Advertisement

Improving the thermal stability of cellobiohydrolase Cel7A from Hypocrea jecorina by directed evolution

Open AccessPublished:August 31, 2017DOI:https://doi.org/10.1074/jbc.M117.803270
      Secreted mixtures of Hypocrea jecorina cellulases are able to efficiently degrade cellulosic biomass to fermentable sugars at large, commercially relevant scales. H. jecorina Cel7A, cellobiohydrolase I, from glycoside hydrolase family 7, is the workhorse enzyme of the process. However, the thermal stability of Cel7A limits its use to processes where temperatures are no higher than 50 °C. Enhanced thermal stability is desirable to enable the use of higher processing temperatures and to improve the economic feasibility of industrial biomass conversion. Here, we enhanced the thermal stability of Cel7A through directed evolution. Sites with increased thermal stability properties were combined, and a Cel7A variant (FCA398) was obtained, which exhibited a 10.4 °C increase in Tm and a 44-fold greater half-life compared with the wild-type enzyme. This Cel7A variant contains 18 mutated sites and is active under application conditions up to at least 75 °C. The X-ray crystal structure of the catalytic domain was determined at 2.1 Å resolution and showed that the effects of the mutations are local and do not introduce major backbone conformational changes. Molecular dynamics simulations revealed that the catalytic domain of wild-type Cel7A and the FCA398 variant exhibit similar behavior at 300 K, whereas at elevated temperature (475 and 525 K), the FCA398 variant fluctuates less and maintains more native contacts over time. Combining the structural and dynamic investigations, rationales were developed for the stabilizing effect at many of the mutated sites.

      Introduction

      Enzymatic conversion of cellulosic biomass, a renewable feedstock rich in carbon, to fermentable sugars is an important process in the production of sustainable alternative fuels and chemicals (
      • Chundawat S.P.S.
      • Beckham G.T.
      • Himmel M.E.
      • Dale B.E.
      Deconstruction of lignocellulosic biomass to fuels and chemicals.
      ). Biomass-utilizing microorganisms employ a variety of enzymes to degrade the complex carbohydrate cellulose, a polymer of β-1,4-glucose (
      • Payne C.M.
      • Knott B.C.
      • Mayes H.B.
      • Hansson H.
      • Himmel M.E.
      • Sandgren M.
      • Ståhlberg J.
      • Beckham G.T.
      Fungal cellulases.
      ). One such microorganism is the filamentous fungus Hypocrea jecorina (also known by the name of its anamorph, Trichoderma reesei), which is among the most prolific producers of biomass-degrading enzymes (
      • Kubicek C.P.
      Systems biological approaches towards understanding cellulase production by Trichoderma reesei.
      ). The cellulolytic system expressed by H. jecorina includes a complex of endoglucanases (Cel7B/EG
      The abbreviations used are: EG
      endoglucanase
      CBH
      cellobiohydrolase
      CBM
      carbohydrate-binding module
      GH7
      glycoside hydrolase family 7
      MD
      molecular dynamics
      MUL
      4-methylumbelliferyl-β-lactoside
      NAG
      N-acetyl glucosamine
      PASC
      phosphoric acid swollen cellulose
      RMSD
      root mean square deviation
      RMSF
      root mean square fluctuation
      PDB
      Protein Data Bank
      CV
      column volumes.
      I, Cel5A/EG II, Cel5B, Cel12A/EG III, Cel45A/EG V, and Cel74A) and two cellobiohydrolases (CBHs), which processively hydrolyze from either the reducing end (Cel7A/CBH I) or the non-reducing end (Cel6A/CBH II) of cellulose chains. Together, these enzymes synergistically break down cellulose into soluble sugars (
      • Okada H.
      • Tada K.
      • Sekiya T.
      • Yokoyama K.
      • Takahashi A.
      • Tohda H.
      • Kumagai H.
      • Morikawa Y.
      Molecular characterization and heterologous expression of the gene encoding a low-molecular-mass endoglucanase from Trichoderma reesei QM9414.
      • Saloheimo M.
      • Lehtovaara P.
      • Penttilä M.
      • Teeri T.T.
      • Ståhlberg J.
      • Johansson G.
      • Pettersson G.
      • Claeyssens M.
      • Tomme P.
      • Knowles J.K.
      EGIII, a new endoglucanase from Trichoderma reesei: the characterization of both gene and enzyme.
      ,
      • Saloheimo A.
      • Henrissat B.
      • Hoffrén A.M.
      • Teleman O.
      • Penttilä M.
      A novel, small endoglucanase gene, egl5, from Trichoderma reesei isolated by expression in yeast.
      ,
      • Saloheimo M.
      • Nakari-Setälä T.
      • Tenkanen M.
      • Penttilä M.
      cDNA cloning of a Trichoderma reesei cellulase and demonstration of endoglucanase activity by expression in yeast.
      ,
      • Teeri T.T.
      • Lehtovaara P.
      • Kauppinen S.
      • Salovuori I.
      • Knowles J.
      Homologous domains in Trichoderma reesei cellulolytic enzymes: gene sequence and expression of cellobiohydrolase II.
      ,
      • Ward M.
      • Wu S.
      • Dauberman J.
      • Weiss G.
      • Larenas E.
      • Bower B.
      • Rey M.
      • Clarkson K.
      • Bott R.
      Cloning, sequence and preliminary structural analysis of a small, high pI endoglucanase (EGIII) from Trichoderma reesei.
      • Foreman P.K.
      • Brown D.
      • Dankmeyer L.
      • Dean R.
      • Diener S.
      • Dunn-Coleman N.S.
      • Goedegebuur F.
      • Houfek T.D.
      • England G.J.
      • Kelley A.S.
      • Meerman H.J.
      • Mitchell T.
      • Mitchinson C.
      • Olivares H.A.
      • Teunissen P.J.
      • et al.
      Transcriptional regulation of biomass-degrading enzymes in the filamentous fungus Trichoderma reesei.
      ). Beyond these primary cellulose-degrading enzymes, H. jecorina is known (
      • Foreman P.K.
      • Brown D.
      • Dankmeyer L.
      • Dean R.
      • Diener S.
      • Dunn-Coleman N.S.
      • Goedegebuur F.
      • Houfek T.D.
      • England G.J.
      • Kelley A.S.
      • Meerman H.J.
      • Mitchell T.
      • Mitchinson C.
      • Olivares H.A.
      • Teunissen P.J.
      • et al.
      Transcriptional regulation of biomass-degrading enzymes in the filamentous fungus Trichoderma reesei.
      ,
      • Barnett C.C.
      • Berka R.M.
      • Fowler T.
      Cloning and amplification of the gene encoding an extracellular β-glucosidase from Trichoderma reesei: evidence for improved rates of saccharification of cellulosic substrates.
      • Takashima S.
      • Iikura H.
      • Nakamura A.
      • Hidaka M.
      • Masaki H.
      • Uozumi T.
      Comparison of gene structures and enzymatic properties between two endoglucanases from Humicola grisea.
      ) to produce seven β-glucosidases (Cel1A, Cel1B, Cel3A Cel3B, Cel3C, Cel3D, and Cel3E), at least two lytic polysaccharide monooxygenases (LPMO9A/Cel61A and LPMO9B/Cel61B), a swollenin, a glucuronyl esterase, and a lyase-like protein (
      • Li X.L.
      • Spániková S.
      • de Vries R.P.
      • Biely P.
      Identification of genes encoding microbial glucuronoyl esterases.
      ,
      • Jacobson F.
      • Karkehabadi S.
      • Hansson H.
      • Goedegebuur F.
      • Wallace L.
      • Mitchinson C.
      • Piens K.
      • Stals I.
      • Sandgren M.
      The crystal structure of the core domain of a cellulose induced protein (Cip1) from Hypocrea jecorina, at 1.5 Å resolution.
      ), all of which play a role in cellulose degradation. The four most abundant components of the cellulolytic system of H. jecorina (Cel7A, Cel6A, Cel7B, and Cel5A) account for >75% of the protein produced by this fungus under inducing conditions and can be secreted in excess of 50 g/liter (
      • Cherry J.R.
      • Fidantsef A.L.
      Directed evolution of industrial enzymes: an update.
      ). The major component, Cel7A, constitutes nearly half of total secreted protein, and it has been shown to be a major rate-limiting factor in cellulose degradation (
      • Payne C.M.
      • Knott B.C.
      • Mayes H.B.
      • Hansson H.
      • Himmel M.E.
      • Sandgren M.
      • Ståhlberg J.
      • Beckham G.T.
      Fungal cellulases.
      ,
      • Suominen P.L.
      • Mäntylä A.L.
      • Karhunen T.
      • Hakola S.
      • Nevalainen H.
      High-frequency one-step gene replacement in Trichoderma reesei. 2. Effects of deletions of individual cellulase genes.
      ), suggesting that GH7 enzymes play a key role in biomass-degrading fungi (
      • Ilmén M.
      • Saloheimo A.
      • Onnela M.L.
      • Penttilä M.E.
      Regulation of cellulase gene expression in the filamentous fungus Trichoderma reesei.
      ). Given its importance in biomass deconstruction, Cel7A has long been a primary component of industrial biomass-degrading enzyme mixtures (
      • Chundawat S.P.S.
      • Lipton M.S.
      • Purvine S.O.
      • Uppugundla N.
      • Gao D.
      • Balan V.
      • Dale B.E.
      Proteomics-based compositional analysis of complex cellulase-hemicellulase mixtures.
      ), and H. jecorina is the predominant host for such applications as a result of its high-titer production of cellulases (
      • Himmel M.E.
      • Ding S.Y.
      • Johnson D.K.
      • Adney W.S.
      • Nimlos M.R.
      • Brady J.W.
      • Foust T.D.
      Biomass recalcitrance: engineering plants and enzymes for biofuels production.
      ).
      Industrial enzymatic conversion efficiency benefits from increases in temperature by virtue of the fundamental dependence of reaction rate on temperature. However, H. jecorina is a mesophilic organism that produces cellulases, including its native Cel7A, that do not have sufficient stability beyond 50 °C (
      • Momeni M.H.
      • Goedegebuur F.
      • Hansson H.
      • Karkehabadi S.
      • Askarieh G.
      • Mitchinson C.
      • Larenas E.A.
      • Ståhlberg J.
      • Sandgren M.
      Expression, crystal structure and cellulase activity of the thermostable cellobiohydrolase Cel7A from the fungus Humicola grisea var. thermoidea.
      ). Thus, process optimization would greatly benefit from identification of increasingly thermally stable cellulase variants, where improvements to Cel7A stability, the most abundant component of industrial enzyme mixtures (
      • Chundawat S.P.S.
      • Lipton M.S.
      • Purvine S.O.
      • Uppugundla N.
      • Gao D.
      • Balan V.
      • Dale B.E.
      Proteomics-based compositional analysis of complex cellulase-hemicellulase mixtures.
      ), offer the largest overall gain.
      There are four commonly implemented approaches to identify stabilizing mutations of a given protein. These approaches include (i) comparison of the amino acid sequence of a protein with that of a single more or less thermally stable homologous protein, followed by replacement of selected amino acids by site-directed mutagenesis (
      • Lehmann M.
      • Loch C.
      • Middendorf A.
      • Studer D.
      • Lassen S.F.
      • Pasamontes L.
      • van Loon A.P.
      • Wyss M.
      The consensus concept for thermostability engineering of proteins: further proof of concept.
      ,
      • van den Burg B.
      • Dijkstra B.W.
      • van der Vinne B.
      • Stulp B.K.
      • Eijsink V.G.
      • Venema G.
      Introduction of disulfide bonds into Bacillus subtilis neutral protease.
      ); (ii) site-directed mutagenesis to introduce amino acids derived from the study of the 3D structure of the protein of interest, where proteins with diverse thermal stabilities are compared and amino acid substitutions identified (
      • Sandgren M.
      • Gualfetti P.J.
      • Shaw A.
      • Gross L.S.
      • Saldajeno M.
      • Day A.G.
      • Jones T.A.
      • Mitchinson C.
      Comparison of family 12 glycoside hydrolases and recruited substitutions important for thermal stability.
      ,
      • Goedegebuur F.
      • Fowler T.
      • Phillips J.
      • van der Kley P.
      • van Solingen P.
      • Dankmeyer L.
      • Power S.D.
      Cloning and relational analysis of 15 novel fungal endoglucanases from family 12 glycosyl hydrolase.
      • Sammond D.W.
      • Kastelowitz N.
      • Himmel M.E.
      • Yin H.
      • Crowley M.F.
      • Bomble Y.J.
      Comparing residue clusters from thermophilic and mesophilic enzymes reveals adaptive mechanisms.
      ); (iii) random mutagenesis followed by selection (
      • Lehmann M.
      • Pasamontes L.
      • Lassen S.F.
      • Wyss M.
      The consensus concept for thermostability engineering of proteins.
      ,
      • Kim P.
      • Yoon S.H.
      • Seo M.J.
      • Oh D.K.
      • Choi J.H.
      Improvement of tagatose conversion rate by genetic evolution of thermostable galactose isomerase.
      • Sakaue R.
      • Kajiyama N.
      Thermostabilization of bacterial fructosyl-amino acid oxidase by directed evolution.
      ); and (iv) a structure-guided approach with recombination of stabilizing fragments (SCHEMA) (
      • Meyer M.M.
      • Hochrein L.
      • Arnold F.H.
      Structure-guided SCHEMA recombination of distantly related β-lactamases.
      ,
      • Heinzelman P.
      • Snow C.D.
      • Wu I.
      • Nguyen C.
      • Villalobos A.
      • Govindarajan S.
      • Minshull J.
      • Arnold F.H.
      A family of thermostable fungal cellulases created by structure-guided recombination.
      ). The latter two approaches rely on the ability to both express the enzyme in a microorganism and assay it using high-throughput screening methods. However, heterologous expression of Cel7A from H. jecorina in Escherichia coli, a traditional high-throughput expression host, has notoriously failed to produce either soluble or active enzyme. In yeast, expression levels are low, and the properties of expressed Cel7s have been unpredictable (
      • Godbole S.
      • Decker S.R.
      • Nieves R.A.
      • Adney W.S.
      • Vinzant T.B.
      • Baker J.O.
      • Thomas S.R.
      • Himmel M.E.
      Cloning and expression of Trichoderma reesei cellobiohydrolase I in Pichia pastoris.
      • Laymon R.A.
      • Adney W.S.
      • Mohagheghi A.
      • Himmel M.E.
      • Thomas S.R.
      Cloning and expression of full-length Trichoderma reesei cellobiohydrolase I cDNAs in Escherichia coli.
      ,
      • Reinikainen T.
      • Ruohonen L.
      • Nevanen T.
      • Laaksonen L.
      • Kraulis P.
      • Jones T.A.
      • Knowles J.K.
      • Teeri T.T.
      Investigation of the function of mutated cellulose-binding domains of Trichoderma reesei cellobiohydrolase I.
      • Penttilä M.E.
      • André L.
      • Lehtovaara P.
      • Bailey M.
      • Teeri T.T.
      • Knowles J.K.
      Efficient secretion of two fungal cellobiohydrolases by Saccharomyces cerevisiae.
      ). The latter obstacle is probably a result of both the lack of N-terminal glutamine cyclization of Cel7A expressed in yeast and the significant variation in protein glycosylation (
      • Dana C.M.
      • Dotson-Fagerstrom A.
      • Roche C.M.
      • Kal S.M.
      • Chokhawala H.A.
      • Blanch H.W.
      • Clark D.S.
      The importance of pyroglutamate in cellulase Cel7A.
      ), highlighting the importance of expression host on enzyme activity.
      To date, approaches ii and iv have been successfully applied in engineering more thermally stable GH7 cellobiohydrolases. Heinzelman et al. (
      • Heinzelman P.
      • Komor R.
      • Kanaan A.
      • Romero P.
      • Yu X.
      • Mohler S.
      • Snow C.
      • Arnold F.
      Efficient screening of fungal cellobiohydrolase class I enzymes for thermostabilizing sequence blocks by SCHEMA structure-guided recombination.
      ) used SCHEMA recombination to increase the half-life of a chimeric GH7 expressed in a glycosylation-deficient Saccharomyces cerevisiae strain. The variant, which was reported to contain an average of 42 unspecified mutations, retained some hydrolytic activity on soluble cellulose at temperatures up to 70 °C. In a follow-up study, the inclusion of an additional eight mutations improved the T50 (temperature at which a 10-min incubation without substrate results in loss of half of the activity) to 72.1 °C (
      • Komor R.S.
      • Romero P.A.
      • Xie C.B.
      • Arnold F.H.
      Highly thermostable fungal cellobiohydrolase I (Cel7A) engineered using predictive methods.
      ). In a similar approach, chimeric libraries of Cel7A (not including H. jecorina Cel7A) were expressed in S. cerevisiae and found to contain stabilized variants (
      • Dana C.M.
      • Saija P.
      • Kal S.M.
      • Bryan M.B.
      • Blanch H.W.
      • Clark D.S.
      Biased clique shuffling reveals stabilizing mutations in cellulase Cel7A.
      ). Voutilainen et al. have also described protein engineering efforts via approach ii, resulting in a 4.5 °C and 9 °C increase in Tm for Melanocarpus albomyces Cel7B and Talaromyces emersonii Cel7A, respectively (
      • Voutilainen S.P.
      • Boer H.
      • Alapuranen M.
      • Jänis J.
      • Vehmaanperä J.
      • Koivula A.
      Improving the thermostability and activity of Melanocarpus albomyces cellobiohydrolase Cel7B.
      ,
      • Voutilainen S.P.
      • Murray P.G.
      • Tuohy M.G.
      • Koivula A.
      Expression of Talaromyces emersonii cellobiohydrolase Cel7A in Saccharomyces cerevisiae and rational mutagenesis to improve its thermostability and activity.
      ). Although these studies demonstrate that GH7 enzyme thermal stability can be improved through a variety of protein engineering approaches, we do not yet understand why these mutations, either singular or multiple, result in improved stability. Moreover, there has only been one report of improved thermal stability of H. jecorina Cel7A in the scientific literature (
      • Smith M.A.
      • Bedbrook C.N.
      • Wu T.
      • Arnold F.H.
      Hypocrea jecorina cellobiohydrolase I stabilizing mutations identified using noncontiguous recombination.
      ).
      As part of a large, broad campaign to find stabilizing mutations in H. jecorina Cel7A (
      • Day A.
      • Goedegebuur F.
      • Gualfetti P.
      • Mitchinson C.
      • Neefe P.
      • Sandgren M.
      • Shaw A.
      • Stahlberg J.
      Variant Hypocrea jecorina CBH1 cellulases.
      ), we developed a host/vector system for heterologous expression of Cel7A in the filamentous fungus Aspergillus niger var. awamori AP4, using a constitutive promoter to limit the background of host proteins and potential interference from other carbohydrases. We demonstrated the successful expression of GH7 CBHs from various fungi in the A. niger AP4 system (
      • Momeni M.H.
      • Goedegebuur F.
      • Hansson H.
      • Karkehabadi S.
      • Askarieh G.
      • Mitchinson C.
      • Larenas E.A.
      • Ståhlberg J.
      • Sandgren M.
      Expression, crystal structure and cellulase activity of the thermostable cellobiohydrolase Cel7A from the fungus Humicola grisea var. thermoidea.
      ). Using approaches i, ii, and iii above, a collection of H. jecorina Cel7A variants was generated and found to be stable by thermal inactivation screening (
      • Day A.
      • Goedegebuur F.
      • Gualfetti P.
      • Mitchinson C.
      • Neefe P.
      • Sandgren M.
      • Shaw A.
      • Stahlberg J.
      Variant Hypocrea jecorina CBH1 cellulases.
      ). In this paper we characterize the most stable of those variants, assessing their thermal stability, residual activity after incubation at elevated temperatures using a model substrate, and activity on amorphous cellulose. We also study combinations of the variants with further improvements in stability. This combinatorial approach has proven effective in other systems (
      • Kanaya S.
      Stabilization of an enzyme with protein engineering technology.
      ,
      • Gray K.A.
      • Richardson T.H.
      • Kretz K.
      • Short J.M.
      • Bartnek F.
      • Knowles R.
      • Kan L.
      • Swanson P.E.
      • Robertson D.E.
      Rapid evolution of reversible denaturation and elevated melting temperature in a microbial haloalkane.
      ). To address the fundamental gap in understanding of structural and dynamic contributions to GH7 thermal stability, specifically that of H. jecorina Cel7A, we focus on the most stable Cel7A variant, FCA398, with 18 mutations, solving its crystal structure and using high-temperature molecular simulation to evaluate changes in its structural dynamics compared with wild type.
      Overall, our study reveals that it is possible to make significant improvements to the thermal stability of H. jecorina Cel7A by the accumulation of many mutations, and through a combination of structural and dynamic approaches, we develop rationales for the beneficial effects of most of the mutations made.

