microchip immobilized enzyme reactors for hydrolysis of methyl cellulose

Microchip Immobilized Enzyme Reactors forHydrolysis of Methyl Cellulose

Claes Melander,† Dane Momcilovic,‡ Carina Nilsson,† Martin Bengtsson,§ Herje Schagerlo1 f,|

Folke Tjerneld,| Thomas Laurell,§ Curt T. Reimann,† and Lo Gorton*,†

Departments of Analytical Chemistry, Technical Analytical Chemistry, and Biochemistry, Lund University, P.O. Box 124,SE-221 00 Lund, Sweden, and Department of Electrical Measurements, LTH, P.O. Box 118, Lund University, Sweden

Microchip immobilized enzyme reactors (µIMERs) withimmobilized endoglucanases were applied for the hy-drolysis of methyl cellulose (MC). MCs of various molec-ular weights were hydrolyzed using two µIMERs contain-ing immobilized celloendoglucanase Cel 5A from Bacillusagaradhaerens (BaCel 5A) connected in series. Hy-drolysis by the µIMER could be confirmed from theaverage molar masses and molar mass distributionsmeasured by size exclusion chromatography (SEC) withonline multiangle light scattering and refractive indexdetection. Methylated cellooligosaccharides with degreesof polymerization (DP) between 1 and 6 formed duringhydrolysis were analyzed by direct infusion electrosprayionization ion-trap mass spectrometry (ESI-ITMS). Massspectra of µIMER- and batch-hydrolyzed samples werecompared and no significant differences were found,indicating that µIMER hydrolysis was as efficient asconventional batch hydrolysis. A fast and automatedhydrolysis with online MS detection was achieved byconnecting the µIMER to high-performance liquid chro-matography and ESI-ITMS. This online separation re-duced the relative intensities of interfering signals andincreased the signal-to-noise ratios in MS. The µIMERhydrolysates were also subjected to SEC interfaced withmatrix-assisted laser desorption/ionization time-of-flightmass spectrometry. With this technique, oligomers withDP 3-30 could be detected. The hydrolysis by the µIMERwas performed within 60 min, i.e. significantly fastercompared with batch hydrolysis usually performed for atleast 24 h. The µIMER also allowed hydrolysis after 10days of continuous use. The method presented in thiswork offers new approaches for the analysis of derivatizedcellulose and provides the possibility of convenient online,fast, and more versatile analysis compared with thetraditional batch method.

Cellulose is a naturally occurring polymer, which is used asraw material in a number of industrial applications, e.g., in thepaper, paint, textile, food, and pharmaceutical industries.1,2 It is a

linear polysaccharide composed of anhydroglucose units, linkedtogether with 1-4 â-D-glycosidic bonds. To fulfill the variousdemands of functionality from industry, the cellulose is oftenmodified by physical, chemical, or enzymatic means. Chemicalmodification implies the substitution of one or more of the freehydroxyl groups of the glucose monomer on carbon 2, 3, or 6.Enzymatic modification is sometimes performed to change theproperties of the cellulose.3,4 The physical and chemical propertiesof the modified cellulose depend on several factors, such as themodification reaction, the type of substituent, the average molarmass (Mw), the molar mass distribution, the degree of substitution(DS), and the distribution of the substituents along the polymerchain. Since the properties of the modified cellulose are correlatedwith their technological usefulness, techniques to accuratelycharacterize them are of great importance.

One approach to estimate the substituent distribution of acellulose polymer is to hydrolyze the polymer using eithercellulose-hydrolyzing enzymes such as endoglucanases or acidsto obtain oligomers that are possible to analyze with variouschromatographic and mass spectrometric techniques.5-7 Endo-glucanases are enzymes capable of hydrolyzing nonterminalglycosidic bonds (i.e., interior parts) of the cellulose chain.8-11

The ability of an endoglucanase to hydrolyze cellulose will dependon the structure of the active site of the endoglucanase, the typeof substituent, and its distribution along the cellulose chain.Different endoglucanases will be more or less hindered bysubstituents depending on the nature of their active site. Theactivity of the endoglucanases on different substrates and thesensitivity of enzyme activity to substituent position and distribu-tion are to some extent unknown, and these enzyme character-istics are currently under extensive investigation.8,12,13 However,

* Corresponding author. E-mail: [email protected]. Tel.: +46 462227582. Fax: +46 46 2224544.

† Department of Analytical Chemistry.‡ Department of Technical Analytical Chemistry.§ Department of Electrical Measurements.| Department of Biochemistry.

(1) Batdorf, J. B.; Desmarais, A. J. Chem. Addit., Symp. 1971, 136-141.(2) Batdorf, J. B.; Rossman, J. M. Industrial Gums, 2nd ed.; 1973; pp 695-729.(3) Miettinen-Oinonen, A. VTT Publ. 2004, 550, 1-96.(4) Miettinen-Oinonen, A.; Paloheimo, M.; Lantto, R.; Suominen, P. J. Biotechnol.

2005, 116, 305-317.(5) Richardson, S.; Gorton, L. Anal. Chim. Acta 2003, 497, 27-65.(6) Cohen, A.; Schagerlof, H.; Nilsson, C.; Melander, C.; Tjerneld, F.; Gorton,

L. J. Chromatogr., A 2004, 1029, 87-95.(7) Mischnick, P.; Heinrich, J.; Gohdes, M. Papier 1999, 53, 739-743.(8) Karlsson, J.; Momcilovic, D.; Wittgren, B.; Schulein, M.; Tjerneld, F.;

Brinkmalm, G. Biopolymers 2002, 63, 32-40.(9) Bayer, E. A.; Chanzy, H.; Lamed, R.; Shoham, Y. Curr. Opin. Struct. Biol.

