PEDS Advance Access published January 26, 2016 Protein Engineering, Design & Selection, 2016, 1–6 doi: 10.1093/protein/gzv066 Short Communication
Short Communication
Engineering ionic liquid-tolerant cellulases for biofuels production Paul W. Wolski1,2,3, Craig M. Dana1,3, Douglas S. Clark1,3,*, and Harvey W. Blanch1,3,* Energy Biosciences Institute, University of California, Berkeley, CA 94720, USA, 2Graduate Group of Comparative Biochemistry at University of California, Berkeley, CA, USA, and 3Department of Chemical and Biomolecular Engineering, University of California, Berkeley, CA 94720, USA *To whom correspondence should be addressed. E-mail: [email protected]; [email protected] Edited by Eva Petersen Received 5 May 2015; Revised 3 December 2015; Accepted 7 December 2015
Abstract Dissolution of lignocellulosic biomass in certain ionic liquids (ILs) can provide an effective pretreatment prior to enzymatic saccharification of cellulose for biofuels production. Toward the goal of combining pretreatment and enzymatic hydrolysis, we evolved enzyme variants of Talaromyces emersonii Cel7A to be more active and stable than wild-type T. emersonii Cel7A or Trichoderma reesei Cel7A in aqueous–IL solutions (up to 43% (w/w) 1,3-dimethylimdazolium dimethylphosphate and 20% (w/w) 1-ethyl-3-methylimidazolium acetate). In general, greater enzyme stability in buffer at elevated temperature corresponded to greater stability in aqueous–ILs. Post-translational modification of the N-terminal glutamine residue to pyroglutamate via glutaminyl cyclase enhanced the stability of T. emersonii Cel7A and variants. Differential scanning calorimetry revealed an increase in melting temperature of 1.9–3.9°C for the variant 1M10 over the wild-type T. emersonii Cel7A in aqueous buffer and in an IL–aqueous mixture. We observed this increase both with and without glutaminyl cyclase treatment of the enzymes. Key words: biased clique, cellulase, directed evolution, ionic liquid, shuffling
Introduction In 2002, Rogers and colleagues identified the room temperature ionic liquid (IL), 1-butyl-3-methylimidazolium (Bmim) chloride, to be an excellent solvent for cellulose (Swatloski et al., 2002). This IL was shown to dissolve up to 10% (w/w) cellulose. When the cellulose suspension in IL was microwaved, up to 25% (w/w) could be dissolved. Subsequently, Dadi et al. (2007) reported that cellulose precipitated from certain ILs by addition of an antisolvent such as water could be readily hydrolyzed by cellulase enzymes. Turner et al. (2003), however, reported that hydrolysis of precipitated cellulose in the presence of the IL 1-butyl-3-methylimidazolium (Bmim) chloride was ineffective, due to denaturation of the cellulases. 1-Ethyl-3-methylimidazolium (Emim) acetate is able to dissolve wood and separate the lignin from the holocellulose (Lee et al., 2009). Ethoxylated ILs with acetate anions (Zhao et al., 2008) were shown to permit enzymatic hydrolysis of the cellulosic content of lignocellulosic
biomass in the presence of the IL. Emim diethylphosphate is able to support cellulase activity at IL concentrations as high as 40% (v/v) (Kamiya et al., 2008). Combining lignocellulose pretreatment by dissolution in ILs with enzymatic hydrolysis in the same vessel offers potential increases in process efficiency. There have been a few other reports of this process, sometimes called one-pot pretreatment and hydrolysis (typically with naturally occurring thermophilic enzymes), with some success (Datta et al., 2010; Gladden et al., 2011; Park et al., 2012; Su et al., 2012). Other studies on selectively removing the glucose and sugars from the IL mixture (Brennan et al., 2010), in situ catalytic conversion to distillable fuels (Chidambaran and Bell, 2010), or even fermentation in the presence of IL (Singer et al., 2011) are contributing toward making such in situ enzymatic hydrolysis a reality. We previously examined protein stability in a variety of ILs using green fluorescent protein as a model for assessing cellulase stability (Wolski et al., 2011). Trichoderma reesei cellulases were shown to
© The Author 2016. Published by Oxford University Press. All rights reserved. For Permissions, please e-mail: [email protected]
1
Downloaded from http://peds.oxfordjournals.org/ at Weizmann Institute of Science on February 1, 2016
1
2
Results and discussion Screening and initial analysis of variants Cel7A variants were selected and purified for further study after screening the activity of each variant toward Emim acetate-pretreated Avicel (see Materials and methods) in aqueous solutions containing 20% (w/w) Mmim DMP, at pH 4.8, and in citrate buffer. The variants were generated via Biased Clique Shuffling (Methods) from the parent Talaromyces emersonii (Te) Cel7A. The variants that were selected all released more sugars from the pretreated Avicel in the presence of 20% (w/w) Mmim DMP than the variant ‘2I13’, described by Dana et al. (2012). When assayed on 4-methylumbelliferyl lactoside (MU-lac) for 10 min as well on Avicel for 22 h the variant 2I13 had an optimum
Fig. 1 Hydrolysis of 2.5 g/L IL-Avicel by Cel7A variant enzymes after 16 h of reaction at 50°C in varying concentrations of Mmim DMP, pH 4.8. The enzyme loading was 0.2 μM. All enzymes were expressed in S. cerevesiae, except T. reesei Cel7A, which was purified from Celluclast. Symbols: T. reesei Cel7A: filled circles; Te Cel7A: open circles; 1G21: open triangles; 2K15: filled triangles; 2I13: filled squares; 1M10: open squares; Te Cel7A QC treated: filled inverted triangles; 2I13 QC treated: open diamonds; 1M10 QC treated: stars. QC indicates treatment with glutaminyl cyclase prior to assay. Dashed lines are shown for the enzymes with N-terminal pyroglutamate (either from adding glutaminyl cyclase or from T. reesei expression host).
temperature (Topt.) of ∼60°C and exhibited the highest residual activity on MU-Lac after incubation at 65°C for 24 h of the mutants studied (Dana et al., 2012). In the present study, two variants (1M10 and 2K15) were also shown to exhibit activity toward MU-lac at 70°C for at least 6 days (data not shown). Of the variants selected (see ‘Selection of enzyme variants’ in Materials and Methods), twelve were expressed for further analysis. After purification, they were assayed for their activity toward Emim acetate-pretreated Avicel in 43% (w/w) Mmim DMP (Fig. 1). Also tested was T. reesei Cel7A, purified from Celluclast. In 43% (w/w) Mmim DMP, two variants, 2K15 and 1M10, produced more glucose than any of the variant or wild-type enzymes tested, including 2I13, 2E10, and 1G21, the most stable variants identified by Dana et al. (2012). T. reesei Cel7A was essentially inactive under this condition. Te Cel7A (expressed in S. cerevisiae) and T. reesei Cel7A ( purified from Celluclast) and variants 1M10, 2K15, 2I13, and 1G21 were assayed in Emim acetate/water solutions under the same conditions as those employed with Mmim DMP. Figure 2 shows the glucose released from the hydrolysis of Emim acetate-pretreated Avicel in 20% (w/w) Emim acetate as well as the glucose released from hydrolysis in buffer. The results are similar to those in 43% (w/w) Mmim DMP, with 1M10 and 2K15 exhibiting the most cellulose hydrolysis in 20% (w/w) Emim acetate relative to buffer. Cellulose hydrolysis is only observed in Emim acetate at up to 30% (w/w) concentration, consistent with our previous finding that Mmim DMP inhibits cellulolytic activity less than Emim acetate (Wolski et al., 2011). In the absence of IL or in low Mmim DMP or Emim acetate concentrations, noticeably more glucose was released by T. reesei Cel7A than by the other enzymes. This effect was also apparent for enzymes expressed in the filamentous fungus, Neurospora crassa. We expressed some of these mutants and wild-type Te Cel7A in Neurospora crassa and assayed them for stability after preincubation as described in the next section. All of the N. crassa-expressed enzymes were more stable than their counterparts expressed in yeast (data not shown). Some of
Downloaded from http://peds.oxfordjournals.org/ at Weizmann Institute of Science on February 1, 2016
remain active in 40% (w/w) 1,3-dimethylimidazolium dimethylphosphate (Mmim DMP) (Wolski et al., 2011). We also showed β-glucosidase to be stable in up to 60% (w/w) solutions of Mmim DMP in water. Mmim DMP has been shown for the pretreatment of barley straw (Mood et al., 2013) and corn cob (Li et al., 2010). Mmim DMP was thus selected as the model solvent to examine the activity of cellulase variants in the present work. Cellulases from Trichoderma reesei comprise endoglucanases and exoglucanases (cellobiohydrolases, CBHs). Endoglucanases break the glycosidic bonds within a polyglucan chain, while CBHs cut from the chain ends (Wilson and Irwin, 1999). Cel7A, also known as CBHI, comprises ∼60% of the T. reesei cellulase mixture, and thus was chosen for the present study (Wood, 1992). Directed evolution is useful in improving enzyme kinetics and substrate specificity, in addition to solvent and/or thermal tolerance (Brayan et al., 1986). For example, Chen and Arnold used directed evolution to engineer subtilisin for increased catalytic activity in dimethylformamide using multiple rounds of error-prone PCR and screening (Chen and Arnold, 1993). Another approach to directed evolution is DNA shuffling, which was employed in the present work. DNA shuffling involves digesting homologous genes and allowing the DNA pieces to align at homologous regions followed by extension via PCR, resulting in chimeric mutant genes (Stemmer, 1994). Directed evolution has been applied to cellulases in efforts to increase their activity and/or optimal temperature. These methods have employed both error-prone PCR and DNA shuffling (Percival Zhang et al., 2006). In the present work, we used Biased Clique Shuffling, where 50% of the library was from Talaromyces emersonii Cel7A, because this method has been demonstrated to produce a more active library than without biasing the library toward any one enzyme (Dana et al., 2012). Previous efforts to improve Cel7A from Talaromyces emersonii used structure-guided site-directed mutagenesis (Heinzelman et al., 2010; Voutilainen et al., 2010; Komor et al., 2012). One report (Voutilainen et al., 2010) showed that adding disulfide bonds with structurally guided information resulted in a higher Tm than any Cel7A reported (84°C), but the activity of the enzymes they created was hampered. The group also generated a mutant with a Tm of 78.5°C with some improved activity using one structurally guided disulfide bond added. In this present work we report here, with the exception of the enzymes with the added disulfide bonds, we believe to have generated the Cel7A with the highest Tm in aqueous buffer while also making it more stable in aqueous–ILs. Additionally, Biased Clique Shuffling does not require any structural information. Recently some groups have employed error-prone PCR to enhance the stability in aqueous–ILs of some non-cellulase enzymes (Liu et al., 2013; Carter et al., 2014; Frauenkron-Machedjou et al., 2015).