      Results

      Expression of thermostabilized H. jecorina Cel7A variants in A. niger var. awamori

      To improve the thermal stability of H. jecorina Cel7A, mutations have been introduced in the protein molecule using protein engineering. Following high-throughput fungal expression and screening, a large set of point mutations resulting in enhanced residual activity after incubation at elevated temperature was identified (
      • Day A.
      • Goedegebuur F.
      • Gualfetti P.
      • Mitchinson C.
      • Neefe P.
      • Sandgren M.
      • Shaw A.
      • Stahlberg J.
      Variant Hypocrea jecorina CBH1 cellulases.
      ). In the present study, a group of these were chosen for larger-scale expression in A. niger var. awamori AP4 and purification for more detailed analyses (Table 1), including wild-type Cel7A (FCA301) as a control. Site-specific combination mutants were then generated, expressed, and screened, leading to a Cel7A variant (FCA398) with 18 mutations and substantially improved thermal stability. Herein, we present the results for the most thermally stable Cel7A variant, FCA398, and key variants in its development (Table 2).
      Table 1Protein Tm for selected variants of H. jecorina Cel7A
      VariantMutationsTm
      Tm is unfolding temperature, measured by CD spectroscopy; ΔTm is Tm value minus the Tm value of FCA301 (wild-type Cel7A).
      ΔTm
      Tm is unfolding temperature, measured by CD spectroscopy; ΔTm is Tm value minus the Tm value of FCA301 (wild-type Cel7A).
      °C°C
      FCA301None62.50
      FCA330S8P63.7+1.2
      FCA328G22D/S278P/T296P63.6+1.1
      FCA335T41I64.2+1.7
      FCA332N49S63.7+1.2
      FCA347A68T63.7+1.2
      FCA377N89D63.5+1.0
      FCA374S92T64.4+1.9
      FCA326S113N64.0+1.5
      FCA375S196T/S411F65.3+2.8
      FCA334P227L65.2+2.7
      FCA373P227A64.8+2.3
      FCA349D249K64.0+1.5
      FCA376T255P64.4+1.9
      FCA384E295K64.0+1.5
      FCA372S297L/V403D/T462I64.5+2.0
      FCA468N301R63.0+0.5
      FCA369T332Y63.3+0.8
      a Tm is unfolding temperature, measured by CD spectroscopy; ΔTm is Tm value minus the Tm value of FCA301 (wild-type Cel7A).
      Table 2Tm and t½ values for selected Cel7A site combination variants
      Cel7A variantMutationst½TmΔTm
      62 °C66 °C69 °C
      min°C°C
      FCA301None19.99.38.662.50
      FCA353G22D/N49S/A68T/P227L/S278P/T296P223ND
      Not determined.
      8.365.9+3.4
      FCA367S8P/G22D/T41I/N49S/A68T/S113N/P227L /D249K/S278P/T296P/N301R59618253.969.9+7.4
      FCA398S8P/T41I/N49S/A68T/N89D/S92T/S113N /S196T/P227L/D249K/T255P/S278P/E295K /T296P/T332Y/V403D/S411F/T462I86631575.372.9+10.4
      a Not determined.

      Determination of protein Tm values

      Thermal stability was assessed by monitoring the irreversible thermal denaturation of the purified Cel7A variants with CD spectroscopy for determination of protein Tm. The obtained Tm values and ΔTm, compared with wild type, are shown in Table 1, Table 2. For comparison, Tm was also determined for Cel7A wild type isolated from a native H. jecorina culture; the Tm of natively expressed Cel7A, 62.4 ± 0.3 °C, was the same as that of FCA301 (wild-type Cel7A expressed in A. niger var. awamori), 62.5 ± 0.3 °C. All the analyzed variants exhibited higher Tm than the wild-type enzyme, as expected from their identification as more thermally stable in the residual activity screening. In general, the stabilizing effect of individual point mutations was small, where ΔTm was in the range of +1 to +2 °C for a majority, up to +2.7 °C for P227L (FCA334), and +2.8 °C for S196F/S411F (FCA375) (Table 1). When a larger number of sites were combined, a significant enhancement of Tm was achieved, as exemplified by FCA353, FCA367, and FCA398 in which 6, 11, and 18 mutations, respectively, increased the Tm by 3.4, 7.4 and 10.4 °C (Table 2).
      The contribution of single-site variants to the thermal stability of the corresponding multi-site combinations was not simply additive. That is, the ΔTm of combinatorial variants was consistently lower than the direct sum of ΔTm values for the contributing single-site variants. For instance, one of our best intermediate variants (supplemental Fig. S1), FCA367, with 11 mutations, had a measured ΔTm of +7.4 °C, whereas the summative ΔTm of the individual 11 mutations would be +11.4 °C if thermal stability gains were purely additive. To rule out the possibility that some mutations may counteract thermal stability when combined with each other, all 11 sites of FCA367 were recombined using QuikChange, generating a library of Cel7A mutants with different combinations of these sites. Around 2000 clones, with an average of 8 mutations/clone (990 possibilities), were screened for retention of activity at high temperature. The most thermally stable clone contained all 11 sites, suggesting that all of the point mutations of FCA367 contribute to increased thermal stability. FCA367 was further improved by the addition or substitution of stability-associated mutations using fusion PCR techniques to generate site-specific combination mutants with enhanced thermal stability. The most improved variant was FCA398, which exhibits a Tm value of 72.9 °C (i.e. 10.4 °C higher than wild-type Cel7A) (Table 2).

      Thermal inactivation and half-life, t½, at elevated temperature

      Residual activity measurements were further used to determine inactivation t½ for the site-combined variants FCA353, FCA367, and FCA398 compared with wild-type FCA301. The activity of the purified enzymes on 4-methylumbelliferyl-β-lactoside (MUL) was measured at 50 °C, before and after incubation at elevated temperatures, 62, 66, and 69 °C. Samples were taken at regular time points up to 20 h for activity measurements on MUL. This experimental design measured irreversible thermal inactivation. The half-life is defined as the time taken for a decrease of activity to 50% of the initial activity. The results show that the Cel7A variants were more stable than wild type, in agreement with measured Tm values. Even a seemingly small increase in Tm substantially improves the lifetime of the enzyme at high temperature (in the case of FCA353), and for FCA398, the most stable variant, we see large improvements (Table 2); the half-life is 44-, 34-, and 9-fold enhanced compared with wild type at 62, 66, and 69 °C, respectively.

