Enzyme selection for synthesis of 1-thio-cellulose
The synthesis of 1-thio-cellulose from 1-thio-β–d-glucose via a cascade reaction of CuCbP and CcCdP involves ∼40% plain cellulose, which is formed as a byproduct [32]. Glucose is required to build plain cellulose, and the αGlc1-P donor substrate is its source. The CcCdP used shows low hydrolase activity against αGlc1-P (20.7 ± 0.4 mU/mg). CuCbP forms cellobiose from the substrate glucose (32.1 U/mg), which shows a 51-fold increase in enzymatic production compared with 1-thio-β–d-glucose (0.63 U/mg). Cellobiose is an excellent substrate of CcCdP (13.3 U/mg), while 1-thio-β–d-glucose (0.008 U/mg) is not. The β-1,4-disaccharide of 1-thio-β–d-glucose may be a better substrate than 1-thio-β–d-glucose, although it is unlikely to be used by CcCdP at a similar activity level as cellobiose. Therefore, limited synthesis precision of the enzymatic polymerization from 1-thio-β–d-glucose arises from “hydrolytic errors” made by CcCdP in the utilization of αGlc1-P [32]. Considering the clear benefit for 1-thio-cellulose synthesis from the availability of a high-fidelity CdP, we searched the sequence database and the relevant literature reports for alternative candidate enzymes from this class.
Sequence-based categorization of biochemically characterized CdPs in the form of a phylogenetic tree (Fig. 1a) shows that the CdP from C. thermocellum (CtCdP; GenBank ID, BAB71818) is quite similar (75% sequence identity) to the CcCdP used in our earlier study [32]. CtCdP has been reported by several groups for the synthesis of cello-oligosaccharides and cellulose [21, 22, 35]. Considering that similarity in the sequence often translates into similarity in the biochemical properties, we focused on enzymes more distant from CcCdP than CtCdP. Among the candidates, CdP from C. stercorarium (CsCdP) was promising, particularly because it shares only 23% sequence identity with CcCdP and is permissive with respect to the acceptor substrate used. β–d-Glucosides with aromatic (e.g., 4-nitrophenol) and aliphatic aglycones (e.g., hexanol) are iteratively glycosylated from αGlc1-P to form the corresponding glycosidic cello-oligosaccharides [36]. The CdPs from Thermosipho africanus TCF52B (TaCdP; GenBank ID, ACJ76363.1) and Ruminococcus albus (RaCdP; GenBank ID, ADU22883.1) are also distantly related to CcCdP (Fig. 1a). However, a biochemical characterization of these enzymes [37, 38] did not show activity with a glycoside acceptor. Thus, CsCdP was the preferred choice for further study.
Phylogenetic analysis of CdPs and isolation of CsCdP. a Phylogenetic tree of selected CdPs previously characterized biochemically. Multiple-sequence alignment was performed using ClustalX (http://www.clustal.org/). CdP gene origin: Ct, C. thermocellum (also known as Ruminiclostridium thermocellum); Cc, C. cellulosi; Ta, T. africanus TCF52B; Ra, R. albus; Cs, C. stercorarium; Ce, C. cellulovorans (GenBank ID, BAV13068.1). b SDS polyacrylamide gel showing purification of CsCdP. Lane Ori shows the original enzyme stock after His-tag purification. The other lanes show the enzyme after heat treatment at the indicated temperature and time (in minutes (min))
Characterization of CsCdP
Recombinant production of CsCdP in E. coli was previously reported [36], although a few points are worth noting here. The native CsCdP gene (GenBank ID, U60580.1) was expressed at a rather low level when the plasmid vector pTrc99A was used in E. coli XL10 Gold [36]. Using codon optimization, the gene expression in E. coli BL21(DE3) with plasmid vector pET-21b(+) was enhanced by a large amount, although most of the produced CsCdP (~85%) was found in inclusion bodies. The tendency of CsCdP to become insoluble was retained under the various conditions of expression examined (Supplementary Fig. S3). An analysis of the cell extract by SDS polyacrylamide was performed to estimate the maximum production yield of soluble CsCdP, which was 8 mg/L bacterial culture. CsCdP is equipped with a His tag at its N-terminus. Supplementary Fig. S2 shows the full amino acid sequence of the enzyme used. Purification led to enrichment of CsCdP in the enzyme preparation, although contaminating proteins remained in substantial amounts (Fig. 1b). Considering that CsCdP is heat resistant (half-life at 60 °C: ∼30 h [39]), we used temperature-induced denaturation of the less thermostable E. coli proteins as a means of further purification of CsCdP (Fig. 1b). The yield of purified CsCdP was ∼6 mg/L of culture. CsCdP did not lose activity during the heat treatment, suggesting that the purified enzyme should be of good quality despite the thermal stress used. The specific activity of the purified CsCdP for the reaction with cellobiose and αGlc1-P (5.1 U/mg) was comparable with that from an earlier study of the enzyme (9.2 U/mg; [39]). This result is also reasonably in line with the specific activities of other CdPs assayed under relevantly comparable conditions (CcCdP: 13.3 U/mg [19]; CtCdP: 5.5 U/mg [40]; RaCdP: 20.4 U/mg [38]).