P.Wolski et al.
Engineering ionic liquid-tolerant cellulases for biofuels production
this effect can be attributed to the natively expressed glutaminyl cyclase (QC) in T. reesei and N. crassa (Dana et al., 2014). The effect of the expression host was explored on heterologously expressed cellulases and determined that post-translation modification of the N-terminal glutamine to pyroglutamate is important for activity and stability of these enzymes (Dana et al., 2014).
Enzyme stability in IL/water solutions and at elevated temperature We examined the stability of selected Cel7A variants in aqueous–IL solutions by incubation in 43% (w/w) Mmim DMP or 20% (w/w) Emim acetate at 50°C. Residual activity at 50°C was measured after preincubation in the aqueous–IL solution, followed by 10-fold dilution into a buffered solution containing the substrate MU-lac. Figure 3 shows the residual activity after the 16-h preincubation at the above conditions for T. reesei Cel7A, Te Cel7A, 1M10, and 2K15 and the following mutants identified previously (Dana et al., 2012): 2I13 and 1G21. Cel7A from T. reesei and Cel7A from T. emersonii (expressed in S. cerevisiae) were the least stable of the enzymes assayed. Figure 3 also shows the thermostability of those variants at 65°C in aqueous buffer (see Materials and methods). The 1M10 variant was stable after 24 h of incubation in 20% (w/w) Emim acetate. 1M10 and 2K15 had the highest residual activity after incubations of 1 h or less (not shown) in 43% (w/w) Mmim DMP, and also exhibited hydrolysis in 43% (w/w) Mmim DMP of Avicel pretreated with Emim acetate (Figs. 1 and 2). The wild-type enzymes were the least stable after incubation in either of the IL solutions described above, becoming irreversibly inactivated in 43% Mmim DMP. In addition, T. reesei Cel7A and Te Cel7A lost activity irreversibly after incubation at 65°C. As a potential indicator of a relationship between thermostability and IL-stability, enzymes that were stable in 20% Emim acetate at 50°C were also stable in buffer at 65°C (Fig. 3). 1M10 was the most stable under each condition (without QC treatment). Stability of all seven of the enzymes tested (without QC treatment) in
Fig. 3 Residual activity of selected Cel7A variants. The assay was based on MU-lac hydrolysis at 50°C in 50 mM sodium acetate after preincubation for 16 h in varying conditions. Symbols for incubation conditions after 16 h: 50 mM citrate pH 4.8 at 65°C: open circles; 20% Emim acetate, pH 4.8, at 50°C: filled squares; and 43% Mmim DMP, pH 4.8, at 50°C: stars. QC indicates treatment with glutaminyl cyclase prior to assay.
20% Emim acetate at 50°C after 16 h of incubation and stability of them in buffer at 65°C after 16 h of incubation have a correlation coefficient (R 2) value of 0.933 when plotted with stability (% activity remaining) in either condition on separate axes (not shown).
Structural basis of stability Differential scanning calorimetry (DSC) data were collected for several Cel7A variants in acetate buffer, pH 5.0, and in aqueous Mmim DMP. The melting temperatures are presented in Table I. Here we define melting temperature (Tm) as the temperature at which the peak of the thermal unfolding curve occurs during DSC, and the Tm’s are presented in Table I. These results demonstrate that the loss of enzymatic activity in the IL is associated, at least partially, with unfolding of the enzyme. For example, the wild-type Te enzyme (without QC treatment) has a Tm in 43% Mmim DMP of 46.0°C, which is below the temperature (50°C) at which it is typically assayed. The 1M10 variant has a Tm of 50.1°C, which may indicate why some activity is observed with 1M10 in 43% Mmim DMP at 50°C. Reversible unfolding was not observed via DSC in 43% Mmim DMP. The mutations present in 2K15 are listed in the Materials and methods section. Full sequence information for 1M10 and Te is given in Supplementary Table SI. Mutation information is given in the Materials and methods. The mutations present in 2K15 and the three previously studied (Dana et al., 2012) are all on the surface of the protein. The only mutations shared by 2K15 and 2I13 are at sites 58 and 60, P58T and Y60L. These are in a loop or turn region of the structure. It has been shown previously that smaller loops and more compact structures can lead to greater protein thermostability (Kotik and Zuber, 1993; Russell and Taylor, 1995). The Y to L mutation could be making this region more hydrophobic, which could lead to burial of this residue and potential stabilization of this region of the protein. The 1M10 variant has mutations in some of the same places as 2K15, but also many others. All but six of its mutations appear to be on the surface of the protein. Future work of interest could include selectively removing the mutations in 1M10 to examine their effect on enzyme stability.
Downloaded from http://peds.oxfordjournals.org/ at Weizmann Institute of Science on February 1, 2016
Fig. 2 Hydrolysis of 2.5 g/L IL-Avicel by Cel7A variant enzymes after 16 h of reaction at 50°C in varying concentrations of Emim acetate, pH 4.8. The enzyme loading was 0.2 μM. All enzymes were expressed in S. cerevesiae, except T. reesei Cel7A, which was purified from Celluclast. Symbols: T. reesei Cel7A: filled circles; Te Cel7A: open circles; 1G21: open triangles; 2K15: filled inverted triangles; 2I13: filled squares; 1M10: open squares; Te Cel7A QC treated: filled triangles; 2I13 QC treated: open diamonds; 1M10 QC treated: stars. QC indicates treatment with glutaminyl cyclase prior to assay. Dashed lines are shown for the enzymes with N-terminal pyroglutamate (either from adding glutaminyl cyclase or from T. reesei expression host).
3
P.Wolski et al.
4
Table I. Melting temperatures of various Cel7A variants in varying concentrations of Mmim DMP measured by DSC % Mmim DMP (w/w)
0 10 20 30 43
Not QC treated
Not QC treated
Not QC treated first peak
Not Qc treated second peak
QC treated
QC treated
QC treated
QC active in vivo
Tm (°C) 1M10
Tm (°C) Te
Tm (°C) 2I13
Tm (°C) 2I13
Tm (°C) 1M10
Tm (°C) Te
Tm (°C) 2I13
Tm (°C) T. reesei Cel7Aa
66.3 65.1 62.5 55.8 50.1
64.0 62.1 60.8 52.1 46.0
65.2
74.6
76.7
74.3
74.6
61.2a
45.7
57.6
58.1
56.2
57.7
The two different peaks for 2I13 correspond to enzyme in the sample with and without cyclization upon storage (second peak). QC treated refers to the addition of exogenous glutaminyl cyclase. a Lantz et al., (2010).
Expression of the wild-type Te enzyme in Neurospora crassa resulted in an enzyme that was more stable than the same enzyme expressed in S. cerevesiae (Dana et al., 2014). This stability difference was due to conversion of the N-terminal glutamine to pyroglutamate (Grassick et al., 2004) by glutaminyl cyclase, a post-translational modification that occurred in N. crassa but not in this S. cerevisiae system. Treatment of the enzymes expressed in S. cerevesiae with exogenous glutaminyl cyclase enhanced their stability, which was verified by DSC (Dana et al., 2014). In order to ensure the presence of pyroglutamate, the 1M10 and 2I13 variants and wild-type Te enzyme expressed in yeast, each with no His tag, were purified and subjected to glutaminyl cyclase treatment (see Materials and methods). Cyclization of the N-terminal glutamine increased the melting temperature as measured by DSC of all enzymes substantially (∼10°C) (see Table I). Even after glutamine cyclization of both variants, the 1M10 variant was still more stable than the Te wild-type enzyme, having a higher Tm both in buffer and in 43% Mmim DMP. In addition, we observed two peaks for the DSC trace of the 2I13 variant without exogenous glutaminyl cyclase treatment. This sample had been stored at 4°C in 20 mM acetate buffer, pH 5, for 18 months (long after the assays with this enzyme were performed). It appears that some of the sample had undergone N-terminal glutamine cyclization spontaneously (Table I). Most stability assays were performed using the QC-treated enzyme and Te without QC treatment as a comparison, and the results are included in Figs. 1–3. As shown, after QC treatment all three enzyme variants exhibited similar activity in aqueous Mmim DMP and Emim acetate solutions toward Avicel, and had similar activity when assayed on 4-MU-lactoside after incubation in 43% Mmim DMP at 50°C or in citrate buffer, pH 4.8, at 65°C. Additionally, QC treated 1M10 had slightly higher activity against Avicel in aqueous buffer than QC treated Te enzyme (Fig. 1). QC treated enzymes retained almost all of their activity after 16 h (and 24 h; data not shown) of incubation at 65°C, whereas Te without QC treatment had lost all of its activity.