      Degradation of phosphoric acid swollen cellulose (PASC) by Cel7A variant FCA398 at elevated temperature

      Wild-type FCA301 and the most stable Cel7A variant, FCA398, were incubated at 53, 65, and 75 °C with 1% PASC as substrate, and the cellulose degradation was monitored over the course of 3 days (72 h) by quantification of released soluble sugar by HPLC (Fig. 1). FCA398 is clearly more active and gave higher sugar yields at all three temperatures than did FCA301 at any temperature. For FCA301, the activity was highest at 53 °C and slightly lower at 65 °C, whereas the enzyme was practically inactive at 75 °C. FCA398 retained relatively high activity even at 75 °C, leveling off after ∼48 h, and its highest activity was observed at 65 °C.
      Figure thumbnail gr1
      Figure 1Hydrolysis of PASC by the most stable Cel7A variant FCA398 (filled symbols) compared with wild type (FCA301; open symbols) at 53 °C (diamonds), 65 °C (triangles), and 75 °C (squares), using 1% PASC and 0.5 mg enzyme/g cellulose at pH 5.5. The amount of soluble sugar was determined by HPLC.

      3D structure of the catalytic domain of the Cel7A FCA398 variant

      To understand the structural basis for improved thermal stability over wild type, the structure of the deglycosylated catalytic domain of FCA398 was determined by X-ray crystallography (Fig. 2). The protein crystallized in space group P21212 with two protein chains in the asymmetric unit, in a new crystal form with unit cell parameters and crystal packing that are different from that of any previous structure of H. jecorina Cel7A. The structure of the FCA398 catalytic domain was solved by molecular replacement and was refined at 2.1 Å resolution to final R and Rfree values of 19.3 and 23.9%, respectively. Statistics on data collection, refinement, and the final structure model are given in Table 3. Ramachandran plots are provided in supplemental Fig. S2). The structure model has been deposited with the Protein Data Bank (PDB accession code 5OA5).
      Figure thumbnail gr2
      Figure 2Overview of the 18 mutation sites (magenta) in the H. jecorina Cel7A FCA398 variant and comparison of the crystal structure of the catalytic domain of FCA398 (yellow) with that of Cel7A (light gray) in complex with cellononaose (green; PDB code 4C4C) (
      • Knott B.C.
      • Haddad Momeni M.
      • Crowley M.F.
      • Mackenzie L.F.
      • Götz A.W.
      • Sandgren M.
      • Withers S.G.
      • Ståhlberg J.
      • Beckham G.T.
      The mechanism of cellulose hydrolysis by a two-step, retaining cellobiohydrolase elucidated by structural and transition path sampling studies.
      ). A, the linker–CBM region (light blue) from a previous model of the full-length Cel7A enzyme from MD simulation (
      • Payne C.M.
      • Jiang W.
      • Shirts M.R.
      • Himmel M.E.
      • Crowley M.F.
      • Beckham G.T.
      Glycoside hydrolase processivity is directly related to oligosaccharide binding free energy.
      ) shows the location of the T462I mutation at the linker–CBM junction. B, view showing the cellulose-binding tunnel entrance side of the catalytic domain. C, view showing the product binding side of the catalytic domain.
      Table 3X-ray data collection and refinement statistics for the structure of H. jecorina Cel7A variant FCA398
      ParameterValue
      PDB code5OA5
      Diffraction sourceESRF ID23-2
      Wavelength (Å)0.8726
      Temperature (K)100
      DetectorCCD MAR225
      Rotation range per image (degrees)0.5
      Space groupP21212
      a, b, c (Å)102.94, 92.10, 102.20
      α, β, γ (degrees)90, 90, 90
      Resolution range (Å)41–2.1 (2.21–2.10)
      Total no. of reflections211,322
      No. of unique reflections56,990
      Completeness (%)99.6 (99.6)
      Redundancy3.71 (3.73)
      I/σ(I)〉7.2 (3.0)
      Rmeas
      Rmeas = Σ √(n/(n − 1))|IhI − 〈Ih〉|/Σ|〈Ih〉|.
      0.15 (0.47)
      Resolution range in refinement (Å)40–2.1 (2.15–2.10)
      σ cut-off2.0
      No. of reflections, working set53,983 (3975)
      No. of reflections, test set2882 (183)
      Final Rcryst0.197 (0.256)
      Final Rfree0.238 (0.306)
      No. of non-hydrogen atoms
          Protein6480
          Non-protein90
          Water546
      RMSD
          Bonds (Å)0.013
          Angles (degrees)1.65
      Average B factors (Å2)
          Overall17.7
          Protein16.9
          Water25.0
      Stringent Ramachandran outliers (%)
      According to the stringent boundary definition of Kleywegt and Jones (101).
      1.7
      a Rmeas = Σ √(n/(n − 1))|IhI − 〈Ih〉|/Σ|〈Ih〉|.
      b According to the stringent boundary definition of Kleywegt and Jones (
      • Kleywegt G.J.
      • Jones T.A.
      Phi/Psi-cology: Ramachandran revisited.
      ).
      Both protein chains, A and B, of the asymmetric unit contain all amino acid residues (positions 1–434) of the FCA398 catalytic domain and three N-acetyl glucosamine residues (NAG), which are covalently linked to Asn64, Asn113, and Asn270, respectively. Glycosylation at Asn64, as observed in this Aspergillus-expressed protein, has not been seen previously in H. jecorina Cel7A structures or glycosylation analyses (
      • Stals I.
      • Sandra K.
      • Devreese B.
      • Van Beeumen J.
      • Claeyssens M.
      Factors influencing glycosylation of Trichoderma reesei cellulases. II: N-glycosylation of Cel7A core protein isolated from different strains.
      ), despite the presence of an N-glycosylation sequence motif at this site also in the wild-type enzyme. At Asn113, a new N-glycosylation motif is introduced by the S113N mutation in FCA398.
      The structures of the two chains are practically identical with a root mean square deviation (RMSD) of 0.13 Å. The FCA398 structure is very similar to previously deposited H. jecorina Cel7A catalytic domain structures; superposition with the Cel7A-cellononaose complex (PDB code 4C4C (
      • Knott B.C.
      • Crowley M.F.
      • Himmel M.E.
      • Ståhlberg J.
      • Beckham G.T.
      Carbohydrate–protein interactions that drive processive polysaccharide translocation in enzymes revealed from a computational study of cellobiohydrolase processivity.
      )) gives RMSD values of 0.35 and 0.34 Å relative to FCA398 chain A and B, respectively. There is only one loop region, 244–254, where the main chain of FCA398 deviates significantly. This loop bends toward and partially encloses the catalytic center of the enzyme. It has previously been referred to as the “exo-loop” (
      • von Ossowski I.
      • Ståhlberg J.
      • Koivula A.
      • Piens K.
      • Becker D.
      • Boer H.
      • Harle R.
      • Harris M.
      • Divne C.
      • Mahdi S.
      • Zhao Y.
      • Driguez H.
      • Claeyssens M.
      • Sinnott M.L.
      • Teeri T.T.
      Engineering the exo-loop of Trichoderma reesei cellobiohydrolase, Cel7A: a comparison with Phanerochaete chrysosporium Cel7D.
      ) and, more recently, loop B3 (
      • Momeni M.H.
      • Payne C.M.
      • Hansson H.
      • Mikkelsen N.E.
      • Svedberg J.
      • Engström Å.
      • Sandgren M.
      • Beckham G.T.
      • Ståhlberg J.
      Structural, biochemical, and computational characterization of the glycoside hydrolase family 7 cellobiohydrolase of the tree-killing fungus Heterobasidion irregulare.
      ), and flexibility in the loop has been reported previously (
      • von Ossowski I.
      • Ståhlberg J.
      • Koivula A.
      • Piens K.
      • Becker D.
      • Boer H.
      • Harle R.
      • Harris M.
      • Divne C.
      • Mahdi S.
      • Zhao Y.
      • Driguez H.
      • Claeyssens M.
      • Sinnott M.L.
      • Teeri T.T.
      Engineering the exo-loop of Trichoderma reesei cellobiohydrolase, Cel7A: a comparison with Phanerochaete chrysosporium Cel7D.
      ,
      • Divne C.
      • Ståhlberg J.
      • Teeri T.T.
      • Jones T.A.
      High-resolution crystal structures reveal how a cellulose chain is bound in the 50 Å long tunnel of cellobiohydrolase I from Trichoderma reesei.
      ,
      • Muńoz I.G.
      • Mowbray S.L.
      • Stahlberg J.
      The catalytic module of Cel7D from Phanerochaete chrysosporium as a chiral selector: structural studies of its complex with the β blocker (R)-propranolol.
      ). The largest backbone shift between the FCA398 and 4C4C structures is seen at residue 249 (4.6 Å between Cα atoms), which is the site of the D249K mutation (Fig. 2, A and C).

      Molecular dynamics (MD) simulations of Cel7A FCA301 and the FCA398 variant

      Molecular dynamics simulations of wild-type Cel7A and FCA398 at 300, 475, and 525 K suggest that the 17 mutations in the FCA398 catalytic domain improve thermal stability by retaining tighter packing of the core domain at elevated temperatures. This is illustrated by two separate analyses of the triplicate simulation trajectories: root mean square fluctuation (RMSF) of the protein backbone and variation in native contacts maintained over time. First, the RMSF of the protein backbone was determined for each of the 18 simulation trajectories. The RMSF of the triplicate simulations were averaged together to obtain an average RMSF of a given enzyme at a given temperature (Fig. 3). The RMSF measures deviation of a residue from its time-averaged position over the course of a simulation. Higher RMSF in the protein backbone typically indicates a more flexible, potentially less thermally stable, region of the protein. At 300 K, the average RMSF of wild type and FCA398 are practically indistinguishable. However, with increasing temperature (475 and 525 K), the FCA398 variant fluctuates less in comparison with wild type, implying that the 17 selected mutations in the catalytic domain increased the stability of the protein. At 475 K, the stabilizing effect of the set of mutations had the most significant impact in the following residue ranges: 243–256, 310–358, and 378–396. At 525 K, we observe the same stabilization effects as in the set of simulations at 475 K, but also the region between 260 and 285 benefits from the stabilizing effects of the FCA398 mutations.
      Figure thumbnail gr3
      Figure 3RMSF of the wild-type Cel7A and FCA398 variant backbone from molecular dynamics simulations at 300, 475, and 525 K. In all cases, the RMSF shown was determined by averaging values from three independent simulation trajectories. At 300 K, both wild type and the variant exhibit nearly identical dynamic flexibility, but at high temperatures, wild-type Cel7A is prone to higher backbone fluctuation relative to the FCA398 variant. In the top right panel, the wild-type Cel7A structure is shown colored by its RMSF at 300 K. The red regions represent backbone segments with the greatest fluctuation, whereas blue indicates the lowest fluctuation. The RMSF values shown on the structure have been scaled to a maximum of 2.5. Loops are labeled in black type on the structure.
      Quantitative determination of the total number of native contacts from MD simulations illustrates that FCA398 retained important molecular interactions within the core of the protein more effectively than wild-type Cel7A when faced with thermal stress. First, the number of native contacts formed by each residue was identified. Here, a native contact was defined as any amino acid whose side chain center of geometry was within 6.5 Å of the reference amino residue's Cα. The total number of native contacts (Fig. 4) is the sum of the native contacts formed by all residues in the protein. The total numbers of native contacts formed by both wild-type Cel7A and FCA398 were identical at the 300 K reference temperature over the entirety of the 50-ns MD simulations, although only 15 ns is shown in Fig. 4 for comparison with high temperature behavior. As expected at high temperatures, the protein began to partially unfold at ∼7 ns, destroying native contacts present at 300 K. Regions having low initial numbers of native contacts began to unfold first during high-temperature simulations (supplemental Fig. S3), corresponding well with the increasing regional fluctuation described above.
      Figure thumbnail gr4
      Figure 4Total number of native contacts retained by wild-type Cel7A and FCA398 mutant over the course of MD simulations at 300, 475, and 525 K. The lines represent the average of three independent MD simulations for each enzyme at a given temperature.
      The relative difference in number of native contacts retained over the 15-ns high-temperature simulations is easiest to see by examining the difference in the number of native contacts formed by each residue over the entire simulation. For both 475 and 525 K, the native contact heat maps of wild-type Cel7A and FCA398 have been subtracted from each other to reveal regions in which FCA398 maintains a greater number of native contacts at high temperature (supplemental Fig. S4). The region between residues 320 and 340 appears to benefit the most from the stabilizing effect of FCA398 mutations, forming and maintaining 2–6 new native contacts in many locations through the region. The FCA398 regions exhibiting the highest number of maintained native contacts relative to wild type largely correspond to the regions of decreased flexibility identified from RMSF. Notably, several of the FCA398 point mutations (P227L, D249K, T255P, and T332Y) were made in these regions. Interestingly, small regions around residues 160 (525 K) and 250 (475 K) exhibit fewer native contacts in FCA398. As described above, the increase in melting temperature due to an added mutation was never simply additive, and this localized destabilization of the FCA398 variant suggests a mechanism for this observation, namely that overall stabilization comes at the cost of localized regions of destabilization.