Figure 2 compares the relevant specific activities of CsCdP with those of CcCdP and CuCbP. Although CsCdP was 2.6-fold less active than CcCdP on cellobiose, it retained considerably higher activities than CcCdP on glucose (11%; 8-fold) and 1-thio-β–d-glucose (3.4%; 21-fold) relative to cellobiose. The specific activity of CsCdP with 1-thio-β–d-glucose thus surpassed that of CcCdP by 21-fold. The presence of the β-thio group diminished the activity of CsCdP (3.35-fold) less than that of CcCdP (8.8-fold) compared to the α/β-hydroxy group in d-glucoside/d-glucose acceptors. The activity of CsCdP with 1-thio-β–d-glucose was reasonably comparable (3.7-fold lower) to that of CuCbP.
Activity characterization of a CcCdP, b CsCdP, and c CuCbP: the synthetic activity on glucose and 1-thio-β–d-glucose (thio-glc) and αGlc1-P hydrolase activity (hydro.) are compared to the activity on cellobiose (at 45 °C). Synthetic activity was measured in MES buffer (pH 7.0) containing αGlc1-P (50 mM) and the corresponding acceptor (50 mM). αGlc1-P hydrolase activity was measured in MES buffer (pH 7.0) containing αGlc1-P (100 mM). The values above the columns indicate the specific activity (U/mg) of the corresponding enzyme
Most importantly, CsCdP showed hydrolase activity toward αGlc1-P that was approximately 11-fold lower than that of CcCdP. We confirmed the result of our earlier paper that CuCbP was inactive toward hydrolyzing αGlc1-P within the limits of detection [32]. Comparing the transfer activity with 1-thio-β–d-glucose to the hydrolase activity with αGlc1-P, CsCdP involved a reaction selectivity of 85 (= 0.17/0.002), which was 224-fold greater than that of CcCdP (0.38 = 0.008/0.021). The use of CdP enzyme selection for a direct (i.e., single-enzyme-catalyzed) iterative β-1,4-glycosylation of 1-thio-β–d-glucose with very high product selectivity was strongly supported by these results. The activity data (Fig. 2) imply that the enzyme cascade with CuCbP can provide only a modest benefit for the conversion of 1-thio-β–d-glucose. A simpler reaction with only a single enzyme (CsCdP) might suffice (Scheme 1b). The transfer/hydrolase selectivity of CsCdP suggests that the formation of plain cellulose should be minimal in the synthesis of cello-oligosaccharides from 1-thio-β–d-glucose. Differences between CsCdP and CcCdP in reactivity might be due to structural changes in the acceptor binding pocket, namely, the +2 subsite, which binds the reducing-end glucose residue of cellobiose, as discussed in other papers [36, 41, 42].
Enzymatic synthesis of 1-thio-cellulose
Cellulose synthesis was performed using 100 mM αGlc1-P and 10 mM 1-thio-β–d-glucose. The concentrations of donor and acceptor substrates were inspired by earlier work [32], in which insoluble product formation was favored under these conditions. Note that the glucosyl units transferred from the αGlc1-P substrate are distributed between soluble and insoluble oligosaccharide products. A high donor/acceptor ratio shifts the distribution toward the desired insoluble material. The reactions of CsCdP alone or in combination with CuCbP were compared to a reference reaction of CcCdP and CuCbP.
As shown in Fig. 3, all reactions involved a whitish precipitate formed with time. Importantly, the blank reaction in the absence of 1-thio-β–d-glucose did not show the precipitate or only showed a slight amount when CsCdP was used with or without CuCbP. In contrast, the blank reaction of the reference produced a substantial amount of precipitate. This finding was consistent with the expectation and supported by the activity data in Fig. 2, which showed that the hydrolysis of αGlc1-P catalyzed by CcCdP promotes cellulose formation via glucose release (Supplementary Fig. S1). These results also imply the formation of 1-thio-cellulose in higher purity when CsCdP is used instead of CcCdP.
