2-(N-morpholino)ethanesulfonic acid (MES) and 25 mM tris at pH 5.95 as buffering agents (Figure 9). Structural characterization of polysaccharides is very important in correlating their composition with physicochemical properties and biological activity. Degree of branching, branching sites, and polydispersity are among the most important structural aspects of glycoconjugates. The enzymes that cleave polysaccharides at specific sites are far more useful tools in structural characterizationthan the methodsemploying acids and bases for polymer degradation. Comparing HPCE ~~
~
~~
~
~
~~
(20) Elyakova, L. A,; Zvyagintscva, T. N . Carbohydr. Res. 1974, 34, 241-248. (21) Di Luzio, N. R.; Williams, D. L. Infect. Immun. 1978, 20, 8W810. (22) Williams, D. L.; Di Luzio, N . R. Science 1980, 208, 61-69. (23) Williams, D. L.; Cook, J. A.; Hoffmann, E. 0.; Di Luzio, N. R. J . Reliculoendothel. Soc. 1978, 23, 4 1 9 4 9 0 . (24) Cook, J. A.; Holbrook. T. W.; Parker, B. W. J. Reticuloendorhel. Soc. 1980, 27, 561-572.
(25) Williams, D. L.; Shewood, E. R.; McNamce, R. B.; Jones, E. L.;Di Luzio, N . R.Hepatology 1985, 5, 198-206.
0
D
6
16 min
Figurela. OugOsacchaddemapsof laminarinafter enzymatic cleavage wtth(A)lamlnarlnase,(B)cellulase,(C)EGI,and(D)EG 111. Conditions In (A) and (B): 0.1 M borate-trls (pH 8.65), -500 Vlcm (10 PA), and 35 cm effective length. Conditions In (C) and (D): 50 mM MES and 25 mM Ms (pH 5.95), -500 V/cm (18 FA) and 85 cm effective length.
oligosaccharidemaps before and after a controlledenzymatic fragmentation of a polysaccharidecan yield useful structural Ana&tical Chemistry, Vol. 66, No. 7, April 1, 1994
1198
information. Such oligosaccharide maps are displayed in Figure 10 for laminarin which had been incubated with different 0- 1,3-glucose-hydrolyzingenzymes. In general, the cellulase and laminarinase preparations seemed to cause more complete hydrolyses than the endoglucanases used. EG I and EG I11 also appear to differ somewhat in their structural preferences. Hence, this application demonstrates the importance of highly selective and sensitive separation systems in the characterization of structural preferences for different enzymes used to hydrolyze polysaccharides and in the measurement of their specific enzymatic activities. This includes formation of oligosaccharides with a similar number of sugar units but differing in degree of branching and type of monosaccharide units in the primary sequence. Furthermore, some of these reaction products can be present at very (26) Jorgcnson, J. W.; Lukacs, K. D. HRC CC, J . High Resolut. Chromatogr. Chromatogr. Commun. 1981, 4,230-235. (27) Cobb, K.; Novotny, M. Anal. Chem. 1989, 61, 2226-2231. (28) Niclsen, R. G.; Richard, E. C. J . Chromatogr. 1990,516, 99-114. (29) Cobb, K.; Novotny, M. Anal. Chem. 1992, 64, 879-886.
1140 Analytical Chemistry, Vol. 66, No. 7, April 1. 1994
low concentrations. Although weinclude here just preliminary results, they suffice to indicate a potential of using a microscale enzymatic treatment in conjunction with HPCE high-resolution analysis of glycoconjugates, much the same as it has been successfullydemonstrated with the peptide mapping proc e d u r e in ~ ~protein ~ ~ ~studies.
ACKNOWLEDGMENT This work was supported by Grant 24349 from theNational Institute of General Medical Sciences, U S . Department of Health and Human Services, and a grant-in-aid from Astra/ Hassle (Mblndal, Sweden). M.S. has been a recipient of fellowships from The Sweden-American Foundation, the foundation “Stiftelsen Blanceflour Boncampagni-Ludovisi”, fbdd Bildt, and The Swedish Academy of Pharmaceutical Sciences. Received for review September 28, 1993. Accepted January 6, 1994.” *Abstract published in Advance ACS Abstracts, February 15, 1994.