      Discussion

      In this work, we examine stabilizing mutations of H. jecorina Cel7A and show that the accumulation of such mutations is an effective strategy in the creation of more stable variants. Although the increase in stability, as measured by thermal melting Tm, by adding stable variants together was not as large as a purely arithmetic addition, the approach was effective in allowing us to progress from single variants with an increase in Tm of up to 2.7 °C (relative to wild type) to combinatorial variants with Tm increases of 3.4 °C, of 7.4 °C and, by accumulating 18 mutations (in FCA398), of 10.4 °C above the starting point. This stability increase led to greatly reduced rates of thermal inactivation and increased hydrolysis of cellulose (PASC) at elevated temperatures.
      Most of the 18 point mutations introduced in the FCA398 variant are located on the surface of the protein and distributed over the whole molecule (Fig. 2). The side chains at six of these sites (N49S, N89D, S92T, S196T, D249K, and T332Y) point out into the surrounding solution, and there are no obvious structural explanations for why these mutations might stabilize the molecule.
      All mutations are found in the catalytic module except one, T462I, which is located at the junction between the linker peptide and the CBM (Fig. 2A). Thr462 corresponds to the first residue, Thr1, in the NMR structure of the CBM from H. jecorina Cel7A (PDB code 1CBH (
      • Kraulis J.
      • Clore G.M.
      • Nilges M.
      • Jones T.A.
      • Pettersson G.
      • Knowles J.
      • Gronenborn A.M.
      Determination of the 3-dimensional solution structure of the C-terminal domain of cellobiohydrolase-I from Trichoderma reesei: a study using nuclear magnetic-resonance and hybrid distance geometry dynamical simulated annealing.
      )). Replacement with isoleucine at this site increases hydrophobic interaction with surrounding residues (Pro16/477, Val18/479, and Ala20/481; 1CBH/FCA398 numbering) and could, thereby, strengthen the linker–CBM connection. However, T462I appeared neither as a single point variant nor in FCA367. Accordingly, the results suggest that T462I is an unlikely contributor to an overall increase in thermal stability for the FCA398 full-length protein.
      None of the mutations in the FCA398 catalytic domain are at sites that are completely conserved in the phylogenetically diverse GH7 CBH sequences, examined in the recent paper by Hobdey et al. (
      • Hobdey S.E.
      • Knott B.C.
      • Haddad Momeni M.
      • Taylor 2nd, L.E.
      • Borisova A.S.
      • Podkaminer K.K.
      • VanderWall T.A.
      • Himmel M.E.
      • Decker S.R.
      • Beckham G.T.
      • Ståhlberg J.
      Biochemical and structural characterizations of two dictyostelium cellobiohydrolases from the amoebozoa kingdom reveal a high level of conservation between distant phylogenetic trees of life.
      ). Only one residue, Asn49, is conserved among the known GH7 CBH structures. Five of the specific amino acid substitutions in FCA398 (Leu227, Pro255, Lys295, Pro296, and Tyr332) are not found in the sequences examined by Hobdey et al. (
      • Hobdey S.E.
      • Knott B.C.
      • Haddad Momeni M.
      • Taylor 2nd, L.E.
      • Borisova A.S.
      • Podkaminer K.K.
      • VanderWall T.A.
      • Himmel M.E.
      • Decker S.R.
      • Beckham G.T.
      • Ståhlberg J.
      Biochemical and structural characterizations of two dictyostelium cellobiohydrolases from the amoebozoa kingdom reveal a high level of conservation between distant phylogenetic trees of life.
      ). Of the remainder, eight (Pro8, Thr68, Asp89, Thr92, Asn113, Thr196, Asp403, and Phe411) are found among the GH7 CBHs with known structures, and four more (Ile41, Ser49, Lys249, and Pro278) are found in the larger set of sequences. Only one of the mutated sites in FCA398 is common with mutations reported in H. jecorina Cel7A literature; the T255P site in FCA398 corresponds to T260V in the study by Smith et al. (
      • Smith M.A.
      • Bedbrook C.N.
      • Wu T.
      • Arnold F.H.
      Hypocrea jecorina cellobiohydrolase I stabilizing mutations identified using noncontiguous recombination.
      ).
      The catalytic domain of Cel7A is built around two β-sheets packed against each other to form a β-sandwich that curves around the active site. Long loops extend from the β-sheets and fold into irregular loop regions, a few α-helices, and short β-strands at the surface of the protein; this is where the MD simulations show elevated backbone fluctuations and stabilization of FCA398 relative to wild type (Fig. 3). The largest differences between FCA398 and wild type at high temperature are seen in surface regions clustered around the side of the protein that harbor the product-binding site (to the right in Fig. 2A).

      Mutation P227L

      Near the product site, we find the P227L mutation, which gave the highest ΔTm of the single point mutations analyzed (+2.7 °C; Table 1). It is the only mutation site in FCA398 where the side chain is completely buried within the protein, sitting on a β-strand anti-parallel to the strand that carries the catalytic residues. The flanking residues, Thr226 and His228, point into the product subsite +1, whereas the 227 side chain points in the opposite direction, into the hydrophobic core (Fig. 5). The mutation does not appear to cause significant conformational changes, although the β-strand is slightly shifted at Leu227 in FCA398 compared with Pro227 in the 4C4C structure (0.8 Å between Cα atoms). The larger leucine side chain seems to fill the void better than proline, with more van der Waals contacts and tighter hydrophobic packing with surrounding side chains (Met213, Ile300, Leu326, Leu349, and Phe352; Fig. 5). The P227L mutation probably contributes to the observed stabilization of the 310–358 segment by strengthening the hydrophobic interaction with several hydrophobic residues in that segment. Furthermore, the peptide nitrogen at Leu227 is accessible for a new hydrogen bond (3.0 Å) to the backbone oxygen of Cys261 in the short 261–263 β-strand at the edge of the inner β-sheet. Cys261, in turn, is bound by a disulfide bridge to Cys331 (next to the T332Y mutation) in the 328–338 α-helix at the surface (Fig. 5).
      Figure thumbnail gr5
      Figure 5Comparison of the Cel7A FCA398 (yellow) and 4C4C (white) structures around the P227L mutation site. Selected side chains are shown in stick representations, with mutated residues in FCA398 in magenta.
      The V403D and S411F sites are at either end of a surface helix (at the top of Fig. 2A), which is among the regions that unfold first and where stabilization of FCA398 is indicated by the high-temperature MD simulations. The V403D mutation introduces a new hydrogen bond in FCA398 between the Asp403 side-chain and the backbone amide of Gln406 in the helix. This type of hydrogen bonding, referred to as N-terminal capping, stabilizes the α-helix (
      • Serrano L.
      • Fersht A.R.
      Capping and α-helix stability.
      ,
      • Bell J.A.
      • Becktel W.J.
      • Sauer U.
      • Baase W.A.
      • Matthews B.W.
      Dissection of helix capping in T4 lysozyme by structural and thermodynamic analysis of six amino acid substitutions at Thr 59.
      ), which in turn should stabilize the whole protein (
      • Mitchinson C.
      • Baldwin R.L.
      The design and production of semisynthetic ribonucleases with increased thermostability by incorporation of S-peptide analogues with enhanced helical stability.
      ). At the other end of the helix, the S411F mutation introduces an aromatic residue partially buried underneath the side chain of Gln410. The Phe411 in FCA398 provides a larger hydrophobic interface with underlying amino acids (Pro137, Leu140, and Val407), which should anchor the helix more firmly at the surface of the catalytic domain. Interestingly, an aromatic residue at this position is also present in the documented thermostable cellobiohydrolases M. albomyces Cel7B and Thermoascus aurantiacus Cel7A (
      • Voutilainen S.P.
      • Boer H.
      • Alapuranen M.
      • Jänis J.
      • Vehmaanperä J.
      • Koivula A.
      Improving the thermostability and activity of Melanocarpus albomyces cellobiohydrolase Cel7B.
      ,
      • Benko Z.
      • Drahos E.
      • Szengyel Z.
      • Puranen T.
      • Vehmaanperä J.
      • Réczey K.
      Thermoascus aurantiacus CBHI/Cel7A production in Trichoderma reesei on alternative carbon sources.
      ).

      Tunnel entrance region

      Near the entrance to the cellulose binding tunnel, at subsite −7, the extended B1 loop (residues 41–63) and the following short α-helix (residues 64–70) cover the protein surface. Three of the FCA398 mutations, T41I, N49S, and A68T, are found in this region. Residue Thr41 is pointing into and is partially buried in a hydrophobic pocket in the structure, surrounded by Tyr51, Trp56, Leu73, Lys181, Phe182, and the Cys50–Cys71 disulfide bridge. The replacement of Thr41 with an isoleucine in FCA398 displaces two water molecules and introduces several new hydrophobic contacts with the surrounding hydrophobic amino acids. These new hydrophobic contacts should contribute to the higher stability of the variant protein, as indicated by the 1.7 °C increase of Tm for the single T41I mutation (Table 1). Our MD simulations also demonstrated that, at high temperatures (supplemental Fig. S3), residue 41 in FCA398 retained more native contacts than in the wild type, indicating more favorable interactions with the neighboring amino acid residues.
      The A68T site is located in the 64–70 α-helix, partially buried at the interface between the helix, the N-terminal pyroglutamate residue, one loop turn formed by residues 183–185, and Thr160 in another loop turn at the base of loop B2. The FCA398 Thr68 side chain makes additional van der Waals contacts with surrounding residues and forms two new hydrogen bonds (with the backbone carbonyl oxygen at position 64 within the helix and with the side chain of Gln182, which in turn bonds to Thr160). The A68T mutation is thus likely to anchor the helix more firmly at the surface and may also contribute to stabilization of the neighboring loop regions. The MD simulations revealed that the A68T mutation in FCA398 variant significantly increased hydrogen bond interactions with Gln186 and Ala187 at native temperature, as well as at the two high temperatures, in comparison with the wild type. These newly formed interactions are assumed to provide additional stability to the disulfide bridge between Cys61 and Cys67. Together, these contributions result in an increase of Tm by about 1.2 °C in the single A68T mutation (Table 1).

      Surface loop turns

      13 of the FCA398 mutations are at the surface of the catalytic module, and 10 of these sites are in, or close to, loop turns at the surface (S8P, N49S, N89D, S113N, S196T, D249K, T255P, S278P, E295K, and T296P). All four of the new proline residues introduced into FCA398 are found in such surface turns. The importance of proline residues in thermal stability of enzymes has been described by Matthews et al. (
      • Matthews B.W.
      • Nicholson H.
      • Becktel W.J.
      Enhanced protein thermostability from site-directed mutations that decrease the entropy of unfolding.
      ). As a prime example, Muslin et al. (
      • Muslin E.H.
      • Clark S.E.
      • Henson C.A.
      The effect of proline insertions on the thermostability of a barley α-glucosidase.
      ) showed the effect of proline insertions on the thermostability of a barley α-glucosidase. Introduction of proline residues in FCA398 at these four positions, S8P, T255P, S278P, and T296P, probably stabilizes the Cel7A molecule by restricting main-chain flexibility and, thus, the mobility at these turns. MD simulations suggest that the T255P and S78P mutations have the largest impact on flexibility (Fig. 3). The S8P, T255P, and T296P mutations also provide a larger area of hydrophobic contact with underlying protein atoms, which may contribute further to stabilization of the respective surface turns. The T296P site is in a hairpin turn (residues 295–298), which also carries the E295K mutation. The Lys295 side chain introduced in FCA398 forms a new salt bridge with Glu325, which may add to stabilization of the 295–298 turn and also contribute to the stabilization effect on the 310–358 region seen in the high-temperature MD simulations (Fig. 3 and supplemental Fig. S4).

      New N-glycosylation at S113N site

      One extra NAG residue in the FCA398 structure compared with the wild-type Cel7A structure was found covalently bound to Asn113 (Fig. 6). The S113N mutation, also at a surface hairpin turn, introduces an additional consensus glycosylation motif (Asn-Asp-Thr) in the amino acid sequence, which has been recognized by the N-glycosylation machinery in the endoplasmic reticulum of the expression host A. niger to produce a new glycosylation. This single mutation resulted in an increase in Tm of 1.5 °C (Table 1). Mutating this residue to an aspartic acid did not result in a significant increase in Tm compared with wild type (data not shown), suggesting that the stability effect of the S113N mutation may be the result of an altered glycosylation pattern of the protein. Interestingly, an N-glycosylation motif is present at the corresponding position in the GH7 CBHs from the close relative Trichoderma harzianum, as well as the distantly related social amoeba Dictyostelium purpureum and D. discoideum. An attached NAG residue is indeed visible at this site in the structures of the two former enzymes (
      • Hobdey S.E.
      • Knott B.C.
      • Haddad Momeni M.
      • Taylor 2nd, L.E.
      • Borisova A.S.
      • Podkaminer K.K.
      • VanderWall T.A.
      • Himmel M.E.
      • Decker S.R.
      • Beckham G.T.
      • Ståhlberg J.
      Biochemical and structural characterizations of two dictyostelium cellobiohydrolases from the amoebozoa kingdom reveal a high level of conservation between distant phylogenetic trees of life.
      ,
      • Textor L.C.
      • Colussi F.
      • Silveira R.L.
      • Serpa V.
      • de Mello B.L.
      • Muniz J.R.C.
      • Squina F.M.
      • Pereira Jr., N.
      • Skaf M.S.
      • Polikarpov I.
      Joint X-ray crystallographic and molecular dynamics study of cellobiohydrolase I from Trichoderma harzianum: deciphering the structural features of cellobiohydrolase catalytic activity.
      ).
      Figure thumbnail gr6
      Figure 6The S113N mutation in the Cel7A FCA398 variant introduces a new N-glycosylation site at a surface hairpin turn. An N-acetyl glucosamine residue attached to Asn113 is visible in the electron density map of the FCA398 crystal structure. The 2FoFcFc map is contoured at a σ level of 1.8. The Cel7A 4C4C structure is in light gray, and FCA398 is shown in yellow with mutated sites in magenta.