Conclusions Using directed evolution via DNA shuffling, several variant cellulases were isolated that were more active and stable in aqueous–IL solutions and at elevated temperatures than the wild-type enzyme. Cellulases with higher melting temperatures were active at higher concentrations
of IL in IL/water mixtures and were more stable at elevated temperature in buffer. Future studies on a greater set of IL types and IL concentrations and temperatures in IL and in buffer are warranted to solidify any correlation between IL stability and thermostabiliy. Based on DSC studies, the 1M10 mutant was more stable than the wild-type enzyme in buffer and in an IL–water mixture, with a higher Tm. The 1M10 mutant was still more stable under both conditions than the wild-type Te enzyme following the removal of the His tag and treatment with glutaminyl cyclase. Future evolution studies including addition of an active glutaminyl cyclase with access to the N-terminal glutamine, either produced by the expression host or added exogenously to all variants, are warranted.
Materials and methods Mutant library construction Biased clique shuffling enriched with previously determined stable variants The library of chimeric Cel7A genes (including catalytic domain, linker, and carbohydrate binding module) was generated via Biased Clique Shuffling as described (Dana et al., 2012). Three thermostable mutants had been generated using that library (Dana et al., 2012). The library used in the study reported here was then generated from a pool, of which 25% was the T. emersonii Cel7A gene. Each of the three thermostable mutants composed 8.33% of the library’s DNA pool, while the 10 genes from the organisms listed below (other than T. emersonii) each composed 5% of the library. This library was cloned, transformed, and selected as for the first library. Briefly, 11 Cel7A genes were selected based on homology to Cel7A from Talaromyces emersonii. The genes selected were from the following organisms: Aspergillus terreus, Aspergillus fischerianus, Penicillium chrysogenum, Thermoascus aurantiacus, Aspergillus nidulans, Aspergillus oryzae, Penicillium decumbens, Aspergillus clavatus, and Talaromyces emersonii. They were synthesized by Genscript for T. emersonii Cel7A and DNA 2.0 for the others. The codons were optimized for expression in S. cerevisiae using the companies’ propriety algorithms. The genes were cloned into the expression vector pCu424. Library DNA was digested using DNaseI. Digestion was quenched with EDTA. The library was then reassembled using PCR and gel purified. After cloning into pCu424 and amplification of the library in E. coli Top10, the library was transformed into S. cerevesiae YVH10, which overexpresses protein disulfide isomerase and was provided by the Wittrup group (Robinson et al., 1994). The PMR1 gene
Downloaded from http://peds.oxfordjournals.org/ at Weizmann Institute of Science on February 1, 2016
Assays of enzyme variants treated with glutaminyl cyclase
Engineering ionic liquid-tolerant cellulases for biofuels production had also been disrupted (Dana et al., 2012). Transformants were selected by plating onto DOBA/SC-Trp (both from Sunrise Science Products, using their recommendations) with 100 mg/L adenine hemisulfate.
Library screening Substrate preparation The chimeric proteins were assayed for hydrolysis of 2.5 g/L IL-pretreated Avicel. The substrate was prepared by dissolution of 1 g of Avicel into 10 g 1-ethyl-3-methylimidazolium acetate (Iolitec), which was incubated for 1 h at 100°C with stirring to produce an optically clear solution. The Avicel was then regenerated from the IL by cooling the IL to 70°C and adding 10 ml of water that was also at 70°C. The resulting material was centrifuged at 3220 g for 5 min, and the supernatant was removed. The wash/centrifuge step was repeated four times over, and then the IL-pretreated Avicel was diluted to 1.25% in 50 mM citrate buffer, pH 4.8. Assay set up In 384-well plates (Corning 3958), 14 μl of 1.25% IL-pretreated Avicel was pipetted using 250-μl wide orifice tips (Axygen) with a Biomek FXP robot (Beckman Coulter) from a slurry that was homogenized using a magnetic tumble stirrer (V&P Scientific) during pipetting. 51 μl of 27.5% (w/w) 1,3-dimethylimidazolium dimethylphosphate (Ioltec), pH 4.8, or 50 mM citrate buffer, pH 4.8, was then pipetted into the plates. The frozen yeast supernatant containing the expressed protein was thawed to room temperature and then 5 μl of it was pipetted with the Biomek FXP robot. To determine background glucose in the yeast supernatant, 65 μl of 50 mM citrate pH 4.8 was added to 5 μl of yeast supernatant from each mutant in the library. The experimental and background plates were then sealed and incubated at 50°C with shaking at 0.24 g in a Thermo Scientific MaxQ6000 incubator shaker for 16 h and then frozen before further analysis. Glucose analysis After thawing the assay plates, the cellulose hydrolysis reaction was diluted with 50 mM citrate 1:1. Then, after centrifugation of the dilution plates at 1610 g for 2 min, 3 μl of the diluted hydrolysis reaction were mixed with 3 μl of β-glucosidase (5 mg/ml, purified from almonds) and incubated at 37°C for 2 h. Glucose was then determined by Amplex Red analysis (Kim et al., 2010). Selection of enzyme variants The variants selected for further study were based on the amount of glucose released in 20% Mmim DMP at 50°C after 16 h. All selected
variants produced more glucose than the 2I13 (Dana et al., 2012) mutant. Ranking was based on sugar released from IL-pretreated Avicel in the presence of IL divided by the sugar released without the IL present. Selected mutants from this ranking had at least a two-fold signal to background ratio.
Characterization of selected enzymes All characterization assays (activity, stability, and DSC analysis) were conducted with purified enzymes at the concentrations given below.
Protein expression of selected mutants and wild-type enzymes The enzymes listed in Fig. 1 were revived from the frozen glycerol stocks by growth in 5 ml of YPAD until saturation. Then, using 2 ml of the saturated culture, 100 ml cultures of SC-Trp in 300 ml Ultra Yield flasks (Thomson Instrument Company) for each gene were inoculated and allowed to shake at 1.12 g at 30°C for 3 days. Protein induction was done as for the high throughput library with centrifugation of the cultures and replacing the supernatant with YPAG with 500 μM CuSO4 and then shaking at 1.12 g at 25°C for 3 days. Sequence analysis of the most stable mutant, 1M10, showed the following mutations: A6L, A8E, N10V, P58E, T59S, Y60L, L73V, G80A, V84I, S87N, S89D, K92T, L105V, L108M, Q109E, N220T, V222F, S301K, I308V, S311G, N312K, Q316N, P317S, N318E, D320T, I321W, T325G, T438N, G439P, T440P, P441G, S442G, H471M. The mutations in the second most stable mutant, 2K15, are as follows: A3V, A6Q, T7K, A8P, N10T, E18K, P22D, P58T, Y60L, N157A, D181N, E183Q.
Purification of selected mutants Each mutant enzyme was collected from the supernatant of the induction culture by centrifugation at 3220 g. The supernatant was concentrated using 10 kDa cut-off 50 ml Sartorius spin concentrators until 90% of the copper had been removed and exchanged with the 150 mM NaCl, 50 mM Tris, pH 7.6. The samples were then equilibrated to 10 mM imidazole as well and then frozen at −20°C until purification. Purification was accomplished using an Äkta Purifier with a 5 ml His Trap column (GE Healthcare) with elution with increasing imidazole concentration. Imidazole was then removed via buffer exchange with 20 mM sodium acetate, pH 5.0, via spin concentration. Enzymes for glutaminyl cyclase treatment had had their His Tag removed. These enzymes were purified via anion exchange chromatography. The first column was a HiPrep Q (GE Healthcare) column, and the second was a Mono Q column, also from GE Healthcare. Elution was performed in 50 mM Na phosphate, pH 7.4, with a gradient of 30–45 min at 4 ml/min of 0–0.5 M NaCl. Active fractions were identified with activity on 4-MU lactoside (see below).
Assays against IL-pretreated Avicel Based on OD280, the mutant proteins were normalized in concentration so that the protein concentration in the hydrolysis assay would be 0.2 μM. IL-pretreated Avicel and aqueous–IL or 50 mM citrate buffer, pH 4.8, were added to 96-well assay plates with the Biomek robot. Then, the purified enzymes were added, and the assay plates were then centrifuged for 2 min at 1610 g. The plates were sealed and shaken at 0.63 g at 50°C for 16 h. They were then analyzed for glucose as for the high throughput assay, using 3 h of incubation time with the β-glucosidase.
