Biomacromolecules 2005, 6, 2793-2799
2793
Improved Chemical Analysis of Cellulose Ethers Using Dialkylamine Derivatization and Mass Spectrometry Dane Momcilovic,† Herje Schagerlo¨f,‡ Bengt Wittgren,§ Karl-Gustav Wahlund,† and Gunnar Brinkmalm*,§ Department of Technical Analytical Chemistry, Lund University, Post Office Box 124, S-221 00 Lund, Sweden, Department of Biochemistry, Lund University, Post Office Box 124, S-221 00 Lund, Sweden, and AstraZeneca R&D Mo¨lndal, S-431 83 Mo¨lndal, Sweden Received April 15, 2005; Revised Manuscript Received June 1, 2005
Oligosaccharides of hydroxypropylmethyl cellulose, hydroxypropyl cellulose, and methyl cellulose were investigated by matrix-assisted laser desorption/ionization mass spectrometry (MALDI-MS). The cellulose ether oligosaccharides were produced either by enzymatic depolymerization utilizing the purified family 5 endoglucanase from Bacillus agaradhaerens or by partial acidic depolymerization. To lower the limit of detection in MALDI-MS three dilakylamines, dimethyl-, diethyl-, and dipropylamine were studied as reagents for reductive amination of the oligosaccharides. All three amines contributed to a significant increase in sensitivity in MALDI-MS, especially for oligosaccharides with a degree of polymerization (DP) < 3. These reagents were also attractive due to their high volatility, which facilitated the purification of the reaction mixtures. It was established that low-mass discrimination in MALDI-MS in the DP range 1-7 was substantially reduced with dialkylamine derivatization. Hence, dialkylamine derivatization of cellulose ether oligosaccharides obtained by endoglucanase depolymerization increased the number of detected analyte components. Dimethylamine was concluded to be the preferred reagent of those evaluated. Introduction Chemical characterization of cellulose ethers is of great interest since these polymers have many applications in a wide range of fields. Because of their nontoxic properties they are used in, for instance, the pharmaceutical industry as barrier providers and emulsion stabilizers and in the food industry as thickeners and foam stabilizing agents. For many applications, the demands on the technological properties of the cellulose ethers are high. Therefore, it is necessary to establish analytical methods by which the technological properties can be correlated to the chemical structure.1-4 Recently, hydroxypropyl cellulose (HPC) was investigated by nuclear magnetic resonance (NMR) spectroscopy, matrixassisted laser desorption/ionization mass spectrometry (MALDI-MS), and high-performance anion-exchange chromatography (HPAEC).5 The degree of substitution (DS) and the substitution heterogeneity could be measured. In another study, the substituent distribution within the anhydroglucose units in methyl cellulose (MC) was measured by electrospray ionization tandem mass spectrometry (ESI-MS/MS).6 During the past decade, ESI- and MALDI-MS have become frequently applied techniques for the identification and structure analysis of oligosaccharides.7-11 For many oligosaccharides, however, the sensitivities in ESI- and MALDI-MS are relatively low. This is mainly due to the * To whom corresponance should be addressed. Tel.: +46-31-7761013. Fax: +46-31-7763834. E-mail:
[email protected]. † Department of Technical Analytical Chemistry, Lund University. ‡ Department of Biochemistry, Lund University. § AstraZeneca.