      Role of cystine residues in stability

      A striking observation is that several mutations are in the vicinity of a cystine residue and interact, either directly or indirectly, with the cystine residues. N49S and A68T are near Cys50, Cys61, Cys67, Cys71, and Cys72, within the 41–74 loop. Four of these cystine residues make short disulfide bridges (Cys50–Cys71 and Cys61–Cys67) within the loop, and the fifth, Cys72, makes a bridge with Cys4 at the N terminus of the enzyme. Another example is at the apex of a loop extending from residue Thr231 to Asp259, where a threonine has been replaced with a proline at position 255. The rigidity of this loop, probably increased by this mutation, in turn, stabilizes the position of Cys256, in a disulfide bond with Cys230. MD simulations support the proposal that the T255P mutation stabilized the protein and the disulfide bond between Cys230 and Cys256 (Fig. 3). The single mutant T255P improves Tm by around 2 °C (Table 1).
      Interactions with cystines could be of importance for the thermal stability of Cel7A; some of the previously determined structures of H. jecorina Cel7A, solved by our group, have shown that the thiol group of some of these cystines can have multiple conformations. The N49S and A68T mutations, for example, may contribute to FCA398 stability by stabilizing low-energy conformations of the disulfides in the 41–74 loop. The H. jecorina Cel7A molecule contains 10 disulfide bridges in the catalytic domain and 2 in the CBM (SwissProt P62694). Disulfide bridges between cystine residues can also be important factors in the thermal stability of enzymes. The cystines involved in disulfide bridges represent about 5% of the residues of the Cel7A catalytic domain, leading one to speculate that disulfides are important for the thermal stability of Cel7A; this idea is supported by examples from the literature. For example, Voutilainen et al. (
      • Voutilainen S.P.
      • Boer H.
      • Alapuranen M.
      • Jänis J.
      • Vehmaanperä J.
      • Koivula A.
      Improving the thermostability and activity of Melanocarpus albomyces cellobiohydrolase Cel7B.
      ) showed that introducing a disulfide bridge found in wild-type H. jecorina Cel7A into the thermostable M. albomyces Cel7B increased the unfolding temperature of this enzyme by 4 °C. Although no disulfides were formed or broken during our experiments or simulations, presumably these covalent bridges limit the possible motions in their vicinity and could indicate regions that would otherwise be sites of local unfolding on the path to thermal denaturation. Thus, mutations near disulfides could further stabilize vulnerable regions and have an indirect effect by stabilizing the disulfide conformation.
      In conclusion, we have developed a robust pathway of mutagenesis of the H. jecorina Cel7A molecule by combining mutated sites and adding their individual influences to improve the thermal stability of Cel7A. The best Cel7A variant obtained, FCA398, inactivated much more slowly than the wild type at elevated temperatures, having a half-life 44 times that of wild type at 62 °C, and had an increase of 10.4 °C in its thermal denaturation melting point. We have shown that this improvement in the thermostability of the Cel7A molecule enables degradation of amorphous cellulose at elevated temperatures, up to 75 °C. The variant FCA398 contained 18 mutated sites compared with the wild-type enzyme, and we developed rationales for the stabilizing effects of most of the mutations on the Cel7A molecule by applying structural and molecular dynamics analyses.

      Experimental procedures

      H. jecorina Cel7A combinatorial variant construction

      Sites with increased thermal stability were identified and selected by screening libraries of H. jecorina Cel7A mutants heterologously expressed in A. niger var. awamori AP4 (
      • Day A.
      • Goedegebuur F.
      • Gualfetti P.
      • Mitchinson C.
      • Neefe P.
      • Sandgren M.
      • Shaw A.
      • Stahlberg J.
      Variant Hypocrea jecorina CBH1 cellulases.
      ). Residual activity of small-scale (100 μl) culture supernatants was assessed by measuring cellobiohydrolase activity against MUL (Sigma, M2405) at 50 °C, as described previously (
      • Momeni M.H.
      • Goedegebuur F.
      • Hansson H.
      • Karkehabadi S.
      • Askarieh G.
      • Mitchinson C.
      • Larenas E.A.
      • Ståhlberg J.
      • Sandgren M.
      Expression, crystal structure and cellulase activity of the thermostable cellobiohydrolase Cel7A from the fungus Humicola grisea var. thermoidea.
      ), before and after heat treatment at elevated temperature (62 °C, for initial rounds, or 69 °C, as more stable variants were introduced as library starting points) for 4 h.
      Specific mutagenized primers were developed, and mutated sites were combined using fusion PCR techniques as described (
      • Turunen O.
      • Etuaho K.
      • Fenel F.
      • Vehmaanperä J.
      • Wu X.
      • Rouvinen J.
      • Leisola M.
      A combination of weakly stabilizing mutations with a disulfide bridge in the α-helix region of Trichoderma reesei endo-1,4-β-xylanase II increases the thermal stability through synergism.
      ). The PCR fragments containing the mutagenized Cel7A gene cbh1 were cloned in E. coli strain MAX Efficiency DH5α (Invitrogen), using plasmid pDONRTM201 (Invitrogen), and then transferred to the E. coli/A. niger shuttle expression vector plasmid pRAXdes, which was subsequently transformed into A. niger var. awamori AP4 (
      • Berka R.M.
      • Barnett C.C.
      The development of gene expression systems for filamentous fungi.
      ), as described previously (
      • Momeni M.H.
      • Goedegebuur F.
      • Hansson H.
      • Karkehabadi S.
      • Askarieh G.
      • Mitchinson C.
      • Larenas E.A.
      • Ståhlberg J.
      • Sandgren M.
      Expression, crystal structure and cellulase activity of the thermostable cellobiohydrolase Cel7A from the fungus Humicola grisea var. thermoidea.
      ,
      • Cao Q.N.
      • Stubbs M.
      • Ngo K.Q.
      • Ward M.
      • Cunningham A.
      • Pai E.F.
      • Tu G.C.
      • Hofmann T.
      Penicillopepsin-JT2, a recombinant enzyme from Penicillium janthinellum and the contribution of a hydrogen bond in subsite S3 to kcat.
      ). General recombinant DNA procedures were adapted from the literature (
      • Sambrook J.
      • Fritsch E.F.
      • Maniatis T.
      ). The mutated genes were sequenced by BaseClear Holding BV (Leiden, The Netherlands), and sequence data were analyzed using the software package VectorNTI®. The nucleotide sequences were translated into protein sequence and compared with the wild-type sequence to identify the mutated sites. The selected Cel7A variants that contributed to the design of the combinatorial variant FCA398 are listed in Table 1. Table 2 lists combinatorial variants. Variant FCA301 is wild-type H. jecorina Cel7A expressed in A. niger var. awamori as a control.

      Recombination of mutated sites of variant FCA367

      All 11 mutated sites of FCA367 were separately introduced into 5′-phosphorylated mutagenized primers. These 11 primers were then used to mutagenize Cel7A to produce a library of mutants with several different combinations, using the QuikChange® multisite-directed mutagenesis kit (Stratagene, La Jolla, CA, catalog no. 200518). A total of around 2000 clones, containing an average of 8 mutations/clone (990 possibilities), were screened for thermal stability, as measured by retention of activity after incubation at high temperature (69 °C).

      Expression and purification of selected Cel7A variants

      The A. niger var. awamori transformants were grown in 500-ml shake flasks for 3 days at 37 °C (
      • Cao Q.N.
      • Stubbs M.
      • Ngo K.Q.
      • Ward M.
      • Cunningham A.
      • Pai E.F.
      • Tu G.C.
      • Hofmann T.
      Penicillopepsin-JT2, a recombinant enzyme from Penicillium janthinellum and the contribution of a hydrogen bond in subsite S3 to kcat.
      ). Expressed Cel7A variants were then purified from culture supernatants by hydrophobic interaction chromatography on phenyl-Sepharose resin (GE Healthcare catalog no. 17-0973-05) equilibrated with 5 column volumes (CV) of 0.020 m sodium phosphate, 0.5 m ammonium sulfate at pH 6.8. Ammonium sulfate was added to the supernatants to a final concentration of 0.5 m, and the pH was adjusted to 6.8. After filtration, the supernatant was loaded onto the column. The column was washed with 10 CV of equilibration buffer and then eluted with a 10-CV gradient from 0.5 to 0 m ammonium sulfate in 0.020 m sodium phosphate, pH 6.8. Cel7A eluted approximately mid-gradient. Fractions were collected and pooled on the basis of SDS-PAGE analysis.

      Determination of Tm

      Protein Tm values were determined for the purified enzymes by CD spectroscopy, as described previously (
      • Momeni M.H.
      • Goedegebuur F.
      • Hansson H.
      • Karkehabadi S.
      • Askarieh G.
      • Mitchinson C.
      • Larenas E.A.
      • Ståhlberg J.
      • Sandgren M.
      Expression, crystal structure and cellulase activity of the thermostable cellobiohydrolase Cel7A from the fungus Humicola grisea var. thermoidea.
      ). The thermal unfolding of the Cel7A variants was not reversible; in our measurements of Cel7A melting points, the native, starting, CD spectrum (and, specifically, the 230 nm value) was not recovered upon cooling.

      Thermal inactivation half-life measurements

      For determination of irreversible inactivation t½, the purified variants, FCA301, FCA353, FCA367, and FCA398, were incubated at 62, 66, and 69 °C in a microtiter plate incubator (iEMSTM microplate incubator/shaker HT, Thermo Fisher Scientific). Samples were taken at regular time intervals up to 20 h, cooled on ice for 10 min, and assayed for residual activity on MUL (measured at 50 °C, as described (
      • Momeni M.H.
      • Goedegebuur F.
      • Hansson H.
      • Karkehabadi S.
      • Askarieh G.
      • Mitchinson C.
      • Larenas E.A.
      • Ståhlberg J.
      • Sandgren M.
      Expression, crystal structure and cellulase activity of the thermostable cellobiohydrolase Cel7A from the fungus Humicola grisea var. thermoidea.
      )). The half-life was derived from the slope (k) of the natural logarithm of residual activity versus time using the formula t½ = ln 2/k.

      Cellulose conversion assay using PASC as substrate

      PASC (1% (w/v)) (
      • Walseth C.S.
      The influence of the fine structure of cellulose on the action of cellulases.
      ,
      • Wood T.M.
      The cellulase of Fusarium solani: purification and specificity of the β-(1→4)-glucanase and the β-d-glucosidase components.
      ) prepared from Sigmacell type 20 (Sigma catalog no. 9004-34-6) was incubated with FCA301 (wild type) or variant FCA398 (0.5 mg enzyme/g cellulose), in 200 μl of 50 mm sodium acetate buffer, pH 5.5, in sealed microtiter plates at 53, 65, and 75 °C in an Eppendorf Thermomixer shaking at 700 rpm. Samples were taken at regular intervals, and soluble sugars were quantified by HPLC as described (
      • Rignall T.R.
      • Baker J.O.
      • McCarter S.L.
      • Adney W.S.
      • Vinzant T.B.
      • Decker S.R.
      • Himmel M.E.
      Effect of single active-site cleft mutation on product specificity in a thermostable bacterial cellulase.
      ).

      Preparation and crystallization of Cel7A variant FCA398 catalytic domain

      Before crystallization, the H. jecorina Cel7A FCA398 variant enzyme (expressed in A. niger var. awamori AP4 and purified as described above) was deglycosylated and treated with papain to remove the C-terminal peptide linker and carbohydrate-binding module from the full-length protein, using a procedure similar to that described previously (
      • Ståhlberg J.
      • Divne C.
      • Koivula A.
      • Piens K.
      • Claeyssens M.
      • Teeri T.T.
      • Jones T.A.
      Activity studies and crystal structures of catalytically deficient mutants of cellobiohydrolase I from Trichoderma reesei.
      ). The FCA398 protein, in 100 mm sodium acetate, pH 5.0, and 20 mm zinc acetate, was mixed with α-mannosidase (Sigma-Aldrich) and endo-N-acetyl glucose aminidase H to a final concentration of 40:2:1 (Cel7A/α-mannosidase/endo-N-acetyl glucose aminidase H, w/w/w) and incubated at room temperature for 24 h. Activated papain (Sigma-Aldrich) was then added to a final concentration of 50:1 (Cel7A/papain, w/w). After overnight incubation at room temperature, the buffer was changed to 20 mm sodium acetate, pH 5.0, and the protein was loaded onto a Source 30Q anion exchange column (GE Healthcare) and eluted by a linear 0.1–1.0 M sodium chloride gradient. Fractions containing FCA398 catalytic domain were collected, and after buffer exchange to 25 mm sodium morpholine ethane sulfonic acid, pH 6.0, the protein was concentrated to 12 mg/ml. The purity of the protein was >95%, as determined by SDS-PAGE. Note that one of the mutations in FCA398, T462I, is in the linker and, hence, is removed by this process.
      Initial screening for crystallization conditions for the FCA398 variant was performed with the JCSG+ screen (Qiagen). Well-diffracting crystals were obtained in drops with 30% JCSG+ solution D9 (25.5% PEG 4000, 0.17 m ammonium sulfate, and 15% glycerol) and 70% protein solution (∼12 mg/ml), equilibrated against D9 solution at 20 °C (
      • McPherson A.
      Preparation and Analysis of Protein Crystals.
      ). Large, single, square-shaped crystals grew to a size of 10 × 200 × 200 μm in sitting drops within 1 week. The crystals belong to the space group P21212 with the cell dimensions a = 102.9 Å, b = 92.1 Å, c = 102.2 Å and have a calculated Vm of 2.48 (
      • Matthews B.W.
      Solvent content of protein crystals.
      ,
      • Kantardjieff K.A.
      • Rupp B.
      Matthews coefficient probabilities: improved estimates for unit cell contents of proteins, DNA, and protein–nucleic acid complex crystals.
      ), a solvent content of 50.5%, and two Cel7A molecules in the asymmetric unit.