1998, 8, 548-557.(10) Tomme, P.; Warren, R. A.; Gilkes, N. R. Adv. Microb. Physiol. 1995, 37,

1-81.(11) Beguin, P.; Aubert, J. P. FEMS Microbiol. Rev. 1994, 13, 25-58.

Anal. Chem. 2005, 77, 3284-3291

3284 Analytical Chemistry, Vol. 77, No. 10, May 15, 2005 10.1021/ac050201r CCC: $30.25 © 2005 American Chemical SocietyPublished on Web 04/12/2005

it is known that unmodified parts of the cellulose chain will behydrolyzed to a higher degree than the modified parts, wherethe substituents typically hinder the enzyme.

The hydrolysis techniques used today are performed insolution, batchwise, with a mixture of enzyme and substrate.6,8

Due to the high cost of producing and purifying enzymes, thetotal concentration of the enzyme in the batch is often low duringthe hydrolysis. The procedure is therefore time-consuming, it isnot possible to automate for online analysis, and it is also verydifficult to recycle the enzyme. This makes the batch methodologyquite inefficient and expensive. To overcome the disadvantagesof the manual batch methods, miniaturization and automation aregenerally good strategies. Silicon micromachining offers severalinteresting possibilities to achieve the goals of low manufacturingcosts and mass production.14,15 Silicon can be made highly porousgiving it a high surface area-to-volume ratio. It is also possible tofabricate well-defined structures with high mechanical strengthand chemical durability, making it feasible to construct analyticalsystems that are extremely small. The main interest in downsizingthe analytical system is to enhance the performance of theanalysis, mainly due to the fact that small dimensions yield verysmall diffusion paths for the reactive species to travel.

In this study, a novel method for hydrolysis of derivatizedcellulose has been developed using a microchip immobilizedenzyme reactor (µIMER) containing the immobilized endogluca-nase from Bacillus agaradhaerens (BaCel 5A). BaCel 5A has amolecular mass of 44 kDa and has been extensively studied withrespect to its crystal structure.16,17 According to these studies, theactive site of BaCel 5A has five subsites, two on the reducing sideof the cleavage site and three on the nonreducing side. Cellobioseis the smallest oligomer that will be produced in any significantamount during hydrolysis, and cellotetraose is the smallestsubstrate for the enzyme. Although the enzyme has five subsites,only four glucose units are necessary for cleavage.12,16,18

The reactors used in this work, developed according to Laurellet al., were constructed from highly porous silicon.19,20 µIMERsof this type have previously been used for sucrose analysis21 andprotein hydrolysis,22 and in the present work, this concept hasbeen expanded to encompass cellulose analysis. The possibilityof immobilizing the enzymes gives a much more efficient method,making it possible to use them for a long period of time with onlya minor consumption of the enzyme. The usage of the µIMERformat also includes the possibility of online connection tochromatographic and MS equipment. The sheer speed of the

µIMER format makes it feasible to carry out extensive measure-ments that are needed to analyze the hydrolysis products that inturn will give information on both the mother polymer and theaction of the enzyme and also provides an increased utilizationand increased cost efficiency. To our best knowledge, endoglu-canases have never before been immobilized on a solid supportfor hydrolysis of derivatized cellulose. We found that a furtheradvantage of using the µIMER format as in the present work isthat this format does not utilize packed material. The open channelstructure reduces the pressure in the system. The highly viscoussolutions that are used would not be able to be pumped througha packed material, such as controlled pore glass or Sephadexsupport materials, without affecting the immobilized enzymes.

EXPERIMENTAL SECTIONMaterials and Methods. Polyethylenimine (PEI; linear, MW

) 750 000, 50% w/v aq) was from Aldrich (St. Louis, MO).Glutardialdehyde (GA, 25% v/v grade I), and NaBH3CN wereobtained from Sigma-Aldrich Corp. (St. Louis, MO). Succinic acid,tris(hydroxymethyl)aminoethane, concentrated NH3 (aq), HCl(aq), and H2O2 were from Merck (Darmstadt, Germany). BaCel5A was a kind gift from the late Dr. Martin Schulein (NovoZymes,Bagsvaerd, Denmark). The full-length enzyme was used includingthe catalytic core and the cellulose binding module.16 The waterused in all experiments was purified in a Milli-Q system fromMillipore (Bedford, MA). The methyl celluloses (MCs) used inthis study were commercial products under the name of Metolose(SM-4, Lot 907540, DS 1.76 (low-mass MC); SM-15, Lot 109614,DS 1.78 (intermediate-mass MC); and SM-1500, Lot 103674, DS1.80 (high-mass MC)) manufactured by Shin-Etsu (Tokyo, Japan).The 1-TIME 2.5-dihydroxybenzoic acid (DHB) precoated MALDIfoils were from LabConnections (Minneapolis, MN). All otherchemicals used were of analytical grade.

