Polymer analytics: SEC-RI-ESI-MS as an alternative to SEC-RI-MALLS
The first step towards a simultaneous analysis of chitosan polymers and oligomers via SEC-RI-ESI-MS was to establish our setup as an alternative to SEC-RI-MALLS to determine the average MW of polymers based on newly established regressions. Figure 2a shows the relationship between the Mw determined via SEC-RI-MALLS and the RT of chitosan polymers and oligomers as well as pullulan standards in the SEC-RI-ESI-MS setup for two SEC columns of different pore size. During their flow through the column of porous beads, smaller oligomers can enter the pores and take a longer route, hence eluting later than larger polymers. When comparing the performance of the two columns, the separation is most efficient around 1 kDa or 80 kDa for the column with 125 Å or 200 Å pore size, respectively, which fits to the optimal MW ranges indicated by the supplier. As a consequence, resolving larger chitosan oligomers (e.g. the Kitostim sample) as well as the pullulans of Mw 0.5-2 kDa into individual peaks was only possible using the 125 Å column, resulting in more data points. Moreover, analytes elute in general faster from the 125 Å column, as expected from its smaller pore size. When comparing different analytes, the pullulan standards elute always later than corresponding chitosans of a similar Mw. This can be explained by the more globular structure of pullulans with their mix of α-1,4 and α-1,6 linkages31 in contrast to the linear and rather stiff chitosans, making the former appear smaller during SEC. Consequently, the connection between Mw and RT is different for different polysaccharides and the calculation of Mw values always needs to be performed using regressions that are based on measurements of the same polysaccharide species. With the y-axis in logarithmic scale, exponential relationships between Mw and RT of chitosans appear as straight lines in the plot. It is not possible to find a suitable exponential equation to connect all Mw values to a RT, but for certain Mw ranges, a common straight line is reasonable: the low Mw range from 0.1-1 kDa, the medium Mw range from 1-10 kDa, and the high Mw range from 10-1000 kDa. The set of polymeric chitosans (HMC samples) in the high Mw range from Figure 2a is depicted as a close-up in Figure 2b. Whereas samples of higher Mw exhibit low standard deviations and fit well to the regression, samples of low Mw deviate more strongly. It needs to be mentioned that due to the low Mw, the MALLS signals of these samples are weak which leads to inaccuracy and high deviations in the SEC-RI-MALLS results. Based on the R2 values of the exponential regressions, the 200 Å column is more suitable for the analysis of chitosan polymers, although both columns give reliable results. An exponential relationship was also established between Mn and RT for both SEC columns with R2 values above 0.96 (Supplementary Fig. S1).
The obtained exponential equations are used to convert signals from measurements of chitosan polymers in the SEC-RI-ESI-MS setup to the corresponding Mw. Figure 2c shows the Mw distribution obtained by SEC-RI-MALLS of five selected samples; these were also measured via SEC-RI-ESI-MS (200 Å column) and based on the corresponding exponential equation, a second x-axis with the Mw was added to the chromatogram (Fig. 2d). As intended, both methods give similar results, but some differences are visible: On the high Mw end above 500 kDa, the SEC-RI-ESI-MS setup is not suitable for efficient separation as samples 75/1000 (Mw: 404 kDa) and 75/3000 (Mw: 535 kDa) elute simultaneously even on the 200 Å column. Moreover, the Mw range above 500 kDa was not covered when establishing the regressions. In contrast, SEC-RI-MALLS shows different results for the two samples as expected from viscosity data of the supplier, indicating a better performance of this method when analyzing very large polymers. However, for small polymers like 75/5 (Mw: 66 kDa), it is difficult to get meaningful results on Mw and Mn via SEC-RI-MALLS because the intensity of the MALLS signal decreases with decreasing MW. On the contrary, SEC-RI-ESI-MS separates small polymers well and is just dependent on the RI signal, making it the better solution on the lower Mw end. Not unexpectedly, it becomes apparent that considering only the single value for the average MW is a strong simplification when characterizing a chitosan sample – it is clearly important to always consider the Mw or Mn distributions from SEC-RI-MALLS or SEC-RI-ESI-MS as well.