Downloaded from http://peds.oxfordjournals.org/ at Weizmann Institute of Science on February 1, 2016
Chimeric cellulase production From the selection plates, colonies were picked using a Genetix Qpix2 colony picker and grown in 250 μl of YPD media (50 g/L) with 100 mg/L of adenine hemisulfate (YPAD) in 96-well plates. After 3 days of shaking at 30°C at 1.75 g, the supernatants were pipetted off and replaced with YPAG with 500 μM CuSO4 (for heterologous protein induction). The YPAG media contains 10 g/L yeast extract, 20 g/L peptone, and 20 g/L galactose (galactose replaced the dextrose to minimize any glucose background in subsequent cellulase activity assays). Of this suspension, 10 μl were resuspended in YPAG with 15% glycerol. After 3 days of growth in the induction media at 25°C, the plates containing the heterologously expressed protein in the supernatant were frozen at −20°C prior to assay. The glycerol stocks were then stored at −80°C.
5
6
Assays against 4-methylumbelliferyl lactoside (MU-lac) These assays were conducted in PCR tubes in thermocyclers. Cellulases were incubated in 50 mM citrate, pH 4.8, or in aqueous– IL at 0.3 μM protein concentration. Cellulase activity against MU-lac (Carbosynth) was measured as follows: the cellulase was diluted 10-fold (to 0.03 μM enzyme reaction concentration) from the incubation conditions (buffer or aqueous–IL) into 90 μl of 930 μM Mu-lac in 50 mM Na acetate, pH 5, resulting in reaction concentration of 837 μM MU-lac at 100 μl. The MU-lac hydrolysis was conducted at 50°C for 10 min. The MU-lac hydrolysis was stopped by addition of 25 μl of 1 M glycine, pH 10.2, and the MU-lac hydrolysis was then determined by fluorescence at an excitation of 365 nm and an emission of 470 nm.
Differential scanning calorimetry
Glutaminyl cyclase treatment of yeast-expressed Cel7A variants Glutaminyl cyclase (QC) was purchased from Sino Biological. 0.2 mg/ ml QC enzyme was diluted 10-fold into a cellulase stock that was 80– 100 μM Cel7A in a 50 mM Na phosphate, pH 7.4. The reaction was then incubated at 30°C for 2 days.
Structural analysis Identification of the location of mutations on the enzymes was done using Visual Molecular Dynamics (VMD) (Humphrey et al., 1996) and Chimera software (Pettersen et al., 2004).
Supplementary data Supplementary data are available at PEDS online.
Funding This work was supported by the Energy Biosciences Institute. We are grateful to Dr. Meera Atreya for providing purified Cel7A from T. reesei Celluclast, to Dr. Harshal Chokhawala for providing T. reesei EGI, to Dr. Joel Graham for providing the EBI 244 enzyme, and to Dr. Christine Roche for expression of enzymes in Neurospora crassa.
References Brayan,P.N., Rollence,M.L., Pantoliano,M.W., Wood,J., Finzel,B.C., Gilliland, G.L., Howard,A.J. and Poulos,T.L. (1986) Proteins Struct. Funct. Bioinform., 1, 326–334. Brennan,T.C.R., Datta,S., Blanch,H.W., Simmons,B.A. and Holmes,B.M. (2010) Bio. Energy Res., 3, 123–133. Carter,J.L.L., Bekhouche,M., Noiriel,A., Blum,L.J. and Doumèche,B. (2014) Chem. Bio. Chem., 15, 2710–2718. Chen,K. and Arnold,F.H. (1993) Proc. Natl Acad. Sci. USA, 90, 5618–5622. Chidambaram,M. and Bell,A.T. (2010) Green Chem, 12, 1253.
Dadi,A.P., Schall,C.A. and Varanasi,S. (2007) Appl. Biochem. Biotechnol., 137–140, 407–421. Dana,C.M., Saija,P., Kal,S.M., Bryan,M.B., Blanch,H.W. and Clark,D.S. (2012) Biotechnol. Bioeng., 109, 2710–2719. Dana,C.M., Dotson-Fagerstrom,A., Roche,C.M., Kal,S.M., Chokhawala,H.A., Blanch,H.W. and Clark,D.S. (2014) Biotechnol. Bioeng., 111, 842–847. Datta,S., Holmes,B., Park,J.I., Chen,Z., Dibble,D.C., Hadi,M., Blanch,H.W., Simmons,B.A. and Sapra,R. (2010) Green Chem., 12, 338–345. Frauenkron-Machedjou,V.J., Fulton,A., Zhu,L., Anker,C., Bocola,M., Jaeger, K.-E. and Schwaneberg,U. (2015) Chem. Bio. Chem., 16, 937–945. Gladden,J.M., Allgaier,M., Miller,C.S., Hazen,T.C., VanderGheynst,J.S., Hugenholtz,P., Simmons,B. and Singer,S.W. (2011) Appl. Environ. Microbiol., 77, 5804–5812. Grassick,A., Murray,P.G., Thompson,R., Collins,C.M., Byrnes,L., Birrane,G., Higgins,T.M. and Tuohy,M.G. (2004) Eur. J. Biochem., 271, 4495–4506. Heinzelman,P., Komor,R., Kanaan,A., Romero,P., Yu,X., Mohler,S., Sno,C. and Arnold,F. (2010) Protein Eng. Design Sel., 23, 871–880. Humphrey,W., Dalke,A. and Schulten,K. (1996) J. Molec. Graphics, 14, 33–38. Kamiya,N., Matsushita,Y., Hanaki,M., Nakashima,K., Narita,M., Goto,M. and Takahashi,H. (2008) Biotechnol. Lett, 30, 1037–1040. Kim,T.-W., Chokhawala,H.A., Nadler,D., Blanch,H.W. and Clark,D.S. (2010) Biotechnol. Bioeng., 107, 601–611. Komor,R.S., Romero,P.A., Xie,C.B. and Arnold,F.H. (2012) Protein Eng.v Design Sel., 25, 827–833. Kotik,M. and Zuber,H. (1993) Eur. J. Biochem., 211, 267–280. Lantz,S.E., Goedegebuur,F., Hommes,R., Kaper,T., Kelemen,B.R., Mitchinson, C., Wallace,L., Ståhlberg,J. and Larenas,E.A. (2010) Biotechnol. Biofuels, 3, 20. Lee,S.H., Doherty,T.V., Linhardt,R.J. and Dordick,J.S. (2009) Biotechnol. Bioeng, 102, 1368–1376. Li,Q., Jiang,X., He,Y., Li,L., Xian,M. and Yang,J. (2010) Appl. Microbiol. Biotechnol., 87, 117–126. Liu,H., Zhu,L., Bocola,M., Chen,N., Spiess,A.C. and Schwaneberg,U. (2013) Green Chem., 15, 1348–1355. Mood,S.H., Golfeshan,A.H., Tabatabaei,M., Abbasalizadeh,S. and Ardjmand, M. (2013) 3 Biotech., 3, 399–406. Park,J.I., Steen,E.J. and Burd,H. (2012) PLoS ONE, 7, e37010. Percival Zhang,Y.H., Himmel,M.E. and Mielenz,J.R. (2006) Biotechnol. Adv., 24, 452–481. Pettersen,E.F., Goddard,T.D., Huang,C.C., Couch,G.S., Greenblatt,D.M., Meng,E.C. and Ferrin,T.E. (2004) J. Comput. Chem., 25, 13, 1605–1612. Robinson,A.S., Hines,V. and Wittrup,K.D. (1994) Nat. Biotech., 12, 381–384. Russell,R.J.M. and Taylor,G.L. (1995) Curr. Opin. Biotechnol., 6, 370–374. Singer,S.W., Reddy,A.P., Gladden,J.M., Guo,H., Hazen,T.C., Simmons,B.A. and VanderGheynst,J.S. (2011) J. Appl. Microbiol., 110, 1023–1031. Stemmer,W.P.C. (1994) Nature, 370, 389–391. Su,C.-H., Chung,M.-H., Hsieh,H-J., Chang,Y.-K., Ding,J.C. and Wu,H.-M. (2012) J. Taiwan Inst. Chem. Eng., 43, 573–577. Swatloski,R.P., Spear,S.K., Holbrey,J.D. and Rogers,R.D. (2002) J. Am. Chem. Soc., 124, 4974–4975. Turner,M.B., Spear,S.K., Huddleston,J.G., Holbrey,J.D. and Rogers,R.D. (2003) Green. Chem., 5, 443–447. Voutilainen,S.P., Murray,P.G., Tuohy,M.G. and Koivula,A. (2010) Protein Eng. Design Sel., 23, 69–79. Wilson,D. and Irwin,D. (1999) In: Tsao,G., Brainard,A., Bungay,H., Cao,N., Cen,P., Chen,Z., Du,J., Foody,B., Gong,C., Hall,P., et al. (eds), Springer, Berlin/Heidelberg, pp. 1–21. Wolski,P.W., Clark,D.S. and Blanch,H.W. (2011) Green. Chem., 13, 3107–3110. Wood,T.M. (1992) Biochem. Soc. Trans., 20, 46–53. Zhao,H., Baker,G.A., Song,Z., Olubajo,O., Crittle,T. and Peters,D. (2008) Green Chem,, 10, 696–705.
Downloaded from http://peds.oxfordjournals.org/ at Weizmann Institute of Science on February 1, 2016
Using a Nano DSC (TA instruments), we measured melting temperatures of cellulases in the presence of aqueous–IL and in buffer alone. The buffer was 50 mM Na acetate, pH 5.0. Enzyme concentration was ∼0.2 mg/ml in the sample. Scans were done at a step rate of 1°C/min after equilibration and prior degassing of the sample.