Silicon Microchip Reactors. Microchip reactors (µIMERs)were fabricated in 3 in., ⟨110⟩, p-type silicon (20-70 Ω cm, SiltronixSA, Geneva, Switzerland), using standard micromachining tech-niques.23 In short, a 1-µm SiO2 wet oxide was grown and patternedin a standard UV-photolithography process. The SiO2 was usedas etch mask in the subsequent anisotropic etching of the parallelchannel microreactor: etch solution, 30% KOH (30 mg/100 mLof H2O) at 80 °C. The reactors were etched to a channel depth of250 µm. To obtain a porous layer on the channel walls, themicrochip reactor was anodized in an ethanol (96%)-HF (40% inwater) solution mixed (1:1) during illumination. The reactordimension was 13.1 × 3.2 mm, comprising 42 porous flowchannels of approximately 235 µm depth and 25 µm width, givinga total internal surface area of 242 mm2 before fabrication of theporous layer (Figure 1).

To ensure sufficient contact with the electrolyte during theanodization process, the wafer backside was previously doped withboron. The current density during anodization was fixed to 50mA/cm2 for 10 min, to achieve an optimized porosity throughoutthe channel walls.24 After anodization, the reactors were rinsedseveral times in ethanol and H2O, removing any remaining HF.After the porous silicon fabrication, all reactors were dried and

(12) Melander, C.; Bengtsson, M.; Schagerlof, H.; Tjerneld, F.; Laurell, T.; Gorton,L. In manuscript.

(13) Momcilovic, D.; Schagerlof, H.; Rome, D.; Jornten-Karlsson, M.; Karlsson,K.-E.; Wittgren, B.; Tjerneld, F.; Wahlund, K.-G.; Brinkmalm, G. Anal. Chem.In press.

(14) He, B.; Tait, N.; Regnier, F. Anal. Chem. 1998, 70, 3790-3797.(15) Fintschenko, Y.; Van Den Berg, A. J. Chromatogr., A 1998, 819, 3-12.(16) Davies, G. J.; Dauter, M.; Brzozowski, A. M.; Bjørnvad, M. E.; Andersen,

K. V.; Schulein, M. Biochemistry 1998, 37, 1926-1932.(17) Varrot, A.; Schulein, M.; Davies, G. J. J. Mol. Biol. 2000, 297, 819-828.(18) Davies, G. J.; Mackenzie, L.; Varrot, A.; Dauter, M.; Brzozowski, A. M.;

Schulein, M.; Withers, S. G. Biochemistry 1998, 37, 11707-11713.:.(19) Laurell, T.; Rosengren, L. Sens. Actuators, B 1994, 19, 614-617.(20) Laurell, T.; Drott, J.; Rosengren, L. Biosens. Bioelectron. 1995, 10, 289-

299.(21) Lendl, B.; Schindler, R.; Frank, J.; Kellner, R.; Drott, J.; Laurell, T. Anal.

Chem. 1997, 69, 2877-2881.(22) Ekstrom, S.; Onnerfjord, P.; Nilsson, J.; Bengtsson, M.; Laurell, T.; Marko-

Varga, G. Anal. Chem. 2000, 72, 286-293.

(23) Drott, J.; Lindstrom, K.; Rosengren, L.; Laurell, T. J. Micromech. Microeng.1997, 7, 14-23.

(24) Bengtsson, M.; Ekstrom, S.; Marko-Varga, G.; Laurell, T. Talanta 2002,56, 341-353.

Analytical Chemistry, Vol. 77, No. 10, May 15, 2005 3285

inspected through an optical microscope and the obtained channelwidths were measured.

Immobilization of Enzymes on PEI-GA Activated Surface.Prior to the immobilization, the silicon microchips were cleanedin a mixture of 25% NH3, 30% H2O2, and H2O (1:1:5 by volume)followed by cleaning in a mixture of 37% HCl, 30% H2O2, and H2O(1:1:5 by volume) at 100 °C for 5 min per cleaning step. Themicrochips were then thoroughly rinsed with water.

To bind the enzyme, a microchip was placed in 0.2% PEI in 10mM succinate adjusted to pH 5.5 with 1 M NaOH and stirred at4 °C overnight. The PEI forms a thin polymer layer of boundamine groups at the surface of the silicon oxide. After rinsing withsuccinate buffer, the reactor was placed under stirring in 2.5% GAin succinate buffer for 1 h. The GA serves as a cross-linkerbetween the amino groups of the enzyme and the amino groupsof PEI. The reactor was rinsed again with succinate buffer andplaced for at least 12 h at 4 °C in an enzyme solution (100 µM)containing the BaCel 5A enzyme. To deactivate unreacted GA,the reactor was placed under stirring for 1 h in 0.1 M tris-(hydroxymethyl)aminoethane adjusted with 18.5% HCl to pH 5.5.To reduce the bond between the enzyme and the GA, the reactorwas placed in 2 mg/mL NaCNBH3 in 10 mM tris(hydroxymethyl)-aminoethane at pH 5.5. The enzyme was now immobilized bymeans of reductive amination. Finally the reactor was rinsed withtris buffer.

Enzymatic Hydrolysis. For batch hydrolysis, the high-massMC at a concentration of 10 g/L was diluted four times in 12.5mM sodium formiate, was incubated with the enzyme for 72 hwith BaCel 5A (1 µM), and then filtered to remove the enzymeand larger derivatized cellulose chains using 5 kDa cutoff MilliporeUltrafree centrifuge filtration tubes (Millipore). The hydrolysateswere stored at +4 °C prior to analysis.