absence of easily ionized functional groups. Therefore, many approaches to increase the sensitivity in ESI-MS and MALDI-MS by means of derivatization have been presented.9,10,12-18 A common strategy has been to introduce a derivative containing an easily ionized functional group at the oligosaccharide reducing end. This increases the ionization efficiency in MALDI and ESI; hence, the sensitivity is increased. Introduction of a permanently charged quarternary amine, such as the Girard’s T reagent, which was introduced by Naven et al., has been demonstrated as a very useful approach for increased sensitivity for oligosaccharides in MALDI-MS.5,19-21 Derivatization by reductive amination is generally performed with a large molar excess, sometimes up to 1000fold, of the amine and reducing agent.15-18 Because of that, removal of excess reagents must be performed prior to MS. For monodisperse oligosaccharides, this can be achieved by, for instance, reverse-phase high-performance liquid chromatography (RP-HPLC). For partially depolymerized cellulose ethers, however, our experience shows that RP-HPLC is not the preferred purification method. This is because partially depolymerized cellulose ethers usually have both wide mass distributions and large variations in hydrophobicity, which gives broad elution windows. Therefore, we have chosen aqueous size-exclusion chromatography (SEC) for the purification of the reaction mixtures. In our hands, however, common reductive amination reagents, such as p-aminobenzoic acid (N,N-diethylamino)ethyl ester13,16 and p-aminobenzoic acid ethyl ester,12,17 are not compatible with SEC. This is probably due to their relatively high hydro-
10.1021/bm050270f CCC: $30.25 © 2005 American Chemical Society Published on Web 07/02/2005
2794
Biomacromolecules, Vol. 6, No. 5, 2005
phobicities, which cause interaction with the column stationary phase. Therefore, one aim of this work was to facilitate SEC purification of reducing-end derivatives by employing volatile derivatization reagents that can be easily removed after the reaction by evaporation. This requirement is fulfilled by the dialkylamines, dimethylamine (DMA), diethylamine (DEA), and dipropylamine (DPA). The other and most important aim was to increase the sensitivity in MALDIMS of partially depolymerized cellulose ethers. This is of great interest since it improves the structure information that can be obtained by combining selective enzymatic depolymerization of cellulose ethers and MALDI-MS. Experimental Section Chemicals. MC SM-1500 (viscosity type: 1500 cP) with the stated DS ) 1.80 and HPC L-HPC LH20 with the stated molar substitution (MoS) ) 0.33 were from Shin-Etsu Chemical Co. (Tokyo, Japan). Hydroxypropylmethyl cellulose (HPMC) Methocel K100LV (viscosity type: 100 cP) with the stated DSmethyl ) 1.49 and MoShydroxypropyl ) 0.21 was from Dow Chemical Co. (Midland, MI). The MALDI matrices 2,5-dihydroxybenzoic acid (DHB) and R-cyano-4hydroxycinnaminic acid (CHCA), as well as dimethylamine (DMA, pKb 3.23 at 25 °C) 2.0 mol L-1 in methanol (MeOH), diethylamine (DEA, pKb 3.07 at 25 °C), and dipropylamine (DPA, pKb not found), were from Aldrich (Steinheim, Germany). Arabinosazone22 was a kind gift from Prof. Milos Novotny (Indiana University, Bloomington, IN). Malto- and cellooligosaccharide standards and NaBH3CN were from Sigma (St. Louis, MO). Trifluoroacetic acid (TFA) was purchased from Riedel-de Hae¨n (Seelze, Germany), and the H2O was from an ELGA Maxima (18.2 MΩ cm, Vivendi Water Systems, High Wycombe, UK). All other chemicals were of analytical grade. All chemicals were used without further purification. Oligosaccharide Standards. The commercial mixture of cellooligosacharide was dissolved to 1 g L-1 in H2O. The equimolar mixture of cellooligosaccharides was prepared by mixing glucose to cellopentaose to a total concentration of 1 g L-1 in H2O. In the same way, the equimolar mixture of maltooligosaccharides was prepared by mixing glucose to maltoheptaose to a total concentration of 1 g L-1 in H2O. Acidic Depolymerization. Partial acidic depolymerization was performed according to Momcilovic et al.23 Enzymatic Depolymerization. The purified cellulose selective family 5 endoglucanase from Bacillus agaradhaerens (Ba Cel5A) with a molar mass of 44 703 g mol-1 was a kind gift from the late Dr. Martin Schu¨lein (Novozymes, Bagsværd, Denmark).24,25 To 2 mL of the HPMC and HPC solutions, 1 g L-1 in H2O, Ba Cel5A was added to a final concentration of 1 µmol L-1. Particles were observed in the HPC solution indicating that a large fraction of this cellulose ether was not dissolved. No particles were observed for the MC and HPMC samples. Depolymerization was carried out for 72 h at room temperature, whereafter the solutions were filtered using a centrifuge and Nanosep 10 kDa Omega filters (Pall, Ann Arbor, MI). The permeate was evaporated using
Momcilovic et al.