P.Wolski et al.
Short Communication
Engineering ionic liquid-tolerant cellulases for biofuels production Paul W. Wolski1,2,3, Craig M. Dana1,3, Douglas S. Clark1,3,*, and Harvey W. Blanch1,3,* Energy Biosciences Institute, University of California, Berkeley, CA 94720, USA, 2Graduate Group of Comparative Biochemistry at University of California, Berkeley, CA, USA, and 3Department of Chemical and Biomolecular Engineering, University of California, Berkeley, CA 94720, USA *To whom correspondence should be addressed. E-mail: [email protected]; [email protected] Edited by Eva Petersen Received 5 May 2015; Revised 3 December 2015; Accepted 7 December 2015
Abstract Dissolution of lignocellulosic biomass in certain ionic liquids (ILs) can provide an effective pretreatment prior to enzymatic saccharification of cellulose for biofuels production. Toward the goal of combining pretreatment and enzymatic hydrolysis, we evolved enzyme variants of Talaromyces emersonii Cel7A to be more active and stable than wild-type T. emersonii Cel7A or Trichoderma reesei Cel7A in aqueous–IL solutions (up to 43% (w/w) 1,3-dimethylimdazolium dimethylphosphate and 20% (w/w) 1-ethyl-3-methylimidazolium acetate). In general, greater enzyme stability in buffer at elevated temperature corresponded to greater stability in aqueous–ILs. Post-translational modification of the N-terminal glutamine residue to pyroglutamate via glutaminyl cyclase enhanced the stability of T. emersonii Cel7A and variants. Differential scanning calorimetry revealed an increase in melting temperature of 1.9–3.9°C for the variant 1M10 over the wild-type T. emersonii Cel7A in aqueous buffer and in an IL–aqueous mixture. We observed this increase both with and without glutaminyl cyclase treatment of the enzymes. Key words: biased clique, cellulase, directed evolution, ionic liquid, shuffling
Introduction In 2002, Rogers and colleagues identified the room temperature ionic liquid (IL), 1-butyl-3-methylimidazolium (Bmim) chloride, to be an excellent solvent for cellulose (Swatloski et al., 2002). This IL was shown to dissolve up to 10% (w/w) cellulose. When the cellulose suspension in IL was microwaved, up to 25% (w/w) could be dissolved. Subsequently, Dadi et al. (2007) reported that cellulose precipitated from certain ILs by addition of an antisolvent such as water could be readily hydrolyzed by cellulase enzymes. Turner et al. (2003), however, reported that hydrolysis of precipitated cellulose in the presence of the IL 1-butyl-3-methylimidazolium (Bmim) chloride was ineffective, due to denaturation of the cellulases. 1-Ethyl-3-methylimidazolium (Emim) acetate is able to dissolve wood and separate the lignin from the holocellulose (Lee et al., 2009). Ethoxylated ILs with acetate anions (Zhao et al., 2008) were shown to permit enzymatic hydrolysis of the cellulosic content of lignocellulosic
biomass in the presence of the IL. Emim diethylphosphate is able to support cellulase activity at IL concentrations as high as 40% (v/v) (Kamiya et al., 2008). Combining lignocellulose pretreatment by dissolution in ILs with enzymatic hydrolysis in the same vessel offers potential increases in process efficiency. There have been a few other reports of this process, sometimes called one-pot pretreatment and hydrolysis (typically with naturally occurring thermophilic enzymes), with some success (Datta et al., 2010; Gladden et al., 2011; Park et al., 2012; Su et al., 2012). Other studies on selectively removing the glucose and sugars from the IL mixture (Brennan et al., 2010), in situ catalytic conversion to distillable fuels (Chidambaran and Bell, 2010), or even fermentation in the presence of IL (Singer et al., 2011) are contributing toward making such in situ enzymatic hydrolysis a reality. We previously examined protein stability in a variety of ILs using green fluorescent protein as a model for assessing cellulase stability (Wolski et al., 2011). Trichoderma reesei cellulases were shown to
© The Author 2016. Published by Oxford University Press. All rights reserved. For Permissions, please e-mail: [email protected]
1
Downloaded from http://peds.oxfordjournals.org/ at Weizmann Institute of Science on February 1, 2016
1
2
Results and discussion Screening and initial analysis of variants Cel7A variants were selected and purified for further study after screening the activity of each variant toward Emim acetate-pretreated Avicel (see Materials and methods) in aqueous solutions containing 20% (w/w) Mmim DMP, at pH 4.8, and in citrate buffer. The variants were generated via Biased Clique Shuffling (Methods) from the parent Talaromyces emersonii (Te) Cel7A. The variants that were selected all released more sugars from the pretreated Avicel in the presence of 20% (w/w) Mmim DMP than the variant ‘2I13’, described by Dana et al. (2012). When assayed on 4-methylumbelliferyl lactoside (MU-lac) for 10 min as well on Avicel for 22 h the variant 2I13 had an optimum
Fig. 1 Hydrolysis of 2.5 g/L IL-Avicel by Cel7A variant enzymes after 16 h of reaction at 50°C in varying concentrations of Mmim DMP, pH 4.8. The enzyme loading was 0.2 μM. All enzymes were expressed in S. cerevesiae, except T. reesei Cel7A, which was purified from Celluclast. Symbols: T. reesei Cel7A: filled circles; Te Cel7A: open circles; 1G21: open triangles; 2K15: filled triangles; 2I13: filled squares; 1M10: open squares; Te Cel7A QC treated: filled inverted triangles; 2I13 QC treated: open diamonds; 1M10 QC treated: stars. QC indicates treatment with glutaminyl cyclase prior to assay. Dashed lines are shown for the enzymes with N-terminal pyroglutamate (either from adding glutaminyl cyclase or from T. reesei expression host).
temperature (Topt.) of ∼60°C and exhibited the highest residual activity on MU-Lac after incubation at 65°C for 24 h of the mutants studied (Dana et al., 2012). In the present study, two variants (1M10 and 2K15) were also shown to exhibit activity toward MU-lac at 70°C for at least 6 days (data not shown). Of the variants selected (see ‘Selection of enzyme variants’ in Materials and Methods), twelve were expressed for further analysis. After purification, they were assayed for their activity toward Emim acetate-pretreated Avicel in 43% (w/w) Mmim DMP (Fig. 1). Also tested was T. reesei Cel7A, purified from Celluclast. In 43% (w/w) Mmim DMP, two variants, 2K15 and 1M10, produced more glucose than any of the variant or wild-type enzymes tested, including 2I13, 2E10, and 1G21, the most stable variants identified by Dana et al. (2012). T. reesei Cel7A was essentially inactive under this condition. Te Cel7A (expressed in S. cerevisiae) and T. reesei Cel7A ( purified from Celluclast) and variants 1M10, 2K15, 2I13, and 1G21 were assayed in Emim acetate/water solutions under the same conditions as those employed with Mmim DMP. Figure 2 shows the glucose released from the hydrolysis of Emim acetate-pretreated Avicel in 20% (w/w) Emim acetate as well as the glucose released from hydrolysis in buffer. The results are similar to those in 43% (w/w) Mmim DMP, with 1M10 and 2K15 exhibiting the most cellulose hydrolysis in 20% (w/w) Emim acetate relative to buffer. Cellulose hydrolysis is only observed in Emim acetate at up to 30% (w/w) concentration, consistent with our previous finding that Mmim DMP inhibits cellulolytic activity less than Emim acetate (Wolski et al., 2011). In the absence of IL or in low Mmim DMP or Emim acetate concentrations, noticeably more glucose was released by T. reesei Cel7A than by the other enzymes. This effect was also apparent for enzymes expressed in the filamentous fungus, Neurospora crassa. We expressed some of these mutants and wild-type Te Cel7A in Neurospora crassa and assayed them for stability after preincubation as described in the next section. All of the N. crassa-expressed enzymes were more stable than their counterparts expressed in yeast (data not shown). Some of
Downloaded from http://peds.oxfordjournals.org/ at Weizmann Institute of Science on February 1, 2016
remain active in 40% (w/w) 1,3-dimethylimidazolium dimethylphosphate (Mmim DMP) (Wolski et al., 2011). We also showed β-glucosidase to be stable in up to 60% (w/w) solutions of Mmim DMP in water. Mmim DMP has been shown for the pretreatment of barley straw (Mood et al., 2013) and corn cob (Li et al., 2010). Mmim DMP was thus selected as the model solvent to examine the activity of cellulase variants in the present work. Cellulases from Trichoderma reesei comprise endoglucanases and exoglucanases (cellobiohydrolases, CBHs). Endoglucanases break the glycosidic bonds within a polyglucan chain, while CBHs cut from the chain ends (Wilson and Irwin, 1999). Cel7A, also known as CBHI, comprises ∼60% of the T. reesei cellulase mixture, and thus was chosen for the present study (Wood, 1992). Directed evolution is useful in improving enzyme kinetics and substrate specificity, in addition to solvent and/or thermal tolerance (Brayan et al., 1986). For example, Chen and Arnold used directed evolution to engineer subtilisin for increased catalytic activity in dimethylformamide using multiple rounds of error-prone PCR and screening (Chen and Arnold, 1993). Another approach to directed evolution is DNA shuffling, which was employed in the present work. DNA shuffling involves digesting homologous genes and allowing the DNA pieces to align at homologous regions followed by extension via PCR, resulting in chimeric mutant genes (Stemmer, 1994). Directed evolution has been applied to cellulases in efforts to increase their activity and/or optimal temperature. These methods have employed both error-prone PCR and DNA shuffling (Percival Zhang et al., 2006). In the present work, we used Biased Clique Shuffling, where 50% of the library was from Talaromyces emersonii Cel7A, because this method has been demonstrated to produce a more active library than without biasing the library toward any one enzyme (Dana et al., 2012). Previous efforts to improve Cel7A from Talaromyces emersonii used structure-guided site-directed mutagenesis (Heinzelman et al., 2010; Voutilainen et al., 2010; Komor et al., 2012). One report (Voutilainen et al., 2010) showed that adding disulfide bonds with structurally guided information resulted in a higher Tm than any Cel7A reported (84°C), but the activity of the enzymes they created was hampered. The group also generated a mutant with a Tm of 78.5°C with some improved activity using one structurally guided disulfide bond added. In this present work we report here, with the exception of the enzymes with the added disulfide bonds, we believe to have generated the Cel7A with the highest Tm in aqueous buffer while also making it more stable in aqueous–ILs. Additionally, Biased Clique Shuffling does not require any structural information. Recently some groups have employed error-prone PCR to enhance the stability in aqueous–ILs of some non-cellulase enzymes (Liu et al., 2013; Carter et al., 2014; Frauenkron-Machedjou et al., 2015).