For µIMER hydrolysis, to lower the viscosity a solution of MCat a concentration of 1 g/L dissolved in 12.5 mM sodium formiatewas pumped using a syringe pump (CMA/Microdialysis, Stock-holm, Sweden) at 0.5 µL/min through the two µIMERs connectedin series. The outlet of the reactors, containing the hydrolysate,was collected in a sample vial or injected directly into a 20 µLsample loop.

ESI-ITMS and LC-ESI-ITMS. Direct infusion electrosprayionization ion-trap mass spectrometry (ESI-ITMS) was performedon an Esquire-LC (Bruker Daltonik, GmbH, Bremen, Germany).Direct infusion of the analyte solution was performed at 3L/min. Nitrogen was used as drying gas at a flow rate of 3L/min with a temperature of 350 °C. Nitrogen was also used asnebulizer gas at 7 psi. The following voltages were used: nebulizerend plate 4000 V, end cap 3500 V, and capillary exit 150 V. Themass spectra from the full-scan experiments were averaged over100 scans. In full-scan mode, the ESI-ITMS was typically set toscan between m/z 100 and 1500. All batch-hydrolyzed sampleswere filtered prior to injection to remove the enzyme. Forcomparison, µIMER-hydrolyzed samples were filtered with thesame filter but at an analyte concentration of 1 g/L.

For the online method, the outlet of two series-connectedµIMERs was connected to a Rheodyne valve with a 20 µL loop(Figure 2). The same settings of the mass spectrometer were usedexcept for nitrogen, which was used as drying gas where the flowwas 7 L/min and the nebulizer at 30 psi. Trap drive was set to47.4, optimizing instrumental sensitivity at m/z ∼700. Carrierliquid was MilliQ water at a flow rate of 0.5 mL/min. Afterinjection, the sample was separated on a TSK GMPWXL sizeexclusion precolumn (40 mm × 7.8 mm2) (TosoHaas Biosepara-tion Specialists, Stuttgart, Germany) and then injected into theESI interface.

SEC/MALDI-TOF-MS. Size exclusion chromatography in-terfaced with matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (SEC/MALDI-TOFMS) was performedusing an LC-Transform Series 500 interface (Round Rock, TX)and a PerSeptive Voyager-DE STR MALDI-TOF mass spectrom-eter (Applied Biosystems, Framingham, MA) according to meth-odology presented by Momcilovic et al. and used with somemodifications.25 In this setup, the SEC effluent is sprayed from aheated steel capillary, by assistance of a sheath gas, onto a MALDIfoil that is coated with DHB. The matrix coating is briefly wettedby the solvent so that the analyte can be incorporated into thematrix crystals. The SEC column was either a TSKgel G 3000PW (30 cm × 7.5 mm i.d., TosoHaas) or a TSK GMPWXL mixed-bed column (30 cm × 7.8 mm i.d., TosoHaas). The µIMER-hydrolyzed MC (at a concentration of 0.6 g/L in 12.5 mM sodiumformiate (pH 5.5) buffer) collected from the two series-connectedµIMERs was injected without any pretreatment. The movementrate of the sample stage that supported the MALDI target in theLC-Transform interface was 4 mm/min.

(25) Momcilovic, D.; Wittgren, B.; Wahlund, K.-G.; Brinkmalm, G. RapidCommun. Mass Spectrom. 2005, 19, 947-954.

Figure 1. Microchip immobilized enzyme reactor. The reactor isapproximately 9 mm long and 3 mm wide. It consists of two reservoirsthat are connected by channels wherein the channel walls are highlyporous. Due to the high porosity of the walls, a high surface area isobtained.

Figure 2. Delivery of the substrate solution to the µIMER performedusing a syringe pump. The outlet is connected directly to a loop foronline injection to the ESI-ITMS system. Alternatively, the hydrolyzedproduct is collected in a sample vial. In this work, online analysis ofthe hydrolyzate has been performed using LC-ESI-ITMS. However,connection to any detection systems with loop injection is possible.

3286 Analytical Chemistry, Vol. 77, No. 10, May 15, 2005

MALDI-TOF mass spectra were acquired in both linear andreflector modes. The instrument settings for the reflector modeacquisitions have been described elsewhere.25 For the linear modeacquisitions, an acceleration voltage of 25 kV was used. The laserintensity was held slightly above threshold, and the lag time was400 ns plus instrument offset. The guide wire voltage was set to0.05% of the acceleration voltage. Mass spectra were accumulatedfor 100-300 laser shots.

SEC-MALS-RI. The hydrolyzed samples were analyzed bySEC coupled to multiangle light scattering and refractive index(SEC-MALS-RI) detection. The separation was carried out usinga TSKgel GMPWXL column (30 cm × 7.8 mm i.d., TosoHaas).The pump in use was a Shimadzu LC-10AD liquid chromatographypump (Shimadzu Corp., Tokyo, Japan), and the degasser was aShimadzu DGU-14A. Injection of the polymer solution was carriedout by a Waters 717 plus autosampler. MALS and RI detectionwas performed utilizing DAWN EOS MALS and Optilab RIdetectors (Wyatt Technology Corp., Santa Barbara, CA), respec-tively. A solution consisting of 10 mM NaCl + 0.02% NaN3 wasused as mobile phase at a flow rate of 0.5 mL/min, and the samplevolume was 100 µL. The analyte concentration was 0.75 g/L, anddouble injections were made. A 0.05 µm VM Millipore filter wasplaced between the pump and the injector. A Sartorius CA0.45 µm filter (Gottingen, Germany) was placed after the columnto improve the MALS signal. Astra for Windows version 4.73.04was used for the data evaluation.