As mentioned above, SEC-RI-MALLS should still be the method of choice when analyzing very large polymers, at least when compared to the SEC columns used in the SEC-RI-ESI-MS setup used here, and it is obviously required to initially generate the regression describing the relationship between RT in SEC-RI-ESI-MS and MW of one type of analyte. Once this is established, using SEC-RI-ESI-MS has certain advantages over SEC-RI-MALLS: First, the amount of sample is multiple times lower and the separation times are much shorter, in our case 3 µg sample and 14 min for SEC-RI-ESI-MS compared to 100 µg sample and 90 min for SEC-RI-MALLS. Second, contaminations or degradation products can directly be identified via MS. And third, the data analysis is less subjective, because the RI peak maximum is used to calculate the sample’s Mw, whereas in SEC-RI-MALLS, the calculation is based on the manual definition of the RI and MALLS peaks’ integrals. Therefore, SEC-RI-ESI-MS should be included into the general toolbox for chitosan polymer analytics as an alternative to SEC-RI-MALLS. But more importantly here, the setup forms an integral part of the live digestion method described below (results section “Live digestion”), addressing the polymer part of the sample.
Oligomer analytics: Prerequisites for semi-quantitative analysis
The second building block of this live digestion method is a simplified, quantitative oligomer analysis. Crucial for this is an adequate separation of COS with the same SEC columns used above for polymer MW determination. As discussed above, the results shown in Figure 2a for COS (Kitostim) and pullulan oligomers indicate a better performance in oligomer separation for the SEC column of pore size 125 Å compared to 200 Å, while maintaining reliable results in polymer analytics. Therefore, the 125 Å column was used in all further experiments. We were able to show that the efficiency of oligomer separation is not impaired by increased flow rates during chromatography (Supplementary Fig. S2). Hence, separation times can be reduced by increasing the flow rate as far as the pressure limit of the SEC column allows without loss of performance.
As mentioned in the introduction, the current main problem in the quantification of different COS are their different signal response factors in MS which in turn derive from their different ionization efficiencies during ESI. To work around this in the currently used quantitative MS sequencing method23, all D-units of COS are N-acetylated with isotope-labeled acetic anhydride before quantitative analysis. This derivatization turns non- or partially acetylated COS into chitin oligomers while still being able to distinguish between original and reacetylated GlcNAc units by MS due to the isotope label. Combined with the addition of defined amounts of double isotope-labeled standards of different DP, this allows for an absolute quantification of each COS of a certain DP and FA via MS. But the derivatization and the use of standards introduce limits into the analysis of oligomers: Because double isotope-labeled standards are only easily available up to DP 6 and even insoluble above DP 8, larger COS cannot be quantitatively analyzed using this method. Moreover, the full N-acetylation requires prior removal of remaining polymeric substrate and enzyme, making a simultaneous analysis of polymers and oligomers within one sample, or sampling live from an ongoing reaction impossible.
In the following, we describe how the combination of MS and RI signals paves the way towards a semi-quantitative analysis of chitosan oligomers without filtration and derivatization. This is only possible if three prerequisites are met: First, different oligomers of the same DP elute simultaneously, second, the RI signal intensity is only dependent on mass concentration, and third, the ionization efficiency of oligomers of one DP in the ESI-MS is similar. The first prerequisite is ensured by a high ammonium acetate concentration in the SEC solvent as shown in the supplementary information (Supplementary Fig. S3). In the following, the other two prerequisites are addressed.
At a constant temperature, the RI shows a linear correlation with a compound’s mass concentration up to high concentrations32. This is commonly used in SEC-RI systems with constant temperature, solvent composition, and flow rate for one analyte species to derive the relative mass concentration of the analyte from the RI peak integral. Indeed, with constant instrument parameters, the RI signal response is mainly dependent on the mass concentration, but also the MW and response factor of the analyte (dn/dc value) play a role33, resulting in an influence of the chitosans’ DP and FA. Even though this is neglected in most SEC-RI applications on chitosans anyway, we confirmed that neither the DP (Fig. 3a) nor the FA or PA (Fig. 3b) have a systematic or considerable influence on the RI signal response of oligomers. Hence, a dependency of the RI signal intensity on the chitosan oligomer mass concentration alone can be assumed.