P.Wolski et al.
Engineering ionic liquid-tolerant cellulases for biofuels production
this effect can be attributed to the natively expressed glutaminyl cyclase (QC) in T. reesei and N. crassa (Dana et al., 2014). The effect of the expression host was explored on heterologously expressed cellulases and determined that post-translation modification of the N-terminal glutamine to pyroglutamate is important for activity and stability of these enzymes (Dana et al., 2014).
Enzyme stability in IL/water solutions and at elevated temperature We examined the stability of selected Cel7A variants in aqueous–IL solutions by incubation in 43% (w/w) Mmim DMP or 20% (w/w) Emim acetate at 50°C. Residual activity at 50°C was measured after preincubation in the aqueous–IL solution, followed by 10-fold dilution into a buffered solution containing the substrate MU-lac. Figure 3 shows the residual activity after the 16-h preincubation at the above conditions for T. reesei Cel7A, Te Cel7A, 1M10, and 2K15 and the following mutants identified previously (Dana et al., 2012): 2I13 and 1G21. Cel7A from T. reesei and Cel7A from T. emersonii (expressed in S. cerevisiae) were the least stable of the enzymes assayed. Figure 3 also shows the thermostability of those variants at 65°C in aqueous buffer (see Materials and methods). The 1M10 variant was stable after 24 h of incubation in 20% (w/w) Emim acetate. 1M10 and 2K15 had the highest residual activity after incubations of 1 h or less (not shown) in 43% (w/w) Mmim DMP, and also exhibited hydrolysis in 43% (w/w) Mmim DMP of Avicel pretreated with Emim acetate (Figs. 1 and 2). The wild-type enzymes were the least stable after incubation in either of the IL solutions described above, becoming irreversibly inactivated in 43% Mmim DMP. In addition, T. reesei Cel7A and Te Cel7A lost activity irreversibly after incubation at 65°C. As a potential indicator of a relationship between thermostability and IL-stability, enzymes that were stable in 20% Emim acetate at 50°C were also stable in buffer at 65°C (Fig. 3). 1M10 was the most stable under each condition (without QC treatment). Stability of all seven of the enzymes tested (without QC treatment) in
Fig. 3 Residual activity of selected Cel7A variants. The assay was based on MU-lac hydrolysis at 50°C in 50 mM sodium acetate after preincubation for 16 h in varying conditions. Symbols for incubation conditions after 16 h: 50 mM citrate pH 4.8 at 65°C: open circles; 20% Emim acetate, pH 4.8, at 50°C: filled squares; and 43% Mmim DMP, pH 4.8, at 50°C: stars. QC indicates treatment with glutaminyl cyclase prior to assay.
20% Emim acetate at 50°C after 16 h of incubation and stability of them in buffer at 65°C after 16 h of incubation have a correlation coefficient (R 2) value of 0.933 when plotted with stability (% activity remaining) in either condition on separate axes (not shown).
Structural basis of stability Differential scanning calorimetry (DSC) data were collected for several Cel7A variants in acetate buffer, pH 5.0, and in aqueous Mmim DMP. The melting temperatures are presented in Table I. Here we define melting temperature (Tm) as the temperature at which the peak of the thermal unfolding curve occurs during DSC, and the Tm’s are presented in Table I. These results demonstrate that the loss of enzymatic activity in the IL is associated, at least partially, with unfolding of the enzyme. For example, the wild-type Te enzyme (without QC treatment) has a Tm in 43% Mmim DMP of 46.0°C, which is below the temperature (50°C) at which it is typically assayed. The 1M10 variant has a Tm of 50.1°C, which may indicate why some activity is observed with 1M10 in 43% Mmim DMP at 50°C. Reversible unfolding was not observed via DSC in 43% Mmim DMP. The mutations present in 2K15 are listed in the Materials and methods section. Full sequence information for 1M10 and Te is given in Supplementary Table SI. Mutation information is given in the Materials and methods. The mutations present in 2K15 and the three previously studied (Dana et al., 2012) are all on the surface of the protein. The only mutations shared by 2K15 and 2I13 are at sites 58 and 60, P58T and Y60L. These are in a loop or turn region of the structure. It has been shown previously that smaller loops and more compact structures can lead to greater protein thermostability (Kotik and Zuber, 1993; Russell and Taylor, 1995). The Y to L mutation could be making this region more hydrophobic, which could lead to burial of this residue and potential stabilization of this region of the protein. The 1M10 variant has mutations in some of the same places as 2K15, but also many others. All but six of its mutations appear to be on the surface of the protein. Future work of interest could include selectively removing the mutations in 1M10 to examine their effect on enzyme stability.
Downloaded from http://peds.oxfordjournals.org/ at Weizmann Institute of Science on February 1, 2016
Fig. 2 Hydrolysis of 2.5 g/L IL-Avicel by Cel7A variant enzymes after 16 h of reaction at 50°C in varying concentrations of Emim acetate, pH 4.8. The enzyme loading was 0.2 μM. All enzymes were expressed in S. cerevesiae, except T. reesei Cel7A, which was purified from Celluclast. Symbols: T. reesei Cel7A: filled circles; Te Cel7A: open circles; 1G21: open triangles; 2K15: filled inverted triangles; 2I13: filled squares; 1M10: open squares; Te Cel7A QC treated: filled triangles; 2I13 QC treated: open diamonds; 1M10 QC treated: stars. QC indicates treatment with glutaminyl cyclase prior to assay. Dashed lines are shown for the enzymes with N-terminal pyroglutamate (either from adding glutaminyl cyclase or from T. reesei expression host).
3
P.Wolski et al.
4
Table I. Melting temperatures of various Cel7A variants in varying concentrations of Mmim DMP measured by DSC % Mmim DMP (w/w)
0 10 20 30 43
Not QC treated
Not QC treated
Not QC treated first peak
Not Qc treated second peak
QC treated
QC treated
QC treated
QC active in vivo
Tm (°C) 1M10
Tm (°C) Te
Tm (°C) 2I13
Tm (°C) 2I13
Tm (°C) 1M10
Tm (°C) Te
Tm (°C) 2I13
Tm (°C) T. reesei Cel7Aa
66.3 65.1 62.5 55.8 50.1
64.0 62.1 60.8 52.1 46.0
65.2
74.6
76.7
74.3
74.6
61.2a
45.7
57.6
58.1
56.2
57.7
The two different peaks for 2I13 correspond to enzyme in the sample with and without cyclization upon storage (second peak). QC treated refers to the addition of exogenous glutaminyl cyclase. a Lantz et al., (2010).
Expression of the wild-type Te enzyme in Neurospora crassa resulted in an enzyme that was more stable than the same enzyme expressed in S. cerevesiae (Dana et al., 2014). This stability difference was due to conversion of the N-terminal glutamine to pyroglutamate (Grassick et al., 2004) by glutaminyl cyclase, a post-translational modification that occurred in N. crassa but not in this S. cerevisiae system. Treatment of the enzymes expressed in S. cerevesiae with exogenous glutaminyl cyclase enhanced their stability, which was verified by DSC (Dana et al., 2014). In order to ensure the presence of pyroglutamate, the 1M10 and 2I13 variants and wild-type Te enzyme expressed in yeast, each with no His tag, were purified and subjected to glutaminyl cyclase treatment (see Materials and methods). Cyclization of the N-terminal glutamine increased the melting temperature as measured by DSC of all enzymes substantially (∼10°C) (see Table I). Even after glutamine cyclization of both variants, the 1M10 variant was still more stable than the Te wild-type enzyme, having a higher Tm both in buffer and in 43% Mmim DMP. In addition, we observed two peaks for the DSC trace of the 2I13 variant without exogenous glutaminyl cyclase treatment. This sample had been stored at 4°C in 20 mM acetate buffer, pH 5, for 18 months (long after the assays with this enzyme were performed). It appears that some of the sample had undergone N-terminal glutamine cyclization spontaneously (Table I). Most stability assays were performed using the QC-treated enzyme and Te without QC treatment as a comparison, and the results are included in Figs. 1–3. As shown, after QC treatment all three enzyme variants exhibited similar activity in aqueous Mmim DMP and Emim acetate solutions toward Avicel, and had similar activity when assayed on 4-MU-lactoside after incubation in 43% Mmim DMP at 50°C or in citrate buffer, pH 4.8, at 65°C. Additionally, QC treated 1M10 had slightly higher activity against Avicel in aqueous buffer than QC treated Te enzyme (Fig. 1). QC treated enzymes retained almost all of their activity after 16 h (and 24 h; data not shown) of incubation at 65°C, whereas Te without QC treatment had lost all of its activity.