      X-ray data collection, structure solution, and refinement

      Single crystals were picked with cryoloops and flash-frozen in liquid nitrogen. X-ray diffraction data were collected at 100 K on beam line ID23-2 at the European Synchrotron Radiation Facility (Grenoble, France). Data collection and processing statistics for the final structure are given in Table 3. The data were integrated with XDS and scaled with SCALA in the CCP4 package (
      • Evans P.
      Scaling and assessment of data quality.
      ,
      • Kabsch W.
      XDS.
      • Winn M.D.
      • Ballard C.C.
      • Cowtan K.D.
      • Dodson E.J.
      • Emsley P.
      • Evans P.R.
      • Keegan R.M.
      • Krissinel E.B.
      • Leslie A.G.W.
      • McCoy A.
      • Leslie A.G.
      • McCoy A.
      • McNicholas S.J.
      • Murshudov G.N.
      • Pannu N.S.
      • et al.
      Overview of the CCP4 suite and current developments.
      ). 5% of the reflections were set aside for calculation of Rfree (
      • Brünger A.T.
      Free R value: a novel statistical quantity for assessing the accuracy of crystal structures.
      ). The structure of the H. jecorina Cel7A FCA398 catalytic domain was solved by molecular replacement using PHASER with a wild-type H. jecorina Cel7A structure as the search model (PDB code 2V3I (
      • Divne C.
      • Ståhlberg J.
      • Reinikainen T.
      • Ruohonen L.
      • Pettersson G.
      • Knowles J.K.
      • Teeri T.T.
      • Jones T.A.
      The three-dimensional crystal structure of the catalytic core of cellobiohydrolase I from Trichoderma reesei.
      )). The structure was refined with alternating cycles of model building using COOT and maximum likelihood refinement in Refmac version 5.0 (
      • Murshudov G.N.
      • Vagin A.A.
      • Dodson E.J.
      Refinement of macromolecular structures by the maximum-likelihood method.
      ). Most water molecules in the structure model were located automatically using water picking protocols in the refinement program. These water molecules were then manually selected for inclusion or discarded by visual inspection. A summary of the data collection and refinement statistics is given in Table 3. All structural comparisons were made with COOT, and figures were prepared with PyMOL (
      • Harris M.
      • Jones T.A.
      Molray: a web interface between O and the POV-Ray ray tracer.
      ). Coordinates and structure-factor amplitudes have been deposited with the Protein Data Bank (
      • Bernstein F.C.
      • Koetzle T.F.
      • Williams G.J.B.
      • Meyer Jr., E.F.
      • Brice M.D.
      • Rodgers J.R.
      • Kennard O.
      • Shimanouchi T.
      • Tasumi M.
      The Protein Data Bank: a computer-based archival file for macromolecular structures.
      ) under PDB accession code 5OA5.

      Molecular dynamics simulation

      To evaluate the role of mutations in enhancing thermal stability, classical MD simulations of wild-type Cel7A and the FCA398 variant were performed at three temperatures: 300, 475, and 525 K. The simulation at 300 K was used to identify baseline dynamics of the two enzymes under standard conditions, providing a point of comparison for behavior at elevated temperatures. High-temperature simulations, at 475 and 525 K, were performed to observe the structural changes that occur as part of the thermal unfolding mechanisms. The selected temperatures are necessarily higher than those the proteins would be exposed to in vivo, so that unfolding was observable within a feasible timeframe for molecular simulation. Although the temperatures are quite high, we do not anticipate adverse impact on predicted thermal unfolding mechanisms; prior studies, conducted at simulation temperatures up to 600 K, have established that unfolding mechanisms are relatively independent of simulation temperature (
      • Day R.
      • Bennion B.J.
      • Ham S.
      • Daggett V.
      Increasing temperature accelerates protein unfolding without changing the pathway of unfolding.
      ,
      • Huang X.
      • Zhou H.X.
      Similarity and difference in the unfolding of thermophilic and mesophilic cold shock proteins studied by molecular dynamics.
      • Sham Y.Y.
      • Ma B.
      • Tsai C.J.
      • Nussinov R.
      Thermal unfolding molecular dynamics simulation of Escherichia coli dihydrofolate reductase: thermal stability of protein domains and unfolding pathway.
      ). Comparisons between an averaged set of trajectories, thus defining the unfolding mechanism, were made on the basis of native contacts, where a native contact is any residue within 6.5 Å of a residue of interest (
      • Sheinerman F.B.
      • Brooks 3rd, C.L.
      Calculations on folding of segment B1 of streptococcal protein G.
      ).
      Wild-type Cel7A and the FCA398 variant were constructed from deposited PDB structures, 4C4C (
      • Knott B.C.
      • Haddad Momeni M.
      • Crowley M.F.
      • Mackenzie L.F.
      • Götz A.W.
      • Sandgren M.
      • Withers S.G.
      • Ståhlberg J.
      • Beckham G.T.
      The mechanism of cellulose hydrolysis by a two-step, retaining cellobiohydrolase elucidated by structural and transition path sampling studies.
      ) and 5OA5, respectively. The E217Q mutation in the 4C4C structure was reversed from Gln to Glu to obtain native Cel7A, and the cellononaose oligomer was removed from the catalytic tunnel. The FCA398 model required no manipulation of the active site. In this study, we modeled only the catalytic domains of the enzymes, maintaining consistency with the crystal structures. Accordingly, the FCA398 model contained only 17 of the 18 mutations (Table 2), as the T462I appears in the linker domain of the full-length protein. The catalytic domains were modeled in a deglycosylated state; based on previous observations, glycosylation effects on protein dynamics are marginal over the MD simulation timescales (
      • Taylor C.B.
      • Payne C.M.
      • Himmel M.E.
      • Crowley M.F.
      • McCabe C.
      • Beckham G.T.
      Binding site dynamics and aromatic-carbohydrate interactions in processive and non-processive family 7 glycoside hydrolases.
      ,
      • Payne C.M.
      • Bomble Y.J.
      • Taylor C.B.
      • McCabe C.
      • Himmel M.E.
      • Crowley M.F.
      • Beckham G.T.
      Multiple functions of aromatic-carbohydrate interactions in a processive cellulase examined with molecular simulation.
      ). Protonation states were determined using H++ at pH 5 with an internal and external dielectric constant of 10 and 80, respectively (
      • Anandakrishnan R.
      • Aguilar B.
      • Onufriev A.V.
      H++ 3.0: automating pK prediction and the preparation of biomolecular structures for atomistic molecular modeling and simulations.
      ,
      • Gordon J.C.
      • Myers J.B.
      • Folta T.
      • Shoja V.
      • Heath L.S.
      • Onufriev A.
      H++: a server for estimating pKas and adding missing hydrogens to macromolecules.
      • Myers J.
      • Grothaus G.
      • Narayanan S.
      • Onufriev A.
      A simple clustering algorithm can be accurate enough for use in calculations of pKs in macromolecules.
      ). Selected protonation states and disulfide bonds are provided in supplemental Table S1. The enzymes were then explicitly solvated in water in an 80 × 80 × 80 Å cubic box, and sodium ions were added to ensure charge neutrality of the systems. The resultant systems contained roughly 52,000 atoms.
      After system construction, the models were minimized, heated, and equilibrated before data collection. First, the water molecules and ions were minimized for 1000 steps of steepest descent in CHARMM, keeping the protein fixed (
      • Brooks B.R.
      • Brooks 3rd, C.L.
      • Mackerell Jr., A.D.
      • Nilsson L.
      • Petrella R.J.
      • Roux B.
      • Won Y.
      • Archontis G.
      • Bartels C.
      • Boresch S.
      • Caflisch A.
      • Caves L.
      • Cui Q.
      • Dinner A.R.
      • Feig M.
      • et al.
      CHARMM: the biomolecular simulation program.
      ). This was followed by 1000 steps of steepest descent in which the entire system was allowed to move freely. After minimization, the systems were heated from 100 to 300 K in the NVE ensemble over 20 ps in 50 K increments. At this point, three independent 300 K simulations were started from different random number seeds for both wild-type Cel7A and FCA398 (supplemental Fig. S4). These six simulations were then density-equilibrated for 100 ps in the NPT ensemble in CHARMM using a Nosè–Hoover thermostat and barostat (
      • Hoover W.G.
      Canonical dynamics: equilibrium phase-space distributions.
      ,
      • Nose S.
      • Klein M.L.
      Constant pressure molecular-dynamics for molecular-systems.
      ). The CHARMM36 all-atom force field with CMAP corrections was used to model the proteins (
      • Brooks B.R.
      • Brooks 3rd, C.L.
      • Mackerell Jr., A.D.
      • Nilsson L.
      • Petrella R.J.
      • Roux B.
      • Won Y.
      • Archontis G.
      • Bartels C.
      • Boresch S.
      • Caflisch A.
      • Caves L.
      • Cui Q.
      • Dinner A.R.
      • Feig M.
      • et al.
      CHARMM: the biomolecular simulation program.
      ,
      • Mackerell Jr., A.D.
      • Feig M.
      • Brooks 3rd, C.L.
      Extending the treatment of backbone energetics in protein force fields: Limitations of gas-phase quantum mechanics in reproducing protein conformational distributions in molecular dynamics simulations.
      ,
      • MacKerell A.D.
      • Bashford D.
      • Bellott M.
      • Dunbrack R.L.
      • Evanseck J.D.
      • Field M.J.
      • Fischer S.
      • Gao J.
      • Guo H.
      • Ha S.
      • Joseph-McCarthy D.
      • Kuchnir L.
      • Kuczera K.
      • Lau F.T.
      • Mattos C.
      • et al.
      All-atom empirical potential for molecular modeling and dynamics studies of proteins.
      ), and the modified TIP3P force field was used to model water (
      • Jorgensen W.L.
      • Chandrasekhar J.
      • Madura J.D.
      • Impey R.W.
      • Klein M.L.
      Comparison of simple potential functions for simulating liquid water.
      ,
      • Durell S.R.
      • Brooks B.R.
      • Bennaim A.
      Solvent-induced forces between 2 hydrophilic groups.
      ). Long-range electrostatic interactions were modeled using the particle mesh Ewald approach. The nonbonded cut-off distance was 10 Å, with a switching distance of 9 Å and a nonbonded pair list distance of 12 Å.
      The equilibrated systems were simulated for 50 ns in NAMD at 300 K in the NVT ensemble using a 2-fs time step (
      • Phillips J.C.
      • Braun R.
      • Wang W.
      • Gumbart J.
      • Tajkhorshid E.
      • Villa E.
      • Chipot C.
      • Skeel R.D.
      • Kalé L.
      • Schulten K.
      Scalable molecular dynamics with NAMD.
      ). High-temperature simulations were started from a 10-ns snapshot of the 300 K simulations and run for 15 ns in the NVT ensemble (supplemental Fig. S5). In all, 18 simulations were performed: three replicates at 300, 475, and 525 K for both wild-type Cel7A and FCA398 (supplemental Fig. S4). VMD was used to visualize the trajectories (
      • Humphrey W.
      • Dalke A.
      • Schulten K.
      VMD: visual molecular dynamics.
      ).

      Author contributions

      C. M. conceived and coordinated the study. F. G., J. S., C. M. P., C. M., and M. S. compiled all results and wrote the paper. F. G., L. D., P. K., and P. J. M. T. conducted the Aspergillus expression, screening of thermostabilized variants, and larger-scale expression and provided protein for further analysis. V. H. and P. G. purified proteins and performed analyses to generate Tm data. B. R. K. and E. A. L. developed and performed the biochemical analyses. S. K., H. H., and M. S. prepared the FCA398 protein for crystallization, crystallized it, and performed the X-ray crystallography part of the study. S. J. and C. M. P. performed, analyzed, and interpreted the MD simulations and prepared the related figures (Figure 3, Figure 4 and supplemental Figs. S3–S5). J. S., C. M. P., and C. M. combined the structural and MD results and developed rationales for stabilizing effect at mutated sites, and J. S. prepared the structural figures (Figs. 2, 5, and 6). All authors reviewed the results and approved the final version of the manuscript.

      Acknowledgments

      Computing resources were provided by the University of Kentucky and National Science Foundation (NSF) Extreme Science and Engineering Discovery Environment (XSEDE), which is supported by NSF Grant ACI-1053575 (under allocation MCB090159). We thank Dr. Nils Mikkelsen (Swedish University of Agricultural Sciences) for help with X-ray data collection.