a Christ rotation vacuum concentrator model 2-18 (Martin Christ GmbH, Osterode am Harz, Germany). Dialkylamine Derivatization. Solutions of DEA and DPA (2.0 mol L-1 in MeOH) were prepared. To 500 µL of 2.0 mol L-1 dialkylamine solution in MeOH, 63 µL acetic acid (HAc) was added. Thereafter, 40 µL of this mixture was mixed with 40 µL of the solution containing the partially depolymerized cellulose ether in H2O (∼2.5 g L-1) and 120 µL MeOH in a 1.5-mL screw-capped glass vial. The resulting solution was heated for 5 h at 80 °C for DMA and 8 h at 80 °C for DEA and DPA in an HLC Heating ThermoMixer 130 R (Haep Labor Consult, Bovenden, Germany) with constant shaking. Thereafter, 10 µL of NaBH3CN (10 g L-1 in MeOH) was added, and the resulting mixture was held at 80 °C for another 4 h. Finally, 500 µL H2O was added, and the solution was shaken in ambient temperature for 12 h. The solvent and dialkylamine were evaporated under a stream of N2 and heating to 50 °C. The residue was dissolved in 140 µL of 300 mmol L-1 NH4Ac (pH 5)/acetonitrile (ACN) 70/30 (v/ v). This solution was then used in the SEC purification. The derivatization of the malto- and cellooligosaccharides was performed using the same procedure as above. The concentrations of the equimolar mixtures of malto- and cellooligosaccharides were 0.21 and 0.40 mmol L-1, respectively. Purification of Dialkylamine Derivatives. The dialkylamine derivatives were purified by SEC. The HPLC pump was a Hewlett-Packard Series 1050 (Agilent Technologies, Palo Alto, CA) fitted with a degasser. A TSKgel G 3000 PW column (30 cm × 7.5 mm inside diameter, TosoHaas Bioseparation Specialists, Stuttgart, Germany) with 300 mmol L-1 NH4Ac (pH 5)/ACN (70/30, v/v) as mobile phase at a flow rate of 0.5 mL min-1 was used for the separation. According to the manufacturer the resolution window of this column is approximately 100-50 000 g mol-1 as calibrated for poly(ethylene oxide)s. It can be expected that for linear oligo- and polysaccharides the range may be shifted to slightly higher molar masses. Samples were injected using a Rheodyne model 7010 valve-injector (Rheodyne, Berkeley, CA) fitted with a 100µL sample loop. The detector, a Shimadzu RID10A refractive index (RI) detector (Shimadzu, Kyoto, Japan), was disconnected after the determination of the elution volume windows for the dialkylamine derivatives and the unwanted components. Thereafter, volume fractions were collected and subsequently evaporated under a stream of N2 and heating to 50 °C. MALDI-MS. MALDI-MS experiments were performed using a PerSeptive Voyager-DE STR time-of-flight mass spectrometer (Applied Biosystems, Framingham, MA) according to Momcilovic et al. with some modifications.26 The detector activation gate was set at m/z 185, and the mass spectra were acquired in positive ion mode with the reflector activated. The guide wire was set to 0.001% of the acceleration voltage. Spectra were accumulated for 150-300 laser shots. Sample preparation using DHB was performed through drying of the samples at reduced pressure or in ambient atmosphere according to a procedure described earlier.27 For preparations with CHCA, the matrix and the analyte were
Reducing-End Derivatization of Cellulose Ethers
Figure 1. Size-exclusion chromatogram, as measured by RI detection, of reaction mixture containing DMA-derivatized cellobiose. The analyte-containing fraction was collected up to 9.3 mL.
dissolved in 0.1% HAc (in H2O)/ACN 50/50 (v/v) at 10 and 1 g L-1, respectively. For preparations with arabinosazone, the matrix and the analyte were dissolved in methanol at 10 and 1 g L-1, respectively. Matrix and analyte solutions were mixed 80/20 (v/v), and 1 µL of this mixture was applied on the MALDI target. Unless stated otherwise, spectra were accumulated from samples prepared using DHB together with H2O and drying at reduced pressure.