Like RI, also the signal response of chitosan oligomers in ESI-MS can differ dependent on the factors DP, FA, or PA because of the different ionization efficiencies during ESI. On the one hand, the ionization efficiency is vastly different for oligomers of different DP values (Fig. 3c): oligomers of DP 3 ionize best and are therefore overrepresented in the MS signal intensities when compared with other DP values. On the other hand, oligomers of the same DP but with different FA and PA values feature similar ionization efficiencies, as shown in Figure 3d.
The RI response factors allow the relative quantification of oligomers of one DP eluting in a single RI peak compared to oligomers of other DPs based on the integrals of the RI signal peaks. The MS response factors enable the relative quantification of oligomers of one FA compared to oligomers of other FAs based on the MS signal intensities, but only within one DP. A distinction between oligomers with the same DP and FA but differing PA is not possible here. Combining both signals makes it possible to quantify each oligomer of a certain DP and FA relative to all others. In comparison to quantitative MS sequencing23, the combination of RI and MS proposed here indeed results in a less accurate quantification because it is based on the integration of RI peaks, because it does not feature absolute quantifications with internal standards, and most importantly, because the ionization of oligomers of the same DP with different FA is similar, but not identical – that is why we call it “semi-quantitative analysis”. However, our developed method is a unique way to easily quantify oligomers of defined DP and FA up to DPs with sufficient signal strength (typically at least up to DP 10) while simultaneously analyzing polymers within the same SEC-RI-ESI-MS run. Considering in addition that no further sample preparation is needed, this allows an easy and fast live analysis of chitosan-cleaving enzymes, as described in the following.
Live digestion: Fast and easy analysis of an enzyme’s action
The newly established methods for the analysis of polymers and oligomers via SEC-RI-ESI-MS were finally combined into a live digestion method for the fast and easy characterization of chitosan-cleaving enzymes. A single injected sample measured with the SEC-RI-ESI-MS setup already gives information on the amount, composition, and average FA of oligomer products as well as on the amount and estimated Mw of the remaining polymer substrate. But to create a full picture of the enzyme’s activity on the substrate, it is necessary to monitor all these parameters over time. Because our method requires no sample preparation at all, the samples containing substrate, enzyme and buffer can be injected directly. That makes it possible to sample automatically over time from a single vial by using the autosampler. As shown in Figure 1, the enzymatic digestion is started by adding the enzyme and from this timepoint on, the autosampler takes live samples from the enzymatic digestion and injects them directly into the SEC-RI-ESI-MS system.
All data generated from a single vial with an enzymatic digestion are summarized in Figure 4, representative for the digestion of a chitosan polymer (651, FA: 0.22, Mw: 130 kDa) with the chitosanase CsnMN, sampled live over 300 min. We used DP 10 as the cutoff between small oligomers on the one hand, and polymers and large oligomers on the other hand, because even at low concentrations of products of DP ≤ 10, distinct RI peaks are formed.
First, one should consider the RI chromatograms (Fig. 4a, Fig. 5) which allow to directly follow the cleavage of the polymers into oligomers. Whereas the substrate contains just polymers as expected, a considerable part was already hydrolyzed after 45 min to smaller polymers (shift of the peak towards later RT), large oligomers (tailing peak) and small oligomers (distinct peaks). Because an exo-enzyme would produce only small oligomers directly from the large polymer substrate, this indicates an endo-cleaving activity which has been reported before for CsnMN27. After 120 min, the majority of the polymers was hydrolyzed and substantial amounts of small oligomers of DP 2-6 were produced. When comparing the chromatograms at timepoints 120 min and 300 min, it becomes apparent that the enzyme degrades already small oligomers of DP 5 and 6 even further. The reaction was far progressed after 300 min, but considering the remaining polymers and large oligomers, probably not yet at its end point. To find the end point, one needs to just continue the ongoing live digestion until the RI chromatograms do not show changes anymore. Then it is useful to add fresh enzyme and take another sample after additional incubation time, to check if the reaction before had stopped due to activity loss of the enzyme, or if the real end point was reached. There, only substrates remain that do not fit to the enzyme’s subsite preferences or required length, or the reaction has come to a standstill due to product inhibition.