Conclusions Using directed evolution via DNA shuffling, several variant cellulases were isolated that were more active and stable in aqueous–IL solutions and at elevated temperatures than the wild-type enzyme. Cellulases with higher melting temperatures were active at higher concentrations
of IL in IL/water mixtures and were more stable at elevated temperature in buffer. Future studies on a greater set of IL types and IL concentrations and temperatures in IL and in buffer are warranted to solidify any correlation between IL stability and thermostabiliy. Based on DSC studies, the 1M10 mutant was more stable than the wild-type enzyme in buffer and in an IL–water mixture, with a higher Tm. The 1M10 mutant was still more stable under both conditions than the wild-type Te enzyme following the removal of the His tag and treatment with glutaminyl cyclase. Future evolution studies including addition of an active glutaminyl cyclase with access to the N-terminal glutamine, either produced by the expression host or added exogenously to all variants, are warranted.
Materials and methods Mutant library construction Biased clique shuffling enriched with previously determined stable variants The library of chimeric Cel7A genes (including catalytic domain, linker, and carbohydrate binding module) was generated via Biased Clique Shuffling as described (Dana et al., 2012). Three thermostable mutants had been generated using that library (Dana et al., 2012). The library used in the study reported here was then generated from a pool, of which 25% was the T. emersonii Cel7A gene. Each of the three thermostable mutants composed 8.33% of the library’s DNA pool, while the 10 genes from the organisms listed below (other than T. emersonii) each composed 5% of the library. This library was cloned, transformed, and selected as for the first library. Briefly, 11 Cel7A genes were selected based on homology to Cel7A from Talaromyces emersonii. The genes selected were from the following organisms: Aspergillus terreus, Aspergillus fischerianus, Penicillium chrysogenum, Thermoascus aurantiacus, Aspergillus nidulans, Aspergillus oryzae, Penicillium decumbens, Aspergillus clavatus, and Talaromyces emersonii. They were synthesized by Genscript for T. emersonii Cel7A and DNA 2.0 for the others. The codons were optimized for expression in S. cerevisiae using the companies’ propriety algorithms. The genes were cloned into the expression vector pCu424. Library DNA was digested using DNaseI. Digestion was quenched with EDTA. The library was then reassembled using PCR and gel purified. After cloning into pCu424 and amplification of the library in E. coli Top10, the library was transformed into S. cerevesiae YVH10, which overexpresses protein disulfide isomerase and was provided by the Wittrup group (Robinson et al., 1994). The PMR1 gene
Downloaded from http://peds.oxfordjournals.org/ at Weizmann Institute of Science on February 1, 2016
Assays of enzyme variants treated with glutaminyl cyclase
Engineering ionic liquid-tolerant cellulases for biofuels production had also been disrupted (Dana et al., 2012). Transformants were selected by plating onto DOBA/SC-Trp (both from Sunrise Science Products, using their recommendations) with 100 mg/L adenine hemisulfate.
Library screening Substrate preparation The chimeric proteins were assayed for hydrolysis of 2.5 g/L IL-pretreated Avicel. The substrate was prepared by dissolution of 1 g of Avicel into 10 g 1-ethyl-3-methylimidazolium acetate (Iolitec), which was incubated for 1 h at 100°C with stirring to produce an optically clear solution. The Avicel was then regenerated from the IL by cooling the IL to 70°C and adding 10 ml of water that was also at 70°C. The resulting material was centrifuged at 3220 g for 5 min, and the supernatant was removed. The wash/centrifuge step was repeated four times over, and then the IL-pretreated Avicel was diluted to 1.25% in 50 mM citrate buffer, pH 4.8. Assay set up In 384-well plates (Corning 3958), 14 μl of 1.25% IL-pretreated Avicel was pipetted using 250-μl wide orifice tips (Axygen) with a Biomek FXP robot (Beckman Coulter) from a slurry that was homogenized using a magnetic tumble stirrer (V&P Scientific) during pipetting. 51 μl of 27.5% (w/w) 1,3-dimethylimidazolium dimethylphosphate (Ioltec), pH 4.8, or 50 mM citrate buffer, pH 4.8, was then pipetted into the plates. The frozen yeast supernatant containing the expressed protein was thawed to room temperature and then 5 μl of it was pipetted with the Biomek FXP robot. To determine background glucose in the yeast supernatant, 65 μl of 50 mM citrate pH 4.8 was added to 5 μl of yeast supernatant from each mutant in the library. The experimental and background plates were then sealed and incubated at 50°C with shaking at 0.24 g in a Thermo Scientific MaxQ6000 incubator shaker for 16 h and then frozen before further analysis. Glucose analysis After thawing the assay plates, the cellulose hydrolysis reaction was diluted with 50 mM citrate 1:1. Then, after centrifugation of the dilution plates at 1610 g for 2 min, 3 μl of the diluted hydrolysis reaction were mixed with 3 μl of β-glucosidase (5 mg/ml, purified from almonds) and incubated at 37°C for 2 h. Glucose was then determined by Amplex Red analysis (Kim et al., 2010). Selection of enzyme variants The variants selected for further study were based on the amount of glucose released in 20% Mmim DMP at 50°C after 16 h. All selected
variants produced more glucose than the 2I13 (Dana et al., 2012) mutant. Ranking was based on sugar released from IL-pretreated Avicel in the presence of IL divided by the sugar released without the IL present. Selected mutants from this ranking had at least a two-fold signal to background ratio.
Characterization of selected enzymes All characterization assays (activity, stability, and DSC analysis) were conducted with purified enzymes at the concentrations given below.
Protein expression of selected mutants and wild-type enzymes The enzymes listed in Fig. 1 were revived from the frozen glycerol stocks by growth in 5 ml of YPAD until saturation. Then, using 2 ml of the saturated culture, 100 ml cultures of SC-Trp in 300 ml Ultra Yield flasks (Thomson Instrument Company) for each gene were inoculated and allowed to shake at 1.12 g at 30°C for 3 days. Protein induction was done as for the high throughput library with centrifugation of the cultures and replacing the supernatant with YPAG with 500 μM CuSO4 and then shaking at 1.12 g at 25°C for 3 days. Sequence analysis of the most stable mutant, 1M10, showed the following mutations: A6L, A8E, N10V, P58E, T59S, Y60L, L73V, G80A, V84I, S87N, S89D, K92T, L105V, L108M, Q109E, N220T, V222F, S301K, I308V, S311G, N312K, Q316N, P317S, N318E, D320T, I321W, T325G, T438N, G439P, T440P, P441G, S442G, H471M. The mutations in the second most stable mutant, 2K15, are as follows: A3V, A6Q, T7K, A8P, N10T, E18K, P22D, P58T, Y60L, N157A, D181N, E183Q.
Purification of selected mutants Each mutant enzyme was collected from the supernatant of the induction culture by centrifugation at 3220 g. The supernatant was concentrated using 10 kDa cut-off 50 ml Sartorius spin concentrators until 90% of the copper had been removed and exchanged with the 150 mM NaCl, 50 mM Tris, pH 7.6. The samples were then equilibrated to 10 mM imidazole as well and then frozen at −20°C until purification. Purification was accomplished using an Äkta Purifier with a 5 ml His Trap column (GE Healthcare) with elution with increasing imidazole concentration. Imidazole was then removed via buffer exchange with 20 mM sodium acetate, pH 5.0, via spin concentration. Enzymes for glutaminyl cyclase treatment had had their His Tag removed. These enzymes were purified via anion exchange chromatography. The first column was a HiPrep Q (GE Healthcare) column, and the second was a Mono Q column, also from GE Healthcare. Elution was performed in 50 mM Na phosphate, pH 7.4, with a gradient of 30–45 min at 4 ml/min of 0–0.5 M NaCl. Active fractions were identified with activity on 4-MU lactoside (see below).
Assays against IL-pretreated Avicel Based on OD280, the mutant proteins were normalized in concentration so that the protein concentration in the hydrolysis assay would be 0.2 μM. IL-pretreated Avicel and aqueous–IL or 50 mM citrate buffer, pH 4.8, were added to 96-well assay plates with the Biomek robot. Then, the purified enzymes were added, and the assay plates were then centrifuged for 2 min at 1610 g. The plates were sealed and shaken at 0.63 g at 50°C for 16 h. They were then analyzed for glucose as for the high throughput assay, using 3 h of incubation time with the β-glucosidase.