Biomacromolecules, Vol. 6, No. 5, 2005 2795
Figure 2. Reflector mode MALDI mass spectra of (a) underivatized cellooligosaccharides and (b) DMA-derivatized cellooligosaccharides. DHB was used as matrix and H2O as solvent. The samples were dried at reduced pressure. The numbers given indicate the DP of the cellooligosaccharides. “H” and “Na” indicate peaks originating from [M + H]+ and [M + Na]+, respectively.
Results and Discussion Purification of Reaction Mixtures. Figure 1 shows the size-exclusion chromatogram, as measured by refractive index (RI) detection, of the reaction mixture containing DMA derivatized cellobiose. By assuming a Gaussian shape of the analyte peak it could be calculated that no less than 90% of this substance could be recovered before the unwanted components started eluting (at ∼9.3 mL). Methyl and hydroxypropyl substituents are expected to increase the hydrodynamic radii of the oligosaccharides. Therefore, it was assumed that for the DMA-derivatized cellulose ethers >90% of the components with DP >1 were recovered. Provided that the DEA and DPA derivatives are separated by the same mechanisms as the DMA derivatives, i.e., without any apparent interactions with the stationary phase, the recovery of those can also be expected to be >90%. Cellooligosaccharide Mixture. Figure 2 shows MALDI mass spectra of a commercial mixture of cellooligosaccharides (Figure 2a) and the corresponding DMA derivatives (Figure 2b). Underivatized cellooligosaccharides form predominantly [M + Na]+ in MALDI. For the DMA derivatives, on the other hand, both [M + H]+ and [M + Na]+ are observed. This can be explained by the fact that the DMA derivatives contain a basic functionality. Therefore, the affinity for protons is higher than for the underivatized oligosaccharide. Oligomers in the DP range 2-7 were detected in MALDIMS of the underivatized cellooligosaccharides (Figure 2a). For the DMA derivatives (Figure 2b), the detected DP range was 1-6 for [M + H]+ and 2-7 for [M + Na]+, respectively. Hence, glucose could only be detected for the DMA derivatives. Apparently, the sensitivity for glucose is significantly higher for the DMA derivatives than for the underivatized oligosaccharides. Moreover, the analyte signal
Figure 3. Relative peak areas in MALDI mass spectra of (a) underivatized cellooligosaccharides dried at reduced pressure, (b) DMA-derivatized cellooligosaccharides dried at reduced pressure, and (c) DMA-derivatized cellooligosaccharides dried in ambient atmosphere. The empty bars correspond to the relative peak areas for [M + Na]+ and the filled bars to the relative peak areas for [M + H]+.
intensities and signal-to-noise ratios in the mass spectra (Figure 2) also indicate that the overall sensitivity in the studied DP range increased by approximately a factor of 3. From Figure 3b, it can be deduced that for the DMA derivatives the [M + H]+/[M + Na]+ peak area ratio is highest at DP 1 and then decreases with increasing DP. For DP g 7, no peaks originating from [M + H]+ could be detected in spectra accumulated from samples with DHB that had been prepared by drying at reduced pressure. Apparently, protonation of the DMA derivatives in MALDI is more favorable than sodium adduct formation for DP < 3 for this preparation method. Interestingly, in spectra that were accumulated from certain positions in samples that had been prepared by drying in ambient atmosphere, the peak areas of [M + H]+ were larger than the corresponding [M + Na]+ for all DPs detected (DP e 7, Figure 3c). The analyte distribution in these samples was, however, highly heterogeneous.26,27 Therefore, such spectra could not be easily reproduced. The reason for the high [M + H]+/[M + Na]+ peak area ratios is unclear, but the preparation procedure plays a significant role. Equimolar Oligosaccharide Mixtures. The variation in sensitivity with varying DP in MALDI-MS of DMA deriva-
2796
Biomacromolecules, Vol. 6, No. 5, 2005
Momcilovic et al.