RESULTS AND DISCUSSIONBy using µIMERs, the possibility of performing hydrolysis with

online analysis, e.g., MS, is provided. In batch experiments,hydrolysis has been performed completely offline and the enzymehad to be removed by filtration. In this work, the µIMER wasconnected directly to LC-MS, where separation of unwantedcomponents could easily be performed. SEC has previously beenused for the determination of the average molar mass and themolar mass distribution of enzymatic hydrolysates of starch andcellulose derivatives.26-28 If SEC is used together with RI andMALS detection, determination of the molar masses of thehydrolysis products can give information about the enzyme actionon the derivatized polysaccharide without any standards.29-31

MALDI-TOF-MS has been employed in a variety of applicationsfor carbohydrate analysis.32 As with many types of mass spec-trometry, MALDI-TOF-MS can provide valuable information onseveral aspects of structural analysis, such as determination ofsequence, branching, and linkage. One great advantage is thatMALDI-TOF-MS can be used to analyze polysaccharides up to10 000 Da with high mass accuracy. The limitations of MALDI-TOF-MS in the lower mass range (m/z <500), due to matrix ionsproduced by the laser pulse, make other MS methods moresuitable.5 In this range, ESI-ITMS is a more convenient analytical

method, providing valuable information on polysaccharides up tom/z of ∼2200.

SEC-MALS-RI. Three MCs with different Mw but similar DSwere hydrolyzed to study the activity of the µIMER in order todetect qualitative variations. By studying the SEC-MALS-RI datait was possible to verify that the enzyme did not denature duringthe immobilization and worked as expected. The diffusion ofsubstrates is highly dependent on molecular weight. Substrateswith high diffusion rate (low molecular weight) can interact withthe immobilized enzymes to a higher degree than those with lowdiffusion rate (high molecular weight). To minimize the influenceof the molecular weight on the efficiency of the hydrolysis, a flowrate of 0.5 µL/min was employed and two reactors were connectedin series. The hydrolysis was monitored with ESI-MS until nofurther hydrolysis could be seen (see below). Then furtheranalyses were performed.

Figure 3 shows the molar mass distributions of the three MCsmeasured by SEC-MALS-RI prior to and after µIMER hydrolysis.For all MCs, the average molar mass decreased significantly afterhydrolysis. This strongly indicates that the immobilized BaCel5A hydrolyzes the MCs. The immobilization of enzyme onto thesurface did not degrade the enzyme function. Table 1 shows thenumber average (Mn) and weight average (Mw) molar masses asmeasured by SEC-MALS-RI. The molar mass distribution of thehigh-mass MC (Figure 3c) has a tailing shape in contrast to theother two that have a more log-normal distribution (Figure 3a,b).The reason for this is unclear, but a possible explanation couldbe that a fraction of the high-mass MC is resistant to hydrolysisby BaCel 5A. This could explain the tailing shape of the molarmass distribution. The Mw/Mn value for the high-mass MC actuallyincreased significantly after the hydrolysis, indicating a broaderrange in polydispersity of formed products. Regions in the high-mass MC with high amounts of substituents might be inaccessiblefor the enzyme.

The distribution of substituents along a cellulose chain canbe divided into three major classes. A random distribution of thesubstituents along the cellulose chain will be the outcome if allglucose monomers in the chain are equally accessible for substitu-tion during manufacturing. In oligomeric sequences, this randompattern can be calculated by binominal distribution statistics. Ifthe glucose units are not equally accessible during the manufac-turing process, e.g., due to crystalline or supramolecular struc-tures, unsubstituted glucose units in the cellulose chain will bethe outcome. This will give a more heterogeneous pattern of thesubstituents. The distribution can also be blocklike, meaning thata part of the cellulose chain is fully or very highly substituted.33

Studies on hydrolysis of carboxymethyl cellulose (CMC)performed by Horner et al.26 showed that the endoglucanaseactivity on the derivatized cellulose is strongly dependent on theDS in different parts of the derivatized cellulose. Areas that arehighly substituted within the cellulose chain were hydrolyzed toonly a small extent, whereas areas that were less substituted werehydrolyzed to a higher extent. As the density of substitution in-creased, the efficiency of the enzyme became more limited. Saakeet al.27 have verified the presence of highly substituted areas withinthe cellulose chain. They showed that the high-mass fractions of

(26) Horner, S.; Puls, J.; Saake, B.; Klohr, E. A.; Thielking, H. Carbohydr. Polym.1999, 40, 1-7.

(27) Saake, B.; Horner, S.; Kruse, T.; Puls, J.; Liebert, T.; Heinze, T. Macromol.Chem. Phys. 2000, 201, 1996-2002.

(28) Steeneken, P. A. M.; Woortman, A. J. J. Carbohydr. Res. 1994, 258, 207-221.

(29) Saake, B.; Horner, S.; Puls, J. ACS Symp. Ser. 1998, No. 688, 201-216.(30) Beri, R. G.; Walker, J.; Reese, E. T.; Rollings, J. E. Carbohydr. Res. 1993,

238, 11-26.(31) Roger, P.; Colonna, P. Carbohydr. Polym. 1993, 21, 83-89.(32) Harvey, D. J. Mass Spectrom. Rev. 1999, 18, 349-450.