Photographic images showing the reaction mixture of the single enzyme (CsCdP) and coupled enzyme (CbP–CdP) reactions. Synthesis reactions containing 100 mM αGlc1-P and 10 mM 1-thio-β–d-glucose in MES buffer (50 mM, pH 7.0) were carried out at 45 °C for 8 h (CsCdP reaction) and 12 h (CbP-CdP reaction). Blank reactions were performed using otherwise exactly identical conditions but lacking 1-thio-β–d-glucose
The αGlc1-P conversion in the synthesis reactions was measured from the released phosphate. Reactions of CsCdP proceeded at a similar rate when CuCdP was absent or present. The conversion yield reached ∼25 mol% after 8 h, which leaves room for optimization. The equilibrium for polymerization from αGlc1-P is complex when an insoluble product is formed. Generally, from the reaction in solution analyzed in detail for CcCdP, one would expect that αGlc1-P conversion of ∼60 mol% should be thermodynamically possible. By comparison, the conversion yield of the CuCbP-CcCdP reference reaction was ∼41 mol%.
Structural characterization of the thio-cello-oligosaccharide chains
The insoluble product of enzymatic synthesis was analyzed by MALDI-TOF-MS. As shown in Fig. 4, the mass spectra showed a series of peaks ([M + Na]+) with peak-to-peak mass differences of 162 Da, which is consistent with the mass of a single glucosyl unit in cellulose chains [35, 43]. The product from the CuCbP-CcCdP reference reaction exhibited two main groups of peaks assigned to the cello-oligosaccharides in the DP range 6–11, and they contained and lacked a single thiol group (Fig. 4a). The product from the CsCdP reaction without CuCbP (Fig. 4b) or with CuCbP (Fig. 4c) showed peak clusters assigned to 1-thio-cellulose chains of DP of 7–17 at an estimated purity of ∼95%. Although plain cello-oligosaccharides were detectable (Fig. 4), their abundance was low (≤5% of total product; single CsCdP reaction). The central idea of this study, that a purer 1-thio-cellulose product can be synthesized by selection of a CdP more suitable than CcCdP, was confirmed by this evidence. Of note, the peak cluster at each DP, with masses differing in m/z by +16/32/48/64 units from the corresponding plain cellulose chain (*), was assigned to different oxidation states of the 1-thio-cellulose chain. As shown elsewhere for the general case of thiol groups in polymers [44], the original −SH group of the cellulose chain can be oxidized to −SOxH under the conditions of the MS analysis, with x varying between 1 and 3. The average polymer DP from the reaction with CsCdP (DPav = 9.7) differed from that of the reference reaction (DPav = 8.1). The longest chain detectable with the MALDI-TOF-MS analysis was larger for the CsCdP reactions (DPmax = 17) compared to the reference reaction (DPmax = 11). The cellulose synthesized by CsCdP exhibited a broader distribution of oligosaccharide DP than the cellulose synthesized by CcCdP (Fig. 4). The difference in the DP distribution of the produced cellulose chains likely reflects the individual specificities of the CdP enzymes used. Kinetic parameters (e.g., maximum rate constant) for each step of glycosyl transfer in the iterative process of polymerization can depend on the DP of the nascent cello-oligosaccharide, which was shown in a recent study of cello-oligosaccharide synthesis by CcCdP [45]. The glycosylation rate dropped substantially (∼3-fold) once the acceptor DP reached a value of 6, thus suggesting an explanation for the particular DP distribution observed. A reaction time course analysis is not available for CsCdP. However, the shift of DP distribution to higher DPav and DPmax compared to CcCdP would imply that CsCdP exhibits a relatively smaller dependence of its activity on the acceptor substrate DP in the relevant DP range from 7 to 17. Previous studies of CdP-catalyzed synthesis of cello-oligosaccharides show that reaction conditions (e.g., molar ratio of donor and acceptor substrate; overall polymerization rate) also affect the DP distribution in the insoluble product. The reaction of CsCdP led to a lower conversion of the αGlc1-P substrate (~25 mol%) than the reference reaction (~40 mol%). The substrate conversion might affect the DP distribution in the insoluble product indirectly via its effect on the concentration and aggregation rate of the growing cello-oligosaccharides in solution [46]. However, an analysis of the CcCdP reaction with cellobiose as the substrate (10 mM; 150 mM αGlc1-P) also gave a DPmax of 11 (data not shown). This result supported the idea that the DP distribution was mainly due to the enzyme used, i.e., CcCdP or CsCdP. The importance of the choice of CdP enzyme was thus supported.