Downloaded from http://peds.oxfordjournals.org/ at Weizmann Institute of Science on February 1, 2016
Chimeric cellulase production From the selection plates, colonies were picked using a Genetix Qpix2 colony picker and grown in 250 μl of YPD media (50 g/L) with 100 mg/L of adenine hemisulfate (YPAD) in 96-well plates. After 3 days of shaking at 30°C at 1.75 g, the supernatants were pipetted off and replaced with YPAG with 500 μM CuSO4 (for heterologous protein induction). The YPAG media contains 10 g/L yeast extract, 20 g/L peptone, and 20 g/L galactose (galactose replaced the dextrose to minimize any glucose background in subsequent cellulase activity assays). Of this suspension, 10 μl were resuspended in YPAG with 15% glycerol. After 3 days of growth in the induction media at 25°C, the plates containing the heterologously expressed protein in the supernatant were frozen at −20°C prior to assay. The glycerol stocks were then stored at −80°C.
5
6
Assays against 4-methylumbelliferyl lactoside (MU-lac) These assays were conducted in PCR tubes in thermocyclers. Cellulases were incubated in 50 mM citrate, pH 4.8, or in aqueous– IL at 0.3 μM protein concentration. Cellulase activity against MU-lac (Carbosynth) was measured as follows: the cellulase was diluted 10-fold (to 0.03 μM enzyme reaction concentration) from the incubation conditions (buffer or aqueous–IL) into 90 μl of 930 μM Mu-lac in 50 mM Na acetate, pH 5, resulting in reaction concentration of 837 μM MU-lac at 100 μl. The MU-lac hydrolysis was conducted at 50°C for 10 min. The MU-lac hydrolysis was stopped by addition of 25 μl of 1 M glycine, pH 10.2, and the MU-lac hydrolysis was then determined by fluorescence at an excitation of 365 nm and an emission of 470 nm.
Differential scanning calorimetry
Glutaminyl cyclase treatment of yeast-expressed Cel7A variants Glutaminyl cyclase (QC) was purchased from Sino Biological. 0.2 mg/ ml QC enzyme was diluted 10-fold into a cellulase stock that was 80– 100 μM Cel7A in a 50 mM Na phosphate, pH 7.4. The reaction was then incubated at 30°C for 2 days.
Structural analysis Identification of the location of mutations on the enzymes was done using Visual Molecular Dynamics (VMD) (Humphrey et al., 1996) and Chimera software (Pettersen et al., 2004).
Supplementary data Supplementary data are available at PEDS online.
Funding This work was supported by the Energy Biosciences Institute. We are grateful to Dr. Meera Atreya for providing purified Cel7A from T. reesei Celluclast, to Dr. Harshal Chokhawala for providing T. reesei EGI, to Dr. Joel Graham for providing the EBI 244 enzyme, and to Dr. Christine Roche for expression of enzymes in Neurospora crassa.
References Brayan,P.N., Rollence,M.L., Pantoliano,M.W., Wood,J., Finzel,B.C., Gilliland, G.L., Howard,A.J. and Poulos,T.L. (1986) Proteins Struct. Funct. Bioinform., 1, 326–334. Brennan,T.C.R., Datta,S., Blanch,H.W., Simmons,B.A. and Holmes,B.M. (2010) Bio. Energy Res., 3, 123–133. Carter,J.L.L., Bekhouche,M., Noiriel,A., Blum,L.J. and Doumèche,B. (2014) Chem. Bio. Chem., 15, 2710–2718. Chen,K. and Arnold,F.H. (1993) Proc. Natl Acad. Sci. USA, 90, 5618–5622. Chidambaram,M. and Bell,A.T. (2010) Green Chem, 12, 1253.
Dadi,A.P., Schall,C.A. and Varanasi,S. (2007) Appl. Biochem. Biotechnol., 137–140, 407–421. Dana,C.M., Saija,P., Kal,S.M., Bryan,M.B., Blanch,H.W. and Clark,D.S. (2012) Biotechnol. Bioeng., 109, 2710–2719. Dana,C.M., Dotson-Fagerstrom,A., Roche,C.M., Kal,S.M., Chokhawala,H.A., Blanch,H.W. and Clark,D.S. (2014) Biotechnol. Bioeng., 111, 842–847. Datta,S., Holmes,B., Park,J.I., Chen,Z., Dibble,D.C., Hadi,M., Blanch,H.W., Simmons,B.A. and Sapra,R. (2010) Green Chem., 12, 338–345. Frauenkron-Machedjou,V.J., Fulton,A., Zhu,L., Anker,C., Bocola,M., Jaeger, K.-E. and Schwaneberg,U. (2015) Chem. Bio. Chem., 16, 937–945. Gladden,J.M., Allgaier,M., Miller,C.S., Hazen,T.C., VanderGheynst,J.S., Hugenholtz,P., Simmons,B. and Singer,S.W. (2011) Appl. Environ. Microbiol., 77, 5804–5812. Grassick,A., Murray,P.G., Thompson,R., Collins,C.M., Byrnes,L., Birrane,G., Higgins,T.M. and Tuohy,M.G. (2004) Eur. J. Biochem., 271, 4495–4506. Heinzelman,P., Komor,R., Kanaan,A., Romero,P., Yu,X., Mohler,S., Sno,C. and Arnold,F. (2010) Protein Eng. Design Sel., 23, 871–880. Humphrey,W., Dalke,A. and Schulten,K. (1996) J. Molec. Graphics, 14, 33–38. Kamiya,N., Matsushita,Y., Hanaki,M., Nakashima,K., Narita,M., Goto,M. and Takahashi,H. (2008) Biotechnol. Lett, 30, 1037–1040. Kim,T.-W., Chokhawala,H.A., Nadler,D., Blanch,H.W. and Clark,D.S. (2010) Biotechnol. Bioeng., 107, 601–611. Komor,R.S., Romero,P.A., Xie,C.B. and Arnold,F.H. (2012) Protein Eng.v Design Sel., 25, 827–833. Kotik,M. and Zuber,H. (1993) Eur. J. Biochem., 211, 267–280. Lantz,S.E., Goedegebuur,F., Hommes,R., Kaper,T., Kelemen,B.R., Mitchinson, C., Wallace,L., Ståhlberg,J. and Larenas,E.A. (2010) Biotechnol. Biofuels, 3, 20. Lee,S.H., Doherty,T.V., Linhardt,R.J. and Dordick,J.S. (2009) Biotechnol. Bioeng, 102, 1368–1376. Li,Q., Jiang,X., He,Y., Li,L., Xian,M. and Yang,J. (2010) Appl. Microbiol. Biotechnol., 87, 117–126. Liu,H., Zhu,L., Bocola,M., Chen,N., Spiess,A.C. and Schwaneberg,U. (2013) Green Chem., 15, 1348–1355. Mood,S.H., Golfeshan,A.H., Tabatabaei,M., Abbasalizadeh,S. and Ardjmand, M. (2013) 3 Biotech., 3, 399–406. Park,J.I., Steen,E.J. and Burd,H. (2012) PLoS ONE, 7, e37010. Percival Zhang,Y.H., Himmel,M.E. and Mielenz,J.R. (2006) Biotechnol. Adv., 24, 452–481. Pettersen,E.F., Goddard,T.D., Huang,C.C., Couch,G.S., Greenblatt,D.M., Meng,E.C. and Ferrin,T.E. (2004) J. Comput. Chem., 25, 13, 1605–1612. Robinson,A.S., Hines,V. and Wittrup,K.D. (1994) Nat. Biotech., 12, 381–384. Russell,R.J.M. and Taylor,G.L. (1995) Curr. Opin. Biotechnol., 6, 370–374. Singer,S.W., Reddy,A.P., Gladden,J.M., Guo,H., Hazen,T.C., Simmons,B.A. and VanderGheynst,J.S. (2011) J. Appl. Microbiol., 110, 1023–1031. Stemmer,W.P.C. (1994) Nature, 370, 389–391. Su,C.-H., Chung,M.-H., Hsieh,H-J., Chang,Y.-K., Ding,J.C. and Wu,H.-M. (2012) J. Taiwan Inst. Chem. Eng., 43, 573–577. Swatloski,R.P., Spear,S.K., Holbrey,J.D. and Rogers,R.D. (2002) J. Am. Chem. Soc., 124, 4974–4975. Turner,M.B., Spear,S.K., Huddleston,J.G., Holbrey,J.D. and Rogers,R.D. (2003) Green. Chem., 5, 443–447. Voutilainen,S.P., Murray,P.G., Tuohy,M.G. and Koivula,A. (2010) Protein Eng. Design Sel., 23, 69–79. Wilson,D. and Irwin,D. (1999) In: Tsao,G., Brainard,A., Bungay,H., Cao,N., Cen,P., Chen,Z., Du,J., Foody,B., Gong,C., Hall,P., et al. (eds), Springer, Berlin/Heidelberg, pp. 1–21. Wolski,P.W., Clark,D.S. and Blanch,H.W. (2011) Green. Chem., 13, 3107–3110. Wood,T.M. (1992) Biochem. Soc. Trans., 20, 46–53. Zhao,H., Baker,G.A., Song,Z., Olubajo,O., Crittle,T. and Peters,D. (2008) Green Chem,, 10, 696–705.
Downloaded from http://peds.oxfordjournals.org/ at Weizmann Institute of Science on February 1, 2016
Using a Nano DSC (TA instruments), we measured melting temperatures of cellulases in the presence of aqueous–IL and in buffer alone. The buffer was 50 mM Na acetate, pH 5.0. Enzyme concentration was ∼0.2 mg/ml in the sample. Scans were done at a step rate of 1°C/min after equilibration and prior degassing of the sample.
P.Wolski et al.