Based on the relationship between Mw and RT established by the regression in section “Polymer analytics” (Fig. 2b), the Mw range of the polymers can be estimated as shown in Figure 4b by introducing a second x-axis. To plot the cleavage of the polymers over the full reaction time, the maximum of the polymer RI peak is converted to the corresponding Mw (Fig. 4c). Both figures show the hydrolytic cleavage of the polymers over time: Whereas the non-digested substrate features a calculated Mw of about 120 kDa (SEC-MALLS-RI of the substrate 651 had given 134 kDa, Supplementary Table 1), the Mw of the RI maximum decreases over the course of the reaction – at first fast, then more slowly until it barely changes anymore. At the same time, the polymer RI peak integral is reduced drastically between start of the reaction and 300 min, indicating an efficient cleavage of the polymers.
The mass concentration-dependent integrals of RI signals can further be used to quantify the enzyme products of distinct size. Figure 4d shows the proportion of polymers and large oligomers (DP > 10) compared to small oligomers, with the latter fraction increasing and the former fraction decreasing linearly during the first 120 min. At this timepoint, half of the substrate was cleaved into small oligomers of DP ≤ 10. Afterwards, the curves level out, either because the enzyme lost its activity over time or because the reaction came close to its end point. Naturally, the same trend is visible when only plotting the amount of the small oligomers in micrograms (Fig. 4e), obtained by multiplying the proportion of small oligomers per timepoint with the total amount of chitosan per injection, in this case 3 µg.
In a second data analysis step, the MS and RI data are combined. The obtained dimensionless amounts of each oligomer of a certain DP and FA can be plotted for each timepoint as oligomer product profiles (Fig. 4f, Fig. 6a) which give insights how the product composition changes over time. At the earliest timepoint of 15 min, all products are fully deacetylated (D3-D9). Over the course of the reaction, two effects can be derived from the product profiles: First, the large fully deacetylated oligomers D6-D10 were further cleaved into even smaller oligomers, resulting in a high share of DP 2-5 products at late timepoints. Second, the proportion of products above DP 5 with one or two acetyl groups increases, for example A1D5, A1D6 or A2D7. This fits exactly to previous reports on CsnMN; the enzyme preferentially cleaves fully deacetylated sites of the substrate due to its high preference for D-units at subsites -3 to +327, resulting in fully deacetylated products. After these sites are hydrolyzed in the initial reaction phase, CsnMN binds increasingly to less acetylated sites, because some of the enzyme’s subsites also accept A-units, but with lower affinity18. The resulting partially acetylated products are again no optimal substrates and, thus, accumulate. This increasing acceptance of acetylated units is also visible in Figure 4g where the average FA of the small oligomer products calculated from the product profiles is plotted over time. Based on the weight percentages of each oligomer shown in Figures 4f and 6a, and the absolute amounts of oligomer products shown in Figure 4e, the total amount of individual oligomers can be followed over time (Fig. 4h, Fig. 6b). For example, D8 and D9 both increased in the beginning, but were further cleaved into smaller oligomers over the course of the reaction. The longer the fully deacetylated substrate, the more likely the enzyme cleaves; hence, the amount of D9 decreased earlier than that of D8. As for the product profiles and the average FA of oligomer products, the time courses of individual oligomers show an increasing acceptance of A-units. For DP 8, A1D7 started to accumulate after 60 min and even A2D6 was produced after 180 min. The longer the oligomer, the more likely is the presence of A-units in the product. Therefore, it is not surprising that both A1D8 and A2D7 were formed already after 60 min. A1D8 was even further degraded as indicated by its decrease after 240 min.
Overall, our results support multiple conclusions: CsnMN is an endo-acting chitosanase with a strong preference for D-units, as described before27. Nevertheless, the enzyme tolerates A-units, especially once the strongly deacetylated parts of the substrate are already cleaved. This fits to previous findings of Weikert et al.18 that show a nearly absolute specificity for D-units at subsites -2 to +2 for initial timepoints of the reaction, whereas A-units are accepted at subsites +1 and +2 only towards the end point of the reaction for more highly acetylated substrates. In contrast, the strong specificity for D-units at subsites -1 and -2 is maintained throughout different substrates and reaction times18,26,27,30. Because no partially acetylated oligomers of DP < 5 are produced, most A-units are accepted at subsites like -3 or +3 rather than close to the catalytic site. Even though after 45 min still less than 20% of the sample was converted to smaller oligomers of DP ≤ 10, the Mw of the polymer peak was reduced drastically. All in all, our proposed live digestion method gives detailed insight into the enzyme based on a single enzymatic digestion.