(33) Mischnick, P.; Heinrich, J.; Gohdes, M.; Wilke, O.; Rogmann, N. Macromol.Chem. Phys. 2000, 201, 1985-1995.


the hydrolyzed CMC (DS 0.8-1.9) had a higher DS than the low-mass fractions.

All the MCs employed in this study have approximately thesame DS but their average molar masses are different. If thesubstituents were distributed in a random manner for the threeMCs, that is, if the MCs have the same heterogeneity, then theenzyme should have the same ability to hydrolyze the cellulosechains. BaCel 5A is an endoglucanase and is most likely notintrinsically sensitive to differences in molecular weight betweenthe MCs. If the enzyme would hydrolyze every possible substrate,i.e., proceed until no further hydrolysis can be seen, then the finalmolar mass of the hydrolysates would be expected to be similarand monomodal. However, if one or more of the MCs haveblocklike parts or parts with very high DS, the enzyme wouldnot be able to hydrolyze these parts. These highly substitutedregions of the MCs would hinder the enzyme and remainunhydrolyzed, whereas the low-substituted areas would have been

hydrolyzed to oligomers. This would result in fractions with ahigher molar mass. The peak apexes of the intermediate- and thehigh-mass MC after exhaustive hydrolysis coincide at ∼10 000Da (Figure 3d). This indicates that a large fraction of the formedhydrolysates for the two MCs has the same molar mass andapproximately the same distribution of the substituents for thesefractions. For the high-mass MC, however, a much broader massrange is obtained with intact fractions of the MC, which areunhydrolyzed or hydrolyzed to a very small extent.

ESI-ITMS of Filtrated Batch and µIMER Hydrolyzed MCs.To improve direct infusion ESI-ITMS, the hydrolysates werefiltered prior to analysis. This reduced the relative signal intensitiesof interfering peaks originating from multicharged componentsand simplified the interpretation of the mass spectra. For samplesobtained from batch hydrolysis, this step also removed theenzyme, which could otherwise impair the quality of the massspectra.

Figure 4 shows full-scan mass spectra of the hydrolysates ofthe high-mass MC obtained from batch hydrolysis (Figure 4a)and µIMER (Figure 4b). Both for batch and for µIMER hydroly-sates substituted oligomers with DPs 1-6 were detected (Figure4). The degree of µIMER and batch hydrolysis of the high-massMC was compared by the relative signal intensities in the massspectra. Figure 5 shows the relative intensities for the twohydrolysis methods. No significant differences can be seen, exceptfor the peak at m/z 393 (DP 2 with two substituents) obtainedfor the µIMER hydrolysate (Figure 4). This peak was, however,also found when blank samples containing no MC were analyzed,and therefore, it could be concluded that it (at least) partly

Table 1. SEC-MALS-RI Analysis of the Intact MC andMC Hydrolyzed by BaCel 5A in a µIMERa

mass

Mw(unhydro/

hydro)

Mn(unhydro/

hydro

Mw/Mn(unhydro/

hydro)

low 22400/10200 12200/6800 1.83/1.50intermediate 42600/19500 24900/9600 1.71/2.02high 200200/40100 79900/10500 2.50/3.81

a Weight-average molar mass Mw of unhydrolyzed (unhydro) andhydrolyzed (hydro) samples, Number-average molar mass, Mn, poly-dispersity index, Mw/Mn.

Figure 3. Molar mass distribution obtained by SEC-MALS-RI of unhydrolyzed and µIMER-hydrolyzed (a) low-mass, (b) intermediate-mass,and (c) high-mass MC. The comparison of distributions after hydrolysis for the different MCs is shown in (d). Each curve is an average of twosamples.


originated from a nonanalyte component and was thereforesubtracted from the spectra. The good correlation between thepeaks is a strong indication that the ability of the enzyme tohydrolyze substrate was not altered after immobilization.

Previous studies have shown that unsubstituted DP 3 (m/z527) is a possible product for this enzyme, but no peak corre-sponding to this molecule was detected in the spectra (Figure5).12 The high DS (1.80) of this MC, however, makes the presenceof unsubstitued DP 3 improbable.34,35

To verify that the µIMER hydrolysis had reached completion,a hydrolyzate was rehydrolyzed by the µIMER and compared by

ESI-MS to those not rehydrolyzed to investigate whether the firsthydrolysis was sufficient. The relative peak intensities in ESI-ITMSof hydrolyzed and rehydrolyzed MC correlated, indicating thatthe µIMER hydrolysis as detected by ESI-MS was complete afterthe first pass through the µIMER. No significant difference inthe spectra could be seen with hydrolysis flow rates below 0.5µL/min.