Table 1. Relative Peak Areas for Equimolar Mixtures of Oligosaccharides and DMA Derivatized Oligosaccharidesa cellooligosaccharides DP
underivatized
DMAderivatized
1 2 3 4 5 6 7
0.07 0.2 0.5 0.7 1.0
1.0 0.8 0.7 0.6 0.3
maltooligosaccharides underivatized
DMAderivatized
0.0 0.04 0.2 0.4 0.6 0.8 1.0
1.0 0.6 0.6 0.6 0.5 0.4 0.4
a For the DMA derivatives, the peak areas from [M + H]+ and [M + Na]+ were summarized, whereas for underivatized oligosaccharides, only [M + Na]+ was quantified.
tives was studied using equimolar mixtures of cellooligosaccharides (DP 1-5) and maltooligosaccharides (DP 1-7). To avoid selective losses of analyte components, the reaction mixtures were analyzed by MALDI-MS after the derivatization without any further purification except for the evaporation of DMA. Table 1 shows the relative peak areas for the underivatized and the DMA-derivatized oligosaccharides. It is clear that, for cellooligosaccharides, the relative peak area increases with increasing DP up to at least DP 5. For maltooligosaccharides, this effect is observed up to at least DP 7. This is in accordance with earlier findings suggesting that oligosaccharides with low DPs have low sensitivity in MALDI-MS.28 For the DMA derivatives the trend is different. Here the relative peak areas indicate that the sensitivity decreases with increasing DP. Judging from Table 1, the ratio between the highest and lowest sensitivities for the DMA derivatives is ∼3. This is a significant improvement compared to for the underivatized oligosaccharides were the ratio is >10. One can argue that the formation of [M + H]+ is suppressed in MALDI of the equimolar DMA derivatives due to the high concentration of Na+ in the sample. Therefore, the relative peak areas for DP 1 and 2, for which there seems to be a preference for protonation (cf. the commercial nonequimolar cellooligosaccharide mixture shown in Figure 3), could be expected to be even higher for a desalted (by SEC) equimolar mixture. Unfortunately, desalting by SEC would ruin the equimolarity and no further information will be obtained than already given by the commercial cellooligosaccharide mixture. Nevertheless, it is clear that for DMA derivatives the discrimination in MALDI is significantly lower than for underivatized oligosaccharides. In a recent study an equimolar mixture of maltooligosaccharides with DP 1-7 was subjected to derivatization with Girard’s T reagent (GT).21 It was found that, in MALDI-MS of the GT derivatives, the discrimination of oligosaccharides with low DP was significantly reduced. The same trend was observed for spectra collected from sample spots dried both at reduced pressure and in ambient atmosphere. Partially Depolymerized Cellulose Ethers. Figure 4 shows MALDI mass spectra of acidically depolymerized MC (Figure 4a), DMA-derivatized MC (MC-DMA, Figure 4b), DEA-derivatized MC (MC-DEA, Figure 4c), and DPAderivatized MC (MC-DPA, Figure 4d). For all dialkylamine derivatives, the [M + Na]+ signal intensities were at least
2-3 times higher than for the underivatized MC. Hence, dialkylamine derivatization of partially depolymerized MC increases the sensitivity in MALDI-MS, which is in accordance with what was observed for the cellooligosaccharides. Moreover, for underivatized MC (Figure 4a), the relative intensities of the peaks originating from molecules with DP 1 were comparably low and no peak corresponding to unsubstituted glucose was detected. For MC-DMA and MC-DEA, on the other hand, the peaks with the highest relative intensities were those originating from molecules with DP 1. Peaks originating from unsubstituted glucose were also observed. Clearly, the sensitivity increased dramatically for DP 1 for MC-DMA and MC-DEA. Surprisingly, for MC-DPA the relative intensities of the peaks