      Supplementary Material

      References

        • Chundawat S.P.S.
        • Beckham G.T.
        • Himmel M.E.
        • Dale B.E.
        Deconstruction of lignocellulosic biomass to fuels and chemicals.
        Annu. Rev. Chem. Biomol. Eng. 2011; 2: 121-145
        • Payne C.M.
        • Knott B.C.
        • Mayes H.B.
        • Hansson H.
        • Himmel M.E.
        • Sandgren M.
        • Ståhlberg J.
        • Beckham G.T.
        Fungal cellulases.
        Chem. Rev. 2015; 115: 1308-1448
        • Kubicek C.P.
        Systems biological approaches towards understanding cellulase production by Trichoderma reesei.
        J. Biotechnol. 2013; 163: 133-142
        • Okada H.
        • Tada K.
        • Sekiya T.
        • Yokoyama K.
        • Takahashi A.
        • Tohda H.
        • Kumagai H.
        • Morikawa Y.
        Molecular characterization and heterologous expression of the gene encoding a low-molecular-mass endoglucanase from Trichoderma reesei QM9414.
        Appl. Environ. Microbiol. 1998; 64: 555-563
        • Saloheimo M.
        • Lehtovaara P.
        • Penttilä M.
        • Teeri T.T.
        • Ståhlberg J.
        • Johansson G.
        • Pettersson G.
        • Claeyssens M.
        • Tomme P.
        • Knowles J.K.
        EGIII, a new endoglucanase from Trichoderma reesei: the characterization of both gene and enzyme.
        Gene. 1988; 63: 11-22
        • Saloheimo A.
        • Henrissat B.
        • Hoffrén A.M.
        • Teleman O.
        • Penttilä M.
        A novel, small endoglucanase gene, egl5, from Trichoderma reesei isolated by expression in yeast.
        Mol. Microbiol. 1994; 13: 219-228
        • Saloheimo M.
        • Nakari-Setälä T.
        • Tenkanen M.
        • Penttilä M.
        cDNA cloning of a Trichoderma reesei cellulase and demonstration of endoglucanase activity by expression in yeast.
        Eur. J. Biochem. 1997; 249: 584-591
        • Teeri T.T.
        • Lehtovaara P.
        • Kauppinen S.
        • Salovuori I.
        • Knowles J.
        Homologous domains in Trichoderma reesei cellulolytic enzymes: gene sequence and expression of cellobiohydrolase II.
        Gene. 1987; 51: 43-52
        • Ward M.
        • Wu S.
        • Dauberman J.
        • Weiss G.
        • Larenas E.
        • Bower B.
        • Rey M.
        • Clarkson K.
        • Bott R.
        Cloning, sequence and preliminary structural analysis of a small, high pI endoglucanase (EGIII) from Trichoderma reesei.
        in: Suominen P. Reinikainen T. The Tricel 93 symposium. Vol. 8. Foundation for Biotechnical and Industrial Fermentation Research, Espoo, Finland1993: 153-158
        • Foreman P.K.
        • Brown D.
        • Dankmeyer L.
        • Dean R.
        • Diener S.
        • Dunn-Coleman N.S.
        • Goedegebuur F.
        • Houfek T.D.
        • England G.J.
        • Kelley A.S.
        • Meerman H.J.
        • Mitchell T.
        • Mitchinson C.
        • Olivares H.A.
        • Teunissen P.J.
        • et al.
        Transcriptional regulation of biomass-degrading enzymes in the filamentous fungus Trichoderma reesei.
        J. Biol. Chem. 2003; 278: 31988-31997
        • Barnett C.C.
        • Berka R.M.
        • Fowler T.
        Cloning and amplification of the gene encoding an extracellular β-glucosidase from Trichoderma reesei: evidence for improved rates of saccharification of cellulosic substrates.
        Biotechnology. 1991; 9: 562-567
        • Takashima S.
        • Iikura H.
        • Nakamura A.
        • Hidaka M.
        • Masaki H.
        • Uozumi T.
        Comparison of gene structures and enzymatic properties between two endoglucanases from Humicola grisea.
        J. Biotechnol. 1999; 67: 85-97
        • Li X.L.
        • Spániková S.
        • de Vries R.P.
        • Biely P.
        Identification of genes encoding microbial glucuronoyl esterases.
        FEBS Lett. 2007; 581: 4029-4035
        • Jacobson F.
        • Karkehabadi S.
        • Hansson H.
        • Goedegebuur F.
        • Wallace L.
        • Mitchinson C.
        • Piens K.
        • Stals I.
        • Sandgren M.
        The crystal structure of the core domain of a cellulose induced protein (Cip1) from Hypocrea jecorina, at 1.5 Å resolution.
        PLoS One. 2013; 8: e70562
        • Cherry J.R.
        • Fidantsef A.L.
        Directed evolution of industrial enzymes: an update.
        Curr. Opin. Biotechnol. 2003; 14: 438-443
        • Suominen P.L.
        • Mäntylä A.L.
        • Karhunen T.
        • Hakola S.
        • Nevalainen H.
        High-frequency one-step gene replacement in Trichoderma reesei. 2. Effects of deletions of individual cellulase genes.
        Mol. Gen. Genet. 1993; 241: 523-530
        • Ilmén M.
        • Saloheimo A.
        • Onnela M.L.
        • Penttilä M.E.
        Regulation of cellulase gene expression in the filamentous fungus Trichoderma reesei.
        Appl. Environ. Microbiol. 1997; 63: 1298-1306
        • Chundawat S.P.S.
        • Lipton M.S.
        • Purvine S.O.
        • Uppugundla N.
        • Gao D.
        • Balan V.
        • Dale B.E.
        Proteomics-based compositional analysis of complex cellulase-hemicellulase mixtures.
        J. Proteome Res. 2011; 10: 4365-4372
        • Himmel M.E.
        • Ding S.Y.
        • Johnson D.K.
        • Adney W.S.
        • Nimlos M.R.
        • Brady J.W.
        • Foust T.D.
        Biomass recalcitrance: engineering plants and enzymes for biofuels production.
        Science. 2007; 315: 804-807
        • Momeni M.H.
        • Goedegebuur F.
        • Hansson H.
        • Karkehabadi S.
        • Askarieh G.
        • Mitchinson C.
        • Larenas E.A.
        • Ståhlberg J.
        • Sandgren M.
        Expression, crystal structure and cellulase activity of the thermostable cellobiohydrolase Cel7A from the fungus Humicola grisea var. thermoidea.
        Acta Crystallogr. D Biol. Crystallogr. 2014; 70: 2356-2366
        • Lehmann M.
        • Loch C.
        • Middendorf A.
        • Studer D.
        • Lassen S.F.
        • Pasamontes L.
        • van Loon A.P.
        • Wyss M.
        The consensus concept for thermostability engineering of proteins: further proof of concept.
        Protein Eng. 2002; 15: 403-411
        • van den Burg B.
        • Dijkstra B.W.
        • van der Vinne B.
        • Stulp B.K.
        • Eijsink V.G.
        • Venema G.
        Introduction of disulfide bonds into Bacillus subtilis neutral protease.
        Protein Eng. 1993; 6: 521-527
        • Sandgren M.
        • Gualfetti P.J.
        • Shaw A.
        • Gross L.S.
        • Saldajeno M.
        • Day A.G.
        • Jones T.A.
        • Mitchinson C.
        Comparison of family 12 glycoside hydrolases and recruited substitutions important for thermal stability.
        Protein Sci. 2003; 12: 848-860
        • Goedegebuur F.
        • Fowler T.
        • Phillips J.
        • van der Kley P.
        • van Solingen P.
        • Dankmeyer L.
        • Power S.D.
        Cloning and relational analysis of 15 novel fungal endoglucanases from family 12 glycosyl hydrolase.
        Curr. Genet. 2002; 41: 89-98
        • Sammond D.W.
        • Kastelowitz N.
        • Himmel M.E.
        • Yin H.
        • Crowley M.F.
        • Bomble Y.J.
        Comparing residue clusters from thermophilic and mesophilic enzymes reveals adaptive mechanisms.
        PLoS One. 2016; 11: e0145848
        • Lehmann M.
        • Pasamontes L.
        • Lassen S.F.
        • Wyss M.
        The consensus concept for thermostability engineering of proteins.
        Biochim. Biophys. Acta. 2000; 1543: 408-415
        • Kim P.
        • Yoon S.H.
        • Seo M.J.
        • Oh D.K.
        • Choi J.H.
        Improvement of tagatose conversion rate by genetic evolution of thermostable galactose isomerase.
        Biotechnol. Appl. Biochem. 2001; 34: 99-102
        • Sakaue R.
        • Kajiyama N.
        Thermostabilization of bacterial fructosyl-amino acid oxidase by directed evolution.
        Appl. Environ. Microbiol. 2003; 69: 139-145
        • Meyer M.M.
        • Hochrein L.
        • Arnold F.H.
        Structure-guided SCHEMA recombination of distantly related β-lactamases.
        Protein Eng. Des. Sel. 2006; 19: 563-570
        • Heinzelman P.
        • Snow C.D.
        • Wu I.
        • Nguyen C.
        • Villalobos A.
        • Govindarajan S.
        • Minshull J.
        • Arnold F.H.
        A family of thermostable fungal cellulases created by structure-guided recombination.
        Proc. Natl. Acad. Sci. U.S.A. 2009; 106: 5610-5615
        • Godbole S.
        • Decker S.R.
        • Nieves R.A.
        • Adney W.S.
        • Vinzant T.B.
        • Baker J.O.
        • Thomas S.R.
        • Himmel M.E.
        Cloning and expression of Trichoderma reesei cellobiohydrolase I in Pichia pastoris.
        Biotechnol. Prog. 1999; 15: 828-833
        • Laymon R.A.
        • Adney W.S.
        • Mohagheghi A.
        • Himmel M.E.
        • Thomas S.R.
        Cloning and expression of full-length Trichoderma reesei cellobiohydrolase I cDNAs in Escherichia coli.
        Appl. Biochem. Biotechnol. 1996; 57: 389-397
        • Reinikainen T.
        • Ruohonen L.
        • Nevanen T.
        • Laaksonen L.
        • Kraulis P.
        • Jones T.A.
        • Knowles J.K.
        • Teeri T.T.
        Investigation of the function of mutated cellulose-binding domains of Trichoderma reesei cellobiohydrolase I.
        Proteins. 1992; 14: 475-482
        • Penttilä M.E.
        • André L.
        • Lehtovaara P.
        • Bailey M.
        • Teeri T.T.
        • Knowles J.K.
        Efficient secretion of two fungal cellobiohydrolases by Saccharomyces cerevisiae.
        Gene. 1988; 63: 103-112
        • Dana C.M.
        • Dotson-Fagerstrom A.
        • Roche C.M.
        • Kal S.M.
        • Chokhawala H.A.
        • Blanch H.W.
        • Clark D.S.
        The importance of pyroglutamate in cellulase Cel7A.
        Biotechnol. Bioeng. 2014; 111: 842-847
        • Heinzelman P.
        • Komor R.
        • Kanaan A.
        • Romero P.
        • Yu X.
        • Mohler S.
        • Snow C.
        • Arnold F.
        Efficient screening of fungal cellobiohydrolase class I enzymes for thermostabilizing sequence blocks by SCHEMA structure-guided recombination.
        Protein Eng. Des. Sel. 2010; 23: 871-880
        • Komor R.S.
        • Romero P.A.
        • Xie C.B.
        • Arnold F.H.
        Highly thermostable fungal cellobiohydrolase I (Cel7A) engineered using predictive methods.
        Protein Eng. Des. Sel. 2012; 25: 827-833
        • Dana C.M.
        • Saija P.
        • Kal S.M.
        • Bryan M.B.
        • Blanch H.W.
        • Clark D.S.
        Biased clique shuffling reveals stabilizing mutations in cellulase Cel7A.
        Biotechnol. Bioeng. 2012; 109: 2710-2719
        • Voutilainen S.P.
        • Boer H.
        • Alapuranen M.
        • Jänis J.
        • Vehmaanperä J.
        • Koivula A.
        Improving the thermostability and activity of Melanocarpus albomyces cellobiohydrolase Cel7B.
        Appl. Microbiol. Biotechnol. 2009; 83: 261-272
        • Voutilainen S.P.
        • Murray P.G.
        • Tuohy M.G.
        • Koivula A.
        Expression of Talaromyces emersonii cellobiohydrolase Cel7A in Saccharomyces cerevisiae and rational mutagenesis to improve its thermostability and activity.
        Protein Eng. Des. Sel. 2010; 23: 69-79
        • Smith M.A.
        • Bedbrook C.N.
        • Wu T.
        • Arnold F.H.
        Hypocrea jecorina cellobiohydrolase I stabilizing mutations identified using noncontiguous recombination.
        ACS Synth. Biol. 2013; 2: 690-696
        • Day A.
        • Goedegebuur F.
        • Gualfetti P.
        • Mitchinson C.
        • Neefe P.
        • Sandgren M.
        • Shaw A.
        • Stahlberg J.
        Variant Hypocrea jecorina CBH1 cellulases.
        January 28 2014 (United States Patent 8,637,294)
        • Kanaya S.
        Stabilization of an enzyme with protein engineering technology.
        in: Biotechnology for Sustainable Utilization of Biological Resources in the Tropics. Vol. 14. International Center for Biotechnology, Osaka University, Osaka, Japan2000: 186-192
        • Gray K.A.
        • Richardson T.H.
        • Kretz K.
        • Short J.M.
        • Bartnek F.
        • Knowles R.
        • Kan L.
        • Swanson P.E.
        • Robertson D.E.
        Rapid evolution of reversible denaturation and elevated melting temperature in a microbial haloalkane.
        Adv. Synth. Catal. 2001; 343: 607-617
        • Stals I.
        • Sandra K.
        • Devreese B.
        • Van Beeumen J.
        • Claeyssens M.
        Factors influencing glycosylation of Trichoderma reesei cellulases. II: N-glycosylation of Cel7A core protein isolated from different strains.
        Glycobiology. 2004; 14: 725-737
        • Knott B.C.
        • Crowley M.F.
        • Himmel M.E.
        • Ståhlberg J.
        • Beckham G.T.
        Carbohydrate–protein interactions that drive processive polysaccharide translocation in enzymes revealed from a computational study of cellobiohydrolase processivity.
        J. Am. Chem. Soc. 2014; 136: 8810-8819
        • von Ossowski I.
        • Ståhlberg J.
        • Koivula A.
        • Piens K.
        • Becker D.
        • Boer H.
        • Harle R.
        • Harris M.
        • Divne C.
        • Mahdi S.
        • Zhao Y.
        • Driguez H.
        • Claeyssens M.
        • Sinnott M.L.
        • Teeri T.T.
        Engineering the exo-loop of Trichoderma reesei cellobiohydrolase, Cel7A: a comparison with Phanerochaete chrysosporium Cel7D.
        J. Mol. Biol. 2003; 333: 817-829
        • Momeni M.H.
        • Payne C.M.
        • Hansson H.
        • Mikkelsen N.E.
        • Svedberg J.
        • Engström Å.
        • Sandgren M.
        • Beckham G.T.
        • Ståhlberg J.