When analyzing the ESI-ITMS spectra of the MCs of differentMw, an apparent difference is exhibited (Figure 6). It can be seenin Figure 6 that the substituent distribution for the low-mass MChas its apex at a significantly higher DS-value than the other twoMCs at all DPs. A possible explanation for the shift for the low-mass MC could be that it has a different substituent distributionthan that for the intermediate- and high-mass MCs, that it isregioselectively substituted, or that the average molar mass inthe unhydrolyzed low-mass MC is significantly lower. Therefore,a larger fraction of the obtained products does not originatefrom the enzymatic hydrolysis than for the other two MCs.13 Inthe low-mass MC, there might also be a contribution from theoriginal reducing ends that may be highly substituted. Sub-stituents may hinder the hydrolysis to different degrees de-pending on the position of the substituent within the four glu-cose units necessary for hydrolysis. The production of sub-stituted DP 2, especially the fully substituted DP 2 (m/z 449),indicates that the enzyme is capable of performing hydrolysisadjacent to a fully substituted glucose unit (Figures 5 and 6).13

Thus, the areas where the enzyme acts do not need to becompletely unsubstituted. Since the MC is only degraded to asmall extent (seen with SEC-MALS-RI) and it is possible to findfully substituted DP 2, it is shown that the enzyme capability tocope with substituents is very complex. Otherwise it is expectedthat the MC would become completely hydrolyzed. According tostudies done on carboxymethyl cellulose by the groups of Saakeand Horner, cleavage can only occur between two adjacentunsubstituted glucose units.26,27 However, another enzyme wasused in that work.

Online Hydrolysis with µIMER and Analysis UsingSEC/ESI-ITMS. µIMER hydrolysis online with chromatographic

(34) Arisz, P. W.; Kauw, H. J. J.; Boon, J. J. Carbohydr. Res. 1995, 271, 1-14.(35) Kern, H.; Choi, S. C.; Wenz, G.; Heinrich, J.; Ehrhardt, L.; Mischnick, P.;

Garidel, P.; Blume, A. Carbohydr. Res. 2000, 326, 67-79.

Figure 4. Full-scan ESI-ITMS spectra of hydrolyzed high-mass MCobtained by batch (a) and µIMER (b) hydrolysis.

Figure 5. Relative peak intensities from analytes with DPs 2-5 inESI-ITMS obtained by µIMER (b) and batch (O) hydrolysis. Intensitydifferences for the samples between the different DPs can beexplained by variation in concentration between the samples. Eachcurve is an average of data obtained from three identical experimentsof the high-mass MC.

Figure 6. Relative peak intensities in ESI-ITMS originating fromhydrolysis products, with DPs 2-5, obtained by µIMER hydrolysis ofdifferent MC molar masses. 1, low-mass MC; O, intermediate-massMC; b, high-mass MC. Each oligomer with varying amount ofsubstituents has been normalized according to the highest peak withinthat oligomer.


separation and mass spectrometric detection was achieved byconnecting the µIMER to an HPLC system consisting of an SECcolumn and the ESI mass spectrometer. The column served asan online filter, removing the high-mass components from theoligomers of interest (DP 1-6). Qualitative data obtained withSEC/ESI-ITMS correlated well with those from ESI-MS of filteredsamples. It can be seen in Figure 7 that the introduction of thecolumn resulted in a 20% decrease in signal intensity comparedto the offline-filtered samples; however, a significant increase insignal-to-noise ratios (∼10 times) was achieved and peaks origi-nating from multicharged ions were significantly reduced. For theoffline-filtered samples, peaks originating from multichargedcellulose products could be observed among the peaks from thesingly charged ions originating from smaller oligomers. Thisincreased the limit of detection for the smaller oligomers.

Analysis of oligomers with DP 1-6 was performed within 5min using the SEC/ESI-ITMS method. From the total ion currentof the expected peaks, it could be determined that the compoundsof interest eluted between 3.6 and 5 min. A large part of theinterfering compounds eluted prior to this interval making itpossible to perform a fast cleanup of the sample.

Applying the SEC column to the system made it possible toperform the hydrolysis and analysis of MC using µIMER andESI-ITMS completely online. Multiple injections of a sample couldbe made with a time spacing of 60 min. New samples werehydrolyzed and analyzed simply by changing the solution in thesyringe pump. The same µIMER could be employed for the differ-ent samples. Introduction of a completely new MC sample could

therefore be performed every 2 h and after that a new injectionof the sample could be done every 60 min. This time can be greatlyreduced if a smaller loop would be used, if a higher flow rate wouldbe used, or if several reactors would be used in parallel. Thehydrolysis products could be detected with stable and reproduciblepeak intensities for over 10 days with the same µIMER and samesurface loading of enzyme, making it possible to analyze a largenumber of samples. Connection to an autosampler would decreasesome of the manual work required to change the substratesolutions. This ability will be tested in forthcoming work.

SEC/MALDI-TOFMS of Unfiltered µIMER Hydrolysates.The µIMER hydrolysates were also subjected to SEC/MALDI-TOFMS analysis according to Momcilovic et al.25 in order toincrease the mass range analyzed by mass spectrometry. Byinterfacing SEC with MALDI-TOFMS, no separation of high-massanalytes through filtration was necessary. Hence, no samplepretreatment of the hydrolysates prior to SEC/MALDI-TOFMSwas necessary. Figure 8 shows mass spectra of µIMER-hydrolyzedhigh-mass MC. Molecules with DPs from 3 up to (at least) 22were detected in reflector mode MALDI-TOFMS. No detectablesignal, in reflector mode analysis, was observed for elutionvolumes of <7.5 mL using the TSK 3000 column. However, inlinear mode analysis, peaks originating from the MC could bedetected at positions on the MALDI targets corresponding toelution volumes of g6.9 mL (data not shown). The spectrumcollected at the position corresponding to the elution volume of6.9 mL contained a single broad peak with a maximum aroundm/z 6300. According to the molar mass distribution measured

Figure 7. ESI-IT mass spectra of hydrolyzed intermediate-massMC obtained using an SEC separation (a) and without a separation(b). The use of the SEC column offers the possibility for an onlinesystem for hydrolysis and detection.