MALDI-TOF MS spectra of cellulose materials synthesized from a CuCbP-CcCdP; b CsCdP; and c CuCbP-CsCdP reaction. At each DP range, the peak of plain cellulose was indicated by the symbol *, and the following clusters (+16/32/48/64 m/z as to the plain cellulose) were assigned to the 1-thio-cellulose and thiol oxidation products thereof: the peak (+16 m/z) marked by the blue line was assigned to cello-oligomer-SH; the peak (+64 m/z) marked by the green line was assigned to cello-oligomer-SO3H; and the area in between were the peaks (+32/48 m/z) of oligomer-SOH/SO2H
The synthesized celluloses were further characterized by 1H-NMR in solution. Solid material dissolved in 4% (w/w) NaOD-D2O was analyzed. The spectra showed signals assigned to the repeating β-glucosyl units of cellulose (Fig. 5) [47]. The dominant doublet at approximately δH 4.30 was assigned to the internal β-1,4 glycosidic linkages (H1) in cellulose [21, 46]. By reference to the acceptor (1-thio-β–d-glucose), the doublet signal at δH 4.35 (J-coupling constant of 8.9 Hz) was ascribed to the anomeric proton (H1’) of the terminal β-1-thio-glucose residue in the 1-thio-cellulose chain. Here, we noticed that the signals at δH 5.1 and 4.5, which are assigned to the α‐ and β‐anomeric protons on the reducing end of plain cellulose [19, 21, 22], were only detected in the spectra of the product from the CuCbP-CcCdP reference reaction (Fig. 5b). No unassigned signals were present in the full 1H-NMR spectra of the product from the CsCdP reaction (Fig. 5a). This result was consistent with the evidence from the mass analysis, thus indicating a highly pure 1-thio-cellulose obtained from the CsCdP reaction. Moreover, based on the signal intensity of the anomeric protons at internal sites (δH 4.30) and at the terminal residue (δH 4.35), a value of 10.3 was determined for DPav in the material from the CsCdP reaction, which is comparable to the value (DPav = 9.7) determined by MALDI-TOF-MS, thus indicating the validity of both analyses. Overall, the results shown confirm the expected chemical structure of the reducing-end thiol-modified cello-oligosaccharides and further support the idea of using CsCdP for precision synthesis of 1-thio-cellulose material.
1H-NMR profile of cellulose materials synthesized from the a CsCdP reaction (profile of acceptor 1-thio-β–d-glucose was included) and b CuCbP-CcCdP reaction. In each panel, the inset showing the enlarged region for anomeric protons was added
Material characterization of the as-synthesized 1-thio-cellulose
XRD and ATR-FTIR were used, and the cellulose from the single-enzyme reaction of CsCdP was analyzed and compared to the sample from the CuCbP-CcCdP reference reaction. Figure 6 shows that the spectral properties of the two celluloses were similar. The XRD patterns of both samples exhibited diffraction peaks (2θ at 12.3°, 20.0°, and 22.1°) that were assigned to the 11̅0, 110 and 020 faces of crystalline cellulose II (Fig. 6a), respectively [48]. The sharp XRD peaks suggest a highly crystalline material. In addition, the ATR-FTIR spectra of both samples showed prominent absorbance bands at 3441 and 3490 cm−1 (Fig. 6b). The two bands are characteristic of the OH-stretching intramolecular hydrogen bonds present in crystalline cellulose II [49, 50]. Overall, therefore, these results indicate self-assembly of the pure 1-thio-cello-oligosaccharides into crystalline cellulose of allomorph II crystal structure. The absence of plain cello-oligosaccharide chains in the synthetic reaction of CsCdP apparently does not affect the recorded characteristics of the material structure in the cellulose obtained. The change in DP distribution in the CsCdP reaction compared to the reference reaction also seems to have no influence on the material properties analyzed. Organization of the synthetic oligomer chains into crystalline material appears to be independent of the CdP enzyme used, which is consistent with the idea of a spontaneous process based on self-assembly-driven chain association. Considering the antiparallel arrangement of the oligosaccharide chains in cellulose II, earlier studies [14, 32] seem to suggest that the functional thiol groups are regularly distributed on the two base planes (surfaces) of the 1-thio-cellulose material. Taken together, the current study demonstrates a highly crystalline material (cellulose II allomorph) composed of pure 1-thio-cellulose chains with an average DP of ∼10. Identification of CsCdP as a specific “cello-oligosaccharide polymerase” was essential to achieve precision synthesis of the material. We expect that access to well-defined 1-thio-cellulose can facilitate further functional diversification of nanocellulose materials in the future.
Structural characterization of the synthetic 1-thio-cellulose: a XRD patterns and b ATR-FTIR spectra. In each panel, the materials synthesized from the CsCdP reaction and the CuCbP-CcCdP reaction are indicated as (i) and (ii), respectively