originating from molecules with DP 1 were lower than for MC-DMA and MC-DEA. This result was unexpected and the reason is unclear. A probable explanation could be that the recovery of DPA derivatives with DP < 3 from the SEC purification was lower than for MC-DMA and MCDEA. Attractive interactions, e.g., hydrophobic or charge interactions, with the column stationary phase could cause an increase of the elution volumes for the DPA derivatives compared to the other derivatives. If this is the case, the preferred reagent of the three will be DMA since it is least hydrophobic and can be expected to be the weakest base. The [M + H]+/[M + Na]+ peak area ratios for the different derivatives can be seen in Figure 5. It is clear that in MALDI the lowest [M + H]+/[M + Na]+ ratios were obtained with MC-DMA. Data also indicated that MC-DPA produced the highest relative intensities of [M + H]+. Thus, the overall trend was that the relative amount of [M + H]+ in MALDI increased with the size of the alkyl group. A possible explanation for this could be the basicity of the dialkylamine functionality, which increases (pKb(DMA) > pKb(DEA), pKb(DPA) not found) with increasing size of the alkyl groups. MALDI mass spectra of MC-DMA, MC-DEA, and MCDPA accumulated from samples with DHB prepared by drying in ambient atmosphere resembled those obtained for the DMA-derivatized cellooligosaccharides. The relative intensities of the [M + H]+ and [M + Na]+ peaks varied significantly between different positions in the samples. For some positions the spectra were similar to those obtained from samples dried at reduced pressure, whereas on other positions spectra were obtained where only [M + H]+ peaks were observed. Thus, also for MC the sample preparation method has an influence on the relative yield of various ion types in MALDI. The suitability of CHCA as MALDI matrix for dialkylamine derivatives was also investigated. With this matrix the [M + H]+/[M + Na]+ ratios were significantly higher than those obtained with DHB. Judging from the even signal intensities, the analyte also appeared to be homogeneously distributed in the sample. However, the absolute signal intensities were only about 20% of those obtained with DHB. Therefore, for applications where high intensities are required such as tandem MS, this matrix seems less appropriate than DHB. Arabinosazone was also studied and with this matrix substance only peaks from [M + Na]+ could be observed.
Reducing-End Derivatization of Cellulose Ethers
Biomacromolecules, Vol. 6, No. 5, 2005 2797
Figure 4. Reflector-mode MALDI mass spectra of (a) underivatized partially depolymerized MC, (b) DMA-derivatized MC, (c) DEA-derivatized MC, and (d) DPA-derivatized MC. DHB was used as matrix and H2O as solvent, and the samples were dried at reduced pressure. Peaks are indicated as in Figure 1 with the addition that the superscripts indicate the number of methyl substituents. Peaks originating from molecules with DP 1 are indicated with an asterisk.
Figure 5. Peak area ratios, from MALDI mass spectra, between [M + H]+ and [M + Na]+ at different DPs for (from left) MC-DMA, MCDEA, and MC-DPA. The error bars indicate the standard deviation (n ) 4).
Although the absolute signal intensities with this matrix also were significantly lower than those obtained with DHB, employment of arabinosazone may in some cases be beneficial since the complexity of the mass spectra decreases. Figure 6 shows MALDI mass spectra of enzymatically depolymerized HPC-DMA (Figure 6a) and HPMC-DMA (Figure 6b). Peaks originating from glucose and monosubstituted glucose molecules were observed for both cellulose ethers. For HPC-DMA, glucose molecules with up to 1 hydroxypropyl substituent were detected. In the HPMCDMA spectrum, peaks from glucose with