Figure 8. Reflector mode MALDI-TOF mass spectra of µIMER-hydrolyzed high-mass MC separated using the TSKgel G 3000 PWcolumn. The spectra were collected at positions on the MALDI foilcorresponding to the elution volumes given in the figure. The numbersindicate the DP of the molecules within each cluster of peaks. Theconcentration of the µIMER-hydrolyzed MC was 0.6 g/L. The columnexclusion and total permeation volumes were ∼5 and ∼10 mL,respectively, according to the manufacturer.


by SEC-MALS-RI (Figure 3), the major fraction of this sample hada molar mass above 6500 g/mol. However, since the sensitivityin MALDI-TOFMS seems to decrease with increasing mass forMC, the major fraction of this sample could not be detected.Nevertheless, the DP range of the molecules detected bySEC/MALDI-TOFMS was significantly larger than that detectedby ESI-ITMS. The advantage with ESI-ITMS lies in that the relativesignal intensities from molecules with DP < 3 are higher than inMALDI-TOFMS. Hence, complementary data can be obtained bythe two techniques as has been shown previously for hydrolyzedstarch.36

µIMER-hydrolyzed high-mass MC was also separated on alinear mixed-bed column and spray deposited onto MALDI targets.Figure 9 shows a linear mode MALDI-TOF mass spectrumobtained through continuous movement of the MALDI targetduring acquisition. In this way, analyte components eluted in awider volume range were subjected to MALDI-TOFMS; hence,the wide mass distribution observed in the spectrum. Peaks fromoligomers with DPs up to (at least) 30, with the various degreesof methylation well resolved, could be detected in linear modeMALDI-TOFMS. The mass range of the analyte componentsobserved in this spectrum is much wider than that observed fromsamples prepared by the conventional dried-droplet samplepreparation method for MALDI.25 This demonstrates the ap-plicability of a continuous spray deposition interface for coupling

SEC online with sample preparation for MALDI when analyzingpartially hydrolyzed cellulose ethers.

CONCLUSIONSThe results presented in this work show that the µIMER with

immobilized endoglucanases can be used for hydrolysis ofderivatized cellulose. To the best of our knowledge, this is thefirst time an endoglucanase has been reported immobilized on asolid support for hydrolysis of methyl cellulose and this is alsothe first time the hydrolysis of cellulose derivatives has beenperformed coupled online to ESI-ITMS.

Comparison of µIMER- and conventional batch-hydrolyzedmethyl cellulose showed that the immobilization procedure didnot affect the selectivity of the enzyme. For molecules with a DPof 2-6, the same hydrolytic products and ratios as in batchhydrolysis could be detected for µIMER-hydrolyzed samples usingESI-ITMS. It was also possible to distinguish differences betweenMCs of various qualities with ESI-ITMS analysis of the hydrolysisproducts formed in the µIMER. By studying the formed hydrolysisproducts, information on the enzymatic activity could also befound. In SEC/MALDI-TOFMS measurement on nonfilteredµIMER hydrolysates, oligomers with a DP of 3 up to at least 30with various degrees of substitution could be detected. Nopretreatment of the hydrolysate solution obtained from the µIMERwas necessary prior to MS. The relative peak intensities foroligomers with DP < 3 were lower in SEC/MALDI-TOFMScompared to in SEC/ESI-ITMS; hence complementary data wereobtained from these techniques.

Size exclusion chromatography with multi-angle light scatteringand refractive index detection (SEC-MALS-RI) data indicate thatthe µIMER could be used for screening of different cellulosederivatives. Differences in the substituent distribution betweenthe MCs could be estimated by comparing the spectra of theµIMER-hydrolyzed MCs with the data from unhydrolyzed MCs.The use of the µIMER method substantially decreased thenecessary time for hydrolysis and can be performed completelyonline. This should be compared with the conventional batchmethod that requires several hours or even days to complete andalso needs extensive manual labor. This makes the µIMER methodapproximately 20-100 times faster than the conventional batchmethod. The low consumption of enzyme taken for immobilizationand the possibility for continuous hydrolysis also make theproposed µIMER hydrolysis method a very cost efficient one.

ACKNOWLEDGMENTWe thank the Centre for Amphiphilic Polymers for financial

support and Lisa Eriksson at AstraZeneca R&D Molndal for helpwith the SEC-MALS-RI.

Received for review January 31, 2005. Accepted March 15,2005.

AC050201R(36) Richardson, S.; Nilsson, G.; Cohen, A.; Momcilovic, D.; Brinkmalm, G.;

Gorton, L. Anal. Chem. 2003, 75, 6499-6508.

Figure 9. Linear mode MALDI-TOF mass spectrum of µIMER-hydrolyzed MC separated using the TSKgel GMPWXL column. Thespectrum was accumulated through continuous movement ofthe MALDI foil at positions corresponding to the elution volumes8.5-9.4 mL. Peaks are indicated as in Figure 8. The concentrationof the µIMER-hydrolyzed MC was 0.6 g/L. The column exclusion andtotal permeation volumes were ∼5 and ∼11 mL, respectively,according to the manufacturer.


microchip immobilized enzyme reactors for hydrolysis of methyl cellulose

Documents