        Structural, biochemical, and computational characterization of the glycoside hydrolase family 7 cellobiohydrolase of the tree-killing fungus Heterobasidion irregulare.
        J. Biol. Chem. 2013; 288: 5861-5872
        • Divne C.
        • Ståhlberg J.
        • Teeri T.T.
        • Jones T.A.
        High-resolution crystal structures reveal how a cellulose chain is bound in the 50 Å long tunnel of cellobiohydrolase I from Trichoderma reesei.
        J. Mol. Biol. 1998; 275: 309-325
        • Muńoz I.G.
        • Mowbray S.L.
        • Stahlberg J.
        The catalytic module of Cel7D from Phanerochaete chrysosporium as a chiral selector: structural studies of its complex with the β blocker (R)-propranolol.
        Acta Crystallogr. D Biol. Crystallogr. 2003; 59: 637-643
        • Kraulis J.
        • Clore G.M.
        • Nilges M.
        • Jones T.A.
        • Pettersson G.
        • Knowles J.
        • Gronenborn A.M.
        Determination of the 3-dimensional solution structure of the C-terminal domain of cellobiohydrolase-I from Trichoderma reesei: a study using nuclear magnetic-resonance and hybrid distance geometry dynamical simulated annealing.
        Biochemistry. 1989; 28: 7241-7257
        • Hobdey S.E.
        • Knott B.C.
        • Haddad Momeni M.
        • Taylor 2nd, L.E.
        • Borisova A.S.
        • Podkaminer K.K.
        • VanderWall T.A.
        • Himmel M.E.
        • Decker S.R.
        • Beckham G.T.
        • Ståhlberg J.
        Biochemical and structural characterizations of two dictyostelium cellobiohydrolases from the amoebozoa kingdom reveal a high level of conservation between distant phylogenetic trees of life.
        Appl. Environ. Microbiol. 2016; 82: 3395-3409
        • Serrano L.
        • Fersht A.R.
        Capping and α-helix stability.
        Nature. 1989; 342: 296-299
        • Bell J.A.
        • Becktel W.J.
        • Sauer U.
        • Baase W.A.
        • Matthews B.W.
        Dissection of helix capping in T4 lysozyme by structural and thermodynamic analysis of six amino acid substitutions at Thr 59.
        Biochemistry. 1992; 31: 3590-3596
        • Mitchinson C.
        • Baldwin R.L.
        The design and production of semisynthetic ribonucleases with increased thermostability by incorporation of S-peptide analogues with enhanced helical stability.
        Proteins. 1986; 1: 23-33
        • Benko Z.
        • Drahos E.
        • Szengyel Z.
        • Puranen T.
        • Vehmaanperä J.
        • Réczey K.
        Thermoascus aurantiacus CBHI/Cel7A production in Trichoderma reesei on alternative carbon sources.
        Appl. Biochem. Biotechnol. 2007; 137: 195-204
        • Matthews B.W.
        • Nicholson H.
        • Becktel W.J.
        Enhanced protein thermostability from site-directed mutations that decrease the entropy of unfolding.
        Proc. Natl. Acad. Sci. U.S.A. 1987; 84: 6663-6667
        • Muslin E.H.
        • Clark S.E.
        • Henson C.A.
        The effect of proline insertions on the thermostability of a barley α-glucosidase.
        Protein Eng. 2002; 15: 29-33
        • Textor L.C.
        • Colussi F.
        • Silveira R.L.
        • Serpa V.
        • de Mello B.L.
        • Muniz J.R.C.
        • Squina F.M.
        • Pereira Jr., N.
        • Skaf M.S.
        • Polikarpov I.
        Joint X-ray crystallographic and molecular dynamics study of cellobiohydrolase I from Trichoderma harzianum: deciphering the structural features of cellobiohydrolase catalytic activity.
        FEBS J. 2013; 280: 56-69
        • Turunen O.
        • Etuaho K.
        • Fenel F.
        • Vehmaanperä J.
        • Wu X.
        • Rouvinen J.
        • Leisola M.
        A combination of weakly stabilizing mutations with a disulfide bridge in the α-helix region of Trichoderma reesei endo-1,4-β-xylanase II increases the thermal stability through synergism.
        J. Biotechnol. 2001; 88: 37-46
        • Berka R.M.
        • Barnett C.C.
        The development of gene expression systems for filamentous fungi.
        Biotechnol. Adv. 1989; 7: 127-154
        • Cao Q.N.
        • Stubbs M.
        • Ngo K.Q.
        • Ward M.
        • Cunningham A.
        • Pai E.F.
        • Tu G.C.
        • Hofmann T.
        Penicillopepsin-JT2, a recombinant enzyme from Penicillium janthinellum and the contribution of a hydrogen bond in subsite S3 to kcat.
        Protein Sci. 2000; 9: 991-1001
        • Sambrook J.
        • Fritsch E.F.
        • Maniatis T.
        Molecular Cloning: A Laboratory Manual. 2nd Ed. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY1989
        • Walseth C.S.
        The influence of the fine structure of cellulose on the action of cellulases.
        TAPPI. 1952; 35: 233-238
        • Wood T.M.
        The cellulase of Fusarium solani: purification and specificity of the β-(1→4)-glucanase and the β-d-glucosidase components.
        Biochem. J. 1971; 121: 353-362
        • Rignall T.R.
        • Baker J.O.
        • McCarter S.L.
        • Adney W.S.
        • Vinzant T.B.
        • Decker S.R.
        • Himmel M.E.
        Effect of single active-site cleft mutation on product specificity in a thermostable bacterial cellulase.
        Appl. Biochem. Biotechnol. 2002; 98: 383-394
        • Ståhlberg J.
        • Divne C.
        • Koivula A.
        • Piens K.
        • Claeyssens M.
        • Teeri T.T.
        • Jones T.A.
        Activity studies and crystal structures of catalytically deficient mutants of cellobiohydrolase I from Trichoderma reesei.
        J. Mol. Biol. 1996; 264: 337-349
        • McPherson A.
        Preparation and Analysis of Protein Crystals.
        Wiley, New York1982
        • Matthews B.W.
        Solvent content of protein crystals.
        J. Mol. Biol. 1968; 33: 491-497
        • Kantardjieff K.A.
        • Rupp B.
        Matthews coefficient probabilities: improved estimates for unit cell contents of proteins, DNA, and protein–nucleic acid complex crystals.
        Protein Sci. 2003; 12: 1865-1871
        • Evans P.
        Scaling and assessment of data quality.
        Acta Crystallogr. D Biol. Crystallogr. 2006; 62: 72-82
        • Kabsch W.
        XDS.
        Acta Crystallogr. D Biol. Crystallogr. 2010; 66: 125-132
        • Winn M.D.
        • Ballard C.C.
        • Cowtan K.D.
        • Dodson E.J.
        • Emsley P.
        • Evans P.R.
        • Keegan R.M.
        • Krissinel E.B.
        • Leslie A.G.W.
        • McCoy A.
        • Leslie A.G.
        • McCoy A.
        • McNicholas S.J.
        • Murshudov G.N.
        • Pannu N.S.
        • et al.
        Overview of the CCP4 suite and current developments.
        Acta Crystallogr. D Biol. Crystallogr. 2011; 67: 235-242
        • Brünger A.T.
        Free R value: a novel statistical quantity for assessing the accuracy of crystal structures.
        Nature. 1992; 355: 472-475
        • Divne C.
        • Ståhlberg J.
        • Reinikainen T.
        • Ruohonen L.
        • Pettersson G.
        • Knowles J.K.
        • Teeri T.T.
        • Jones T.A.
        The three-dimensional crystal structure of the catalytic core of cellobiohydrolase I from Trichoderma reesei.
        Science. 1994; 265: 524-528
        • Murshudov G.N.
        • Vagin A.A.
        • Dodson E.J.
        Refinement of macromolecular structures by the maximum-likelihood method.
        Acta Crystallogr. D Biol. Crystallogr. 1997; 53: 240-255
        • Harris M.
        • Jones T.A.
        Molray: a web interface between O and the POV-Ray ray tracer.
        Acta Crystallogr. D Biol. Crystallogr. 2001; 57: 1201-1203
        • Bernstein F.C.
        • Koetzle T.F.
        • Williams G.J.B.
        • Meyer Jr., E.F.
        • Brice M.D.
        • Rodgers J.R.
        • Kennard O.
        • Shimanouchi T.
        • Tasumi M.
        The Protein Data Bank: a computer-based archival file for macromolecular structures.
        J. Mol. Biol. 1977; 112: 535-542
        • Day R.
        • Bennion B.J.
        • Ham S.
        • Daggett V.
        Increasing temperature accelerates protein unfolding without changing the pathway of unfolding.
        J. Mol. Biol. 2002; 322: 189-203
        • Huang X.
        • Zhou H.X.
        Similarity and difference in the unfolding of thermophilic and mesophilic cold shock proteins studied by molecular dynamics.
        Biophys. J. 2006; 91: 2451-2463
        • Sham Y.Y.
        • Ma B.
        • Tsai C.J.
        • Nussinov R.
        Thermal unfolding molecular dynamics simulation of Escherichia coli dihydrofolate reductase: thermal stability of protein domains and unfolding pathway.
        Proteins Struct. Funct. Genet. 2002; 46: 308-320
        • Sheinerman F.B.
        • Brooks 3rd, C.L.
        Calculations on folding of segment B1 of streptococcal protein G.
        J. Mol. Biol. 1998; 278: 439-456
        • Knott B.C.
        • Haddad Momeni M.
        • Crowley M.F.
        • Mackenzie L.F.
        • Götz A.W.
        • Sandgren M.
        • Withers S.G.
        • Ståhlberg J.
        • Beckham G.T.
        The mechanism of cellulose hydrolysis by a two-step, retaining cellobiohydrolase elucidated by structural and transition path sampling studies.
        J. Am. Chem. Soc. 2014; 136: 321-329
        • Taylor C.B.
        • Payne C.M.
        • Himmel M.E.
        • Crowley M.F.
        • McCabe C.
        • Beckham G.T.
        Binding site dynamics and aromatic-carbohydrate interactions in processive and non-processive family 7 glycoside hydrolases.
        J. Phys. Chem. B. 2013; 117: 4924-4933
        • Payne C.M.
        • Bomble Y.J.
        • Taylor C.B.
        • McCabe C.
        • Himmel M.E.
        • Crowley M.F.
        • Beckham G.T.
        Multiple functions of aromatic-carbohydrate interactions in a processive cellulase examined with molecular simulation.
        J. Biol. Chem. 2011; 286: 41028-41035
        • Anandakrishnan R.
        • Aguilar B.
        • Onufriev A.V.
        H++ 3.0: automating pK prediction and the preparation of biomolecular structures for atomistic molecular modeling and simulations.
        Nucleic Acids Res. 2012; 40: W537-W541
        • Gordon J.C.
        • Myers J.B.
        • Folta T.
        • Shoja V.
        • Heath L.S.
        • Onufriev A.
        H++: a server for estimating pKas and adding missing hydrogens to macromolecules.
        Nucleic Acids Res. 2005; 33: W368-W371
        • Myers J.
        • Grothaus G.
        • Narayanan S.
        • Onufriev A.
        A simple clustering algorithm can be accurate enough for use in calculations of pKs in macromolecules.
        Proteins. 2006; 63: 928-938
        • Brooks B.R.
        • Brooks 3rd, C.L.
        • Mackerell Jr., A.D.
        • Nilsson L.
        • Petrella R.J.
        • Roux B.
        • Won Y.
        • Archontis G.
        • Bartels C.
        • Boresch S.
        • Caflisch A.
        • Caves L.
        • Cui Q.
        • Dinner A.R.
        • Feig M.
        • et al.
        CHARMM: the biomolecular simulation program.
        J. Comput. Chem. 2009; 30: 1545-1614
        • Hoover W.G.
        Canonical dynamics: equilibrium phase-space distributions.
        Phys. Rev. A. 1985; 31: 1695-1697
        • Nose S.
        • Klein M.L.
        Constant pressure molecular-dynamics for molecular-systems.
        Mol. Phys. 1983; 50: 1055-1076
        • Mackerell Jr., A.D.
        • Feig M.
        • Brooks 3rd, C.L.
        Extending the treatment of backbone energetics in protein force fields: Limitations of gas-phase quantum mechanics in reproducing protein conformational distributions in molecular dynamics simulations.
        J. Comput. Chem. 2004; 25: 1400-1415
        • MacKerell A.D.
        • Bashford D.
        • Bellott M.
        • Dunbrack R.L.
        • Evanseck J.D.
        • Field M.J.
        • Fischer S.
        • Gao J.
        • Guo H.
        • Ha S.
        • Joseph-McCarthy D.
        • Kuchnir L.
        • Kuczera K.
        • Lau F.T.
        • Mattos C.
        • et al.
        All-atom empirical potential for molecular modeling and dynamics studies of proteins.
        J. Phys. Chem. B. 1998; 102: 3586-3616
        • Jorgensen W.L.
        • Chandrasekhar J.
        • Madura J.D.
        • Impey R.W.
        • Klein M.L.
        Comparison of simple potential functions for simulating liquid water.
        J. Chem. Phys. 1983; 79: 926-935
        • Durell S.R.
        • Brooks B.R.
        • Bennaim A.
        Solvent-induced forces between 2 hydrophilic groups.
        J. Phys. Chem. 1994; 98: 2198-2202
        • Phillips J.C.
        • Braun R.
        • Wang W.
        • Gumbart J.
        • Tajkhorshid E.
        • Villa E.
        • Chipot C.
        • Skeel R.D.
        • Kalé L.
        • Schulten K.
        Scalable molecular dynamics with NAMD.
        J. Comput. Chem. 2005; 26: 1781-1802
        • Humphrey W.
        • Dalke A.
        • Schulten K.
        VMD: visual molecular dynamics.
        J. Mol. Graph. Model. 1996; 14: 33-38
        • Jones T.A.
        • Zou J.-Y.
        • Cowan S.W.
        • Kjeldgaard M.
        Improved methods for building protein models in electron density maps and the location of errors in these models.
        Acta Crystallogr. A. 1991; 47: 110-119
        • Kleywegt G.J.
        • Jones T.A.
        xdlMAPMAN and xdlDATAMAN: programs for reformatting, analysis and manipulation of biomacromolecular electron-density maps and reflection data sets.
        Acta Crystallogr. D Biol. Crystallogr. 1996; 52: 826-828
        • Kleywegt G.J.
        • Jones T.A.
        Detecting folding motifs and similarities in protein structures.
        Methods Enzymol. 1997; 277: 525-545
        • Kleywegt G.J.
        • Jones T.A.
        Phi/Psi-cology: Ramachandran revisited.
        Structure. 1996; 4: 1395-1400
        • Payne C.M.
        • Jiang W.
        • Shirts M.R.
        • Himmel M.E.
        • Crowley M.F.
        • Beckham G.T.
        Glycoside hydrolase processivity is directly related to oligosaccharide binding free energy.
        J. Am. Chem. Soc. 2013; 135: 18831-18839