BIOTECHNOLOGICAL PRODUCTS AND PROCESS ENGINEERING

Influence of culture aeration on the cellulase activityof Thermobifida fusca

Yu Deng & Stephen S. Fong

Received: 26 May 2009 /Revised: 10 July 2009 /Accepted: 15 July 2009 /Published online: 21 August 2009# Springer-Verlag 2009

Abstract Currently, one of the hurdles hindering efficientproduction of cellulosic biofuel is the recalcitrant nature ofcellulose to hydrolysis. A wide variety of cellulase enzymesare found natively in microorganisms that can potentiallybe used to effectively hydrolyze cellulose to fermentablesugars. In this study, phenomenological and mechanisticparameters affecting cellulase activity were studied usingthe moderately thermophilic, aerobic, and cellulolyticmicroogainsm Thermobifida fusca. Two major sets ofexperiments were conducted to (1) study the mechanisticdifferences in growth in a flask compared to a bioreactorand (2) study the cell culture parameters influencingcellulase activity using a series of bioreactor experiments.Specific cellulase and specific endoglucanase activitieswere found to be higher in the bioreactor as compared toflask growth. Measurements of messenger RNA transcriptlevels of 18 cellulase-related genes and intracellular ATPlevels indicated that measured enzyme activity was likelymore influenced by post-transcriptional energetics ratherthan transcriptional regulation. By delineating the effects ofculture aeration and stir speed using a bioreactor, it wasfound that cellulase activity increased with increasingaeration and increasing stir speeds (highest Kla) with atradeoff of decreased cellular growth at the highest stirspeeds tested (400 rpm). Overall, these results allude to aconnection between aeration and oxidative respiration thatlead to increased ATP allowing for increased cellulasesynthesis as the primary constraint on overall cellulaseactivity.

Keywords Cellulase . Gene expression . Thermobifidafusca . Bioreactor

Introduction

Cellulose is the most common organic compound on theEarth and can be potentially used as source material forbiofuel production. Cellulose is a linear condensationpolymer consisting of D-anhydro-glucopyranose joinedtogether by β-1, 4-glycosidic bonds with a degree ofpolymerization from 100 to 20,000(OSullivan 1997).Efficient processes for converting cellulose to biofuel canpotentially provide a sustainable energy source; however,cellulose is a difficult polymer to use for chemicalprocessing because the molecules form tightly packed,extensively hydrogen-bonded microfibrils, which are em-bedded in the plant cell wall matrix (Lao and Wilson 1996).Currently, biofuel production processes incorporate avariety of pretreatment and hydrolysis steps to convertcellulose into glucose monomers using cellulase enzymes.The most common source of cellulase enzymes are micro-organisms such as fungi, yeast, and bacteria (Demain et al.2005; Zhang et al. 2006) where cellulases are nativelyproduced.

Thermobifida fusca (formerly Thermomonaspora fusca)is an aerobic, moderately thermophilic, filamentous soilbacterium (Lykidis et al. 2007). It is a major degrader ofplant cell walls in heated organic materials, such ascompost piles, rotting hay, or manure piles (Bachmannand Mccarthy 1991). The extracellular enzymes producedby T. fusca, including cellulases, have been studiedextensively because of their thermostability, utility througha broad pH range (4–10) and high activity (Lykidis et al.2007; Wilson 2004; Saloheimo et al. 1988).

Y. Deng : S. S. Fong (*)Department of Chemical and Life Science Engineering,Virginia Commonwealth University,601 W. Main Street,Richmond, VA 23284, USAe-mail: ssfong@vcu.edu

Appl Microbiol Biotechnol (2010) 85:965–974DOI 10.1007/s00253-009-2155-9

The widely accepted mechanism for enzymatic hydro-lysis of cellulose involves synergistic actions by endoglu-canase (EC 3.2.1.4), exoglucanase (or cellobiohydrolase)(EC 3.2.1.91), and β-glucosidase (EC3.2.1.21). Endo-glucanases hydrolyze accessible intramolecular β-1,4-glucosidic bonds of cellulose chains randomly to producenew chain ends; exoglucanases processively cleave cellu-lose chains at the ends to release soluble cellobiose orglucose(Zhang et al. 2006) and β-glucosidases hydrolyzecellobiose to glucose(Spiridonov and Wilson 2001). Todate, six different cellulases have beeen purified andidentified in T. fusca. Among them, three of them areendocellulases, Cel9B, Cel6A, and Cel5A (Ai and Wilson2002; Bevillard et al. 2004; Jung et al. 2003), two of themare exocellulases, Cel6B and Cel48A (Irwin et al. 2000;Watson et al. 2002), and one is a novel processiveendocellulase, Cel9A (Zhou et al. 2004). In addition tocellulases, T. fusca also produces and secretes hemi-cellulases. Of the hemicellulases found in T. fusca, twoare xylanases (xyl11A and xyl10B), one is a xyloglucanase(xg74A), and one is a mannanase (manB) (Beki et al.2003; Kim et al. 2004; Wilson 2004). The cellulases in T.fusca act synergistically to convert insoluble cellulose tocellobiose, glucose, or other soluble sugars.

Biosynthesis of cellulases in T. fusca is thought to beinfluenced by cellobiose induction and catabolite repression,with any sugar that can be metabolized acting as a repressor(Spiridonov and Wilson 1999). This regulatory mechanismwas partially confirmed by studying the relationship betweenCel5A transcript and the level of the protein expression ofthis gene (Lin and Wilson 1988). By identification of allcellulase-related extracellular proteins produced by T. fuscagrown on cellobiose and glucose, cellobiose was found to bean inducer of cellulolytic genes (Chen and Wilson 2007).Endocellulases and exocellulases showed distinctly differentregulation patterns, with exocellulases showing the highestlevel of induction (Spiridonov and Wilson 1998). Transcrip-tional regulation of some specific cel genes was found to becorrelated to a 14-bp inverted repeat DNA sequences (5′-TGGGAGCGCTCCCA) that acts as a cis-acting element(Lin and Wilson 1988). The gene celR has been isolated andidentified as the transcription regulator that binds to the14-bp DNA sequence (Spiridonov and Wilson 1999).

Although most of the recent research on T. fusca hasconcentrated on regulation of cellulase-related genes andsynergies in cellulase activity, few studies about the effect ofgeneral culture conditions on cellulase production have beenpublished. Based upon an initial observation of differentgrowth characteristics of T. fusca cultured in a flask ascompared to a bioreactor, a series of experiments wasconducted to investigate cellular mechanisms resulting indifferences in cellulase activity as a consequence of thedifferent culture systems. Our initial goal was to confirm

differences in growth characteristics and cellulase activitydue to growth in a shaken Erlenmeyer flask as compared to astirred bioreactor. Two series of more detailed experimentswere conducted to identify mechanistic changes in differentgrowth conditions and to delineate the contributions ofaeration and stir speed to cellulase activity. In the course ofthese experiments, specific cellulase activity, specific endo-glucanase activity, intracellular ATP levels, messenger RNA(mRNA) transcript levels for 18 cellulase-related genes, andthe oxygen transport coefficient, Kla, were determined.

Materials and methods

Culture conditions

T. fusca ATCC BAA-629 was grown in Hagerdahl medium(Ferchak and Pye 1983) containing 1.0% cellobiose. Forexperiments conducted in Erlenmeyer flasks, 50 mL pre-cultures of T. fusca were grown at 55°C and 250 rpm for24 h in a 500 mL Erlenmeyer flask. Growth cultures fortesting were inoculated using 5% of the pre-culture andgrown at 55°C and 250 rpm for 42–48 h. For bioreactorexperiments, 200 mL pre-cultures were grown for 24 h, andused to inoculate the bioreactor (B. Braun BiotechInternational, Allentown, PA) with a 4-L working volume.The bioreactor had a jacketed glass vessel with stainlesssteel-top plate and baffle insert. The bioreactor experimentswere conducted with four baffles and two six-bladed-diskimpellers for mixing. Cells were cultured in the bioreactorat 55°C for 42–48 h. Stir speeds and aeration rates varieddepending upon the specific experiment. To determine theoxygen supply in the bioreactor, the volumetric oxygentransfer coefficient (Kla) was calculated at different stirspeeds and aeration rates (Bandyopadhyay et al. 1967).

Enzyme activity assay

Two assays were used to measure the overall cellulase andendoglucanase activity of culture supernatants. Filter paperwas used as the starting material to measure overallcellulase activity, and endoglucanase activity was assayedby the measurement of reducing sugars generated from0.5% medium-viscosity carboxymethylcellulose (CMC).All the enzyme activity assays were measured by amircoplate-based assay (Xiao et al. 2004). Assays wereconducted in a 60-μl volume as follows. A 20-μl aliquot ofraw enzyme was added into the wells of PCR platescontaining 40 μl of 50 mM NaAc buffer and a filter paperdisk (7-mm diameter) for cellulase activity or 40 μl of50 mM NaAc buffer with 0.5% CMC for endoglucanaseactivity. After 60 min of incubation at 50°C, 120 μl of 3,5-dinitrosalicylic acid solution was added into each reaction

966 Appl Microbiol Biotechnol (2010) 85:965–974

and incubated at 95°C in a thermal cycler (iCycler®Ther-mocycler, BioRad, Hercules, CA) for 5 min. Finally, a36-μl aliquot of each sample was transferred to the wells ofa flat-bottom plate containing 160 μl of H2O, and theabsorbance at 540 nm was measured on VersaMax EXTmicroplate reader (Molecular Devices, Sunnyvale, CA).One enzyme unit (U) is defined as an average of 1 μmol ofcellobiose equivalent released per min in the assay reaction.All the enzyme activity values presented were averagesobtained from triplicate measurements.

Cell density measurement

Due to the growth physiology of T. fusca (filamentous cellsthat aggregate), the culture density of T. fusca was determinedby measuring cytoplasmic protein content. One milliliterculture was centrifuged at 10,000×g for 5 min. The pelletswere re-suspended in fresh media and centrifuged at10,000×g for 5 min again. Sediments were dissolved in200 μl 50 mM Tris–HCl buffer (pH6.8) containing 2%sodium dodecyl sulfate, 0.1 M dithiothreitol and 50%glycerol. Samples were then pulsar sonicated at 70% strengthfor 10 min in an ice bath. After centrifuging at 10,000×g for5 min, the proteins in the supernatant were measured by theBradford protein assay (Bradford 1976). The dry cell weight(DCW) is proportionally related to the overall protein content.

ATP determination

ATP was measured by the ATP determination kit fromInvitrogen (Carlsbad, CA) using the protocol suggested bythe manufacturer. Specific ATP levels were defined as theamount of ATP divided by the amount of cytoplasmicproteins in T. fusca. For ATP assays, cells were sampledduring mid-log phase growth. Experiments were conductedto assay for ATP levels for growth in a bioreactor at 55°Cwith the 0.375 vvm aeration and a stir speed of 250 rpmand in 500-ml flasks at 55°C that were shaken at 250 rpm.One milliliter culture was centrifuged at 10,000×g for10 min. The pellets were resuspended in the purified waterand centrifuged at 10,000×g for 5 min again. Samples werethen pulsar sonicated at 10% strength for 5 min in the icebath to release the ATP. After centrifuging at 10,000×g for5 min, 10 μl supernatant was added into the wells of PCRplates containing 90 μl reaction solutions. After incubatingin the VersaMax EXT microplate reader (MolecularDevices, Sunnyvale, CA) for 30 s at room temperature,the absorbance was measured at 560 nm.

RNA preparation and real-time PCR

To study molecular-level differences in the cultures of T.fusca, gene expression was studied using real-time PCR

using cells harvested at the mid-log growth phase and earlystationary phase corresponding to experimental pointsexhibiting different cellulase activity. Cells at the selectedpoints were centrifuged at 10,000×g for 5 min. The cellpellets were resuspended in RNAprotect bacteria reagent(Qiagen, Valencia, CA) as proscribed by the manufacturer.After incubation at room temperature for 5 min, the cellswere pulsar sonicated at 10% strength for 2 min. RNeasyMidi kits (Qiagen, Valencia, CA) were then used to isolateRNA using the protocol suggested by the manufacturer.The real-time PCR measurements were performed in theABI Prism® 7900 Sequence Detection System (AppliedBiosystems, Foster City, CA) using the TaqMan® One StepPCR Master Mix Reagents Kit (P/N, 4309169). The cyclingconditions were 48°C/30 min, 95°C/10 min, and 40 cyclesof 95°C/15 s and 60°C/1 min. The cycle threshold wasdetermined to provide the optimal standard curve values(0.98–1.0).The probes and primers (Table 1) were designedusing Primer Express® 3.0. 18 different cellulase-relatedgenes were measured along one housekeeping gene(Tfu_2950) that was chosen as a control to be consistentwith previously published real-time PCR measurements inT. fusca (Chen and Wilson 2007). All reported transcriptlevels were normalized to this housekeeping gene. Theprobes were labeled at the 5′ end with 6-carboxyfluoresceineand at the 3′ end with 6-carboxytetramethylrhodamine.Triplicate measurements of all data points were measuredusing independent biological replicates.

Results

In this study, the filamentous actinomycete, T. fusca, wasstudied to characterize differences in growth characteristicsand cellulase activity resulting from growth in either shakenErlenmeyer flasks or in a stirred bioreactor. Measurementsof growth characteristics, cellulase activity, endoglucanaseactivity, and mRNA transcript levels were taken to compareflask and bioreactor cultures. mRNA transcript levels for 18cellulase-related genes(Chen and Wilson 2007) weremeasured using real-time PCR to associate molecularmechanisms with measured cellulase activity. Finally, aseries of controlled bioreactor experiments were conductedto determine the affects of aeration and stir speed oncellulase activity and cell growth of T. fusca.

Flask versus bioreactor experiments

T. fusca growth in bioreactor and flask

Initial growth experiments were conducted to comparegrowth phenotypes of T. fusca in a bioreactor and a shakenErlenmeyer flask. T. fusca was cultured in a bioreactor

Appl Microbiol Biotechnol (2010) 85:965–974 967

system at 55°C in 1 wt.% cellobiose Hagerdahl mediumwith 0.375 vvm aeration and a stir speed of 250 rpm.Growth in 500-mL flasks was conducted at 55°C in 1 wt.%cellobiose Hagerdahl medium with flasks shaken at250 rpm. The cultures in the flask and bioreactorwere intentionally carbon source limited because a highcellobiose concentration would inhibit cellulase synthesis(Zhang et al. 2006). At a macroscopic level, there werenoticeable differences in the gross morphology of T. fuscain the two culture conditions. Cells grown in the bioreactoraggregated into small balls/clumps of cells, whereas cellsgrown in flasks produced mycelia that connected cells in aweb-like fashion. In terms of growth phenotypes, cellobiosewas utilized at similar rates in the flask and bioreactor, butthe cellular growth rates and DCW yields were significantlydifferent (Fig. 1). The doubling time of cells in shakenflasks and bioreactors was 3.37 and 4.59 h, respectively,and growth in flasks produced a higher maximum yield ofcell mass (7.51 mg/mL in flask vs. 6.39 mg/mL inbioreactor).

T. fusca cellulase activity in bioreactor and flask

Concurrent with the growth phenotype characterization ofT. fusca grown in a bioreactor or flask, measurements forboth overall cellulase activity and endoglucanase activitywere taken. All known cellulases in T. fusca are secreted(Wilson 2004), so cellulase and endoglucanase activity

were measured using culture supernatants. Specifically,overall cellulase activity was assayed through the hydro-lysis of filter paper, and endoglucanase activity was assayedthrough the hydrolysis of CMC.

There was a consistent trend where the total cellulaseand endoglucanase activity in the flask was higher than thatin the bioreactor (Fig. 2a, b). For overall cellulase activity,flask cultures reached a maximum cellulase activity of1.96±0.13 U/mL (compared to a maximum cellulase

Table 1 Primers used for real-time PCR

Locus Forward primer Reverse primer Probe

Tfu_1627 5′CTCCGCTGTGAGGTCATCCT 5′GGATCGAGATCTGGGTCTCCTA 5′CAGTTCGAATCCCCCTACGGAAGGAC

Tfu_1074 5′CCGTCCCGCCTACATCATC 5′TGGACGTGCTGCATGCA 5′CGGACCTGATCTCGCTGATGTCGA

Tfu_0620 5′GCTGGGTCCCGATGAACTC 5′CAGGTTCTCGTAGTCTGCGAAGT 5′ACATCGACCCGATCGCCGACA

Tfu_2176 5′CTTGGCGAAGGCCGAGTA 5′ATTTTTGCTTGCCGGTTTCC 5′CTCTCCACCGAGCAGCAGACCGA

Tfu_0901 5′GACGCCCTGGCCTACGA 5′AGCCGTCTTCCTGGATGTACA 5′TGGAAGGCCGACATCATCCGC

Tfu_1959 5′CGTCGCCGAAGTCAAGGT 5′GCCGCTCTGGCTGAACAC 5′TCCCCGGTAACCAGCAGATCACCA

Tfu_1268 5′CACCGACGTCAACAACACCTT 5′TGGTGACGTAGACCAGGAAGTAGTC 5′CCATCCACCTGTACGACCAGGCCA

Tfu_1665 5′AACCCCGCTTGCCAGAA 5′GAGGTTGCCGAACCAGTTGT 5′CTGCTTGCGGAGAACGGCAACTACC

Tfu_0934 5′AGGTCGGTAAGAAGTTCCAGGAA 5′CGTTGACGGCGCTGAAG 5′CCTTCATCTCCACGACCGTCCCCT

Tfu_0937 5′TCTCGAGGAGACTCCCAAACC 5′AGCCTTCGATCTGGAACGAA 5′ATATCCGCTTCCCGTCCGATTTCGT

Tfu_0938 5′CTGGTGGTGTCGGAAAACAA 5′CCCGTTCGGACAGAGCAA 5′AACCCTTCTATGCCGGGATCGTGC

Tfu_2923 5′GCTCAACAGTGACGCCCAGTA 5′TCTCGTTTTCGTGGGTGATG 5′CGCAACATCGCGGCTACCCAGT

Tfu_2788 5′GGACTACACCGTGGTCAATGACT 5′CTGCAGGGTCCAGTTGTTGAT 5′TCACCGTCTCCAACACCGGATCCT

Tfu_1213 5′TGGAGTACTACATCGTCGAAAGCT 5′CGTGGTCTTGTAGATGTCGTAGGT 5′ACCTACATGGGCACGGTGACCACC

Tfu_2791 5′CCCCGGAAAACCAGATGAA 5′CGGCCATGGCGAAGTC 5′TCCATCCTGAGCCGGACCGCTA

Tfu_1612 5′CACGGAGGAACCGTCCACT 5′AGGAGACAGCGCAGGCG 5′CGCCAACGGAGGAGCCGC

Tfu_2990 5′CTACAGCACGCTGGTCAGTCA 5′TGATCTGGGCCACGAGGAT 5′ATGCGGGCCAACAACCCGAAC

Tfu_0900 5′ACGAACACCATGCGGAACA 5′CCGTACATGTGGATCGAGAAGA 5′CAGGTGTACGCCAGCGACCCCA

Housekeeping(Tfu_2950)

5′CATCGCCTGCCTGATGCT 5′GCCATCACCGCCTTGAAG 5′TTCGCCGATGACGACGCGC

Fig. 1 Dry cell weight (DCW, g/L) and amount of reducing sugar (g/L)during growth in a flask or bioreactor. Cellobiose was chosen as thestandard reducing sugar, and dry cell weight was measured by assayingfor total protein content

968 Appl Microbiol Biotechnol (2010) 85:965–974

activity of 1.69±0.12 U/mL in the bioreactor) and main-tained a higher level of activity throughout the stationaryphase (Fig. 2a). Endoglucanase activity was comparable inboth growth systems (maximum activity of 1.41±0.18 U/mLfor bioreactor and 1.59±0.54 U/mL for flask) as shown inFig. 2b.

Due to the observed differences in cell yield, specificcellulase and specific endoglucanase activity were alsocalculated to determine enzyme activity per microgram ofDCW. The specific cellulase and endoglucanase activitieswere significantly different for growth in a bioreactorcompared to growth in a flask (Fig. 2c, d) with cells inthe bioreactor exhibiting higher per cell overall cellulaseand endoglucanase activity. T. fusca in the bioreactorexhibited a maximum specific cellulase activity of 0.718±0.079 U/mg and cells grown in flasks showed a maximumspecific cellulase activity of 0.327±0.003 U/mg.

A similar result was found for specific endoglucanaseactivity, where cells in the bioreactor had a maximumspecific endoglucanase activity of 0.560±0.021 U/mg(compared to a maximum of 0.287±0.040 U/mg in flasks)and an average specific endoglucanase activity of 0.429±0.026 U/mg (compared to an average of 0.199±0.033 U/mgin flasks).

Cytoplasmic ATP level of T. fusca in bioreactor and flask

Due to the potentially large metabolic cost associated withproducing high levels of cellulases, measurements of

intracellular ATP levels were also conducted. IntracellularATP levels were measured for cultures grown in the flasksshaken at 250 rpm and in a bioreactor with aeration at0.375 vvm and stirring at 250 rpm (Fig. 3). Duringexponential growth, the specific ATP level in the bioreactor(maximum 2.898±0.462 ng/mg) was significantly higherthan that in the flask (maximum 1.086±0.328 ng/mg). Asexpected, once cells reached stationary phase, the intracel-lular ATP levels were similar between the two cultureconditions.

Gene expression of cellulase genes

In addition to the growth phenotypes and cellulase andendoglucanase assays, real-time PCR was used to studymolecular-level changes. According to the genome se-quence of T. fusca (Lykidis et al. 2007), 18 genes (Chenand Wilson 2007) were identified to be potentially relatedto cellulase synthesis. Gene expression of these 18 genes(Table 2) was studied using triplicate real-time PCRmeasurements during exponential growth and stationaryphase to correspond with measured differences in cellulaseactivity. In general, the measured mRNA transcript levelsfor the 18 genes studied were higher in the flask culturesthan in the bioreactor. During exponential growth, 16 of the18 genes had statistically higher transcript levels (p<0.01)in the flask as compared to the bioreactor. The remainingtwo genes showed no statistical difference between the twoconditions. In stationary phase, ten of the 18 genes

Fig. 2 Cellulolytic enzymeactivities in the flask andbioreactor of T. fusca YX.Cellulase activity (U/mL) wasassayed using filter paper, andendoglucanase activity (U/mL)was measured using CMCas the substrate. Specificenzyme activity (U/mg) wascalculated by dividing enzymeactivities by dry cell weight.a Cellulase activity, b endoglu-canase activity, c specificcellulase activity; d specificendoglucanase activity

Appl Microbiol Biotechnol (2010) 85:965–974 969

maintained higher mRNA levels (p<0.01) in the flask ascompared to the bioreactor with four genes showing nostatistical difference and four genes having statisticallyhigher transcript levels in the bioreactor. The four genes

that showed higher mRNA transcript levels in stationaryphase in the bioreactor were two exocellulases (Tfu_0620and Tfu_1959), one transporter (Tfu_0934), and onemannanase (Tfu_0900).

One gene of specific interest is celR (Tfu_0938), whichhas previously been identified as a regulator of severalcellulase genes. Results for both exponential growth andstationary phase showed that there were consistently highertranscript levels of celR in flask growth experiments than inbioreactor growth (p<0.01). Eight out of 18 genes in thisstudy have the perfect 14-bp palindrome on their 5′-upstream region that is recognized by celR: Tfu_1627,Tfu_1074, Tfu_0620, Tfu_2176, Tfu_0901, Tfu_1959,Tfu_1665, and Tfu_0934. It was observed that transcriptlevels of all of these genes mirrored the transcript levels ofcelR (in this case, increased with increased amounts ofcelR). The gene Tfu_2923 has an imperfect 14-bp palin-drome that has been proposed to be repressed by CelR. Ourresults show that transcript levels of Tfu_2923 increased asthe amount of celR increased, thereby indicating that celRlikely did not repress expression of Tfu_2923 as has beenpreviously suggested (Lykidis et al. 2007).

Fig. 3 Measured specific ATP levels (ng ATP/mg T. fusca protein) ofT. fusca in bioreactor and flask growth. Experiments were conductedat 55°C with bioreactor experiments aerated at 0.375 vvm and a stirspeed of 250 rpm and flask experiments shaken at 250 rpm

Table 2 Gene expressions in flask and bioreactor at exponential and stationary phase of Thermobifida fusca YX

No. Locus Gene Product (protein designation) Transcript ratio in

Exponential phase Stationary phase

Bioreactor Flask P>Fa Bioreactor Flask P>Fa

1 Tfu_1627 β-1,4-endoglucanase (Cel9B) 21.07±2.48 59.36±1.68 <0.01 12.04±0.82 12.11±0.82 NS

2 Tfu_1074 β-1,4-endoglucanase (Cel6A) 35.32±1.44 93.02±5.28 <0.01 10.93±0.93 19.07±0.27 <0.01

3 Tfu_0620 β-1,4-exocellulase (Cel6B) 17.63±0.62 34.94±1.08 <0.01 12.48±0.48 0.86±0.07 <0.01

4 Tfu_2176 β-1,4-endoglucanase (Cel9A), Progressive 72.95±5.74 169.51±17.82 <0.01 12.49±0.32 28.94±1.92 <0.01

5 Tfu_0901 Endo-1,4-β-glucanase (Cel5A) 26.98±0.67 51.85±3.81 <0.01 15.75±1.52 25.82±1.04 <0.01

6 Tfu_1959 β-1,4-exocellulase (Cel48A) 20.59±0.33 47.64±3.27 <0.01 15.17±2.15 5.30±0.26 <0.01

7 Tfu_1268 Predicted cellulose-binding protein (E7) 32.12±0.97 82.08±10.07 <0.01 14.16±1.04 17.8±1.47 NS

8 Tfu_1665 Predicted cellulose-binding protein (E8) 14.48±0.70 50.19±2.63 <0.01 12.96±0.30 17.62±1.59 <0.01

9 Tfu_0934 Predicted ABC-type sugar transportsystem, periplasmic component forcellobiose/cellotriose

43.76±0.75 138.12±4.02 <0.01 16.12±0.51 3.47±0.21 <0.01

10 Tfu_0937 β-glucosidase (BglC) 96.20±7.36 335.90±7.13 <0.01 15.84±0.44 15.27±1.99 NS

11 Tfu_0938 CelR 35.93±3.22 83.23±5.17 <0.01 8.23±0.57 24.65±2.64 <0.01

12 Tfu_2923 β-1,4-endoxylanase (Xyl10A) 16.86±1.62 19.53±1.69 NS 15.11±1.67 38.67±2.10 <0.01

13 Tfu_2788 Putative β-1,4-endoxylanase 41.94±0.88 80.57±9.57 <0.01 14.69±0.55 13.60±1.29 NS

14 Tfu_1213 β-1,4-xylosidase 26.67±2.06 33.40±2.49 NS 11.85±0.15 31.69±3.13 <0.01

15 Tfu_2791 β-1,4-endoxylanase (Xyl10B) 18.28±1.21 25.14±1.04 <0.01 15.74±0.69 40.16±4.82 <0.01

16 Tfu_1612 Putative secreted xyloglucanase 18.32±2.21 31.02±2.29 <0.01 14.83±1.07 50.99±0.64 <0.01

17 Tfu_2990 Putative secreted xylanase 17.67±1.18 32.89±1.19 <0.01 15.93±0.52 34.53±0.70 <0.01

18 Tfu_0900 Putative secreted β-mannanase 2.45±0.071 5.35±0.23 <0.01 17.62±1.16 5.64±0.70 <0.01

NS not significant (P>0.01), F variance ratioaP values for differences in expression between flask and bioreactor

970 Appl Microbiol Biotechnol (2010) 85:965–974

Bioreactor experiments

Factors influencing T. fusca phenotypes

Based upon our results that indicated different levelsof overall cellulase activity, endoglucanase activity, andcell growth between the growth in a shaken flask and astirred bioreactor, we conducted a series of controlledbioreactor experiments in an attempt to delineate thefactors influencing cellulase production and cell growth.While there may be additional parameters affectingcellulase activity, we focused on two of them that wereclearly different in the two conditions, agitation andaeration. Because T. fusca is a filamentous bacterium,the amount of agitation (and resulting shear forces)can have a direct impact on physiological functions(Gusek et al. 1991). Another major difference forgrowth in a flask versus a bioreactor is the level ofoxygen available to cells. All of the experiments in thefollowing sections were conducted using a bioreactorto allow for more precise control over each cultureparameter.

Bioreactor with different aeration rates

To determine if the supply of oxygen was a key factoraffecting cellulase production, a series of growth experi-ments with different aeration rates was conducted. Theculture stir speed was kept constant at 200 rpm, andaeration rates of 0.2, 0.5, and 1 vvm were tested. To helpdetermine the amount of oxygen available in each exper-iment, the mass transfer coefficient for oxygen, Kla, wasdetermined. Cells grown with 0.5 vvm sparging appearedto grow faster than the other two cultures (Fig. 4a).Kla was determined to be 10.43 h−1 at 0.2 vvm and16.02 and 16.13 for the 0.5 and 1 vvm experiments,respectively. Overall cell yields at 0.5 and 1 vvm werealmost the same but higher than at 0.2 vvm. It appearedthat above a critical aeration rate (0.5 vvm in this study),the cell growth of T. fusca was not appreciably affected byaeration rate.

The maximum observed overall cellulase activity at1 vvm (2.067±0.310 U/mL) was almost twofold higherthan at 0.5 vvm (1.079±0.439 U/mL) and 0.2 vvm (0.99±0.01 U/mL), as shown in Fig. 4b. Endoglucanase activitydid not appear to be strongly influenced by changes inaeration rate (Fig. 4b), with the highest measured endo-glucanase activities of 1.528±0.198 and 1.537±0.078 U/mLat 1 and 0.5 vvm, respectively. These results indicated thatthe overall cellulase activity (combined exocellulase andendoglucanase activity) in T. fusca increased with higheraeration rates.

Bioreactor with different stir speeds

As T. fusca is a filamentous actinomycete, it is possible thatdifferent stir speeds (and resulting shear forces) couldinfluence both cellular oxygenation and cellular physiologyrelated to morphology. Based upon the previous set ofresults with aeration rates, an aeration rate of 1 vvm wasselected for all experiments and stir speed was varied.Growth and cellulase activity were tested at three stirspeeds: 400, 200, and 50 rpm (all at 1 vvm aeration). Interms of cell growth rate and yield, cells grown at 50 and200 rpm showed similar results (Fig. 5a). Cells grown witha stir speed of 400 rpm grew slower and resulted insignificantly decreased cell yield (4.80 g/L compared to7.39 g/L at 50 rpm and 7.98 g/L at 200 rpm). As could beexpected, the highest Kla was 17.62 h−1 at 1 vvm andstirring at 400 rpm, but the high shear stress associated withthe high stir speed appeared to be greatly detrimental tocellular growth.

At a constant aeration rate, there were measurabledifferences in both overall cellulase (Fig. 5b) and endo-glucanase activities (Fig. 5b) at the different stir speeds.The highest overall cellulase activity was observed at400 rpm (2.086±0.207 U/mL) and 200 rpm (2.067±0.310 U/mL). These levels of cellulase activity were morethan 1.5 times higher than the cellulase activity at 50 rpm(1.317±0.129 U/mL). The highest endoglucanse activity(1.646±0.206 U/mL) was observed at 50 rpm (Fig. 5b).The endoglucanase at 200 and 400 rpm were 1.528±0.298and 1.262±0.013 U/mL, respectively. Because the oxygensupply was sufficient, cellulase production did not changeappreciably at stir speeds above 200 rpm. Endoglucanaseactivity appeared to be adversely affected by the high stirspeed.

Discussion

Improved understanding of cellulose degrading processesvia cellulase enzymes may lead to numerous bioprocessingapplications with cellulose as a starting material. Here, wefocus on the thermophilic, aerobic, and cellulolytic actino-mycete, T. fusca, to study how different culture parameters(aeration rate and stir speed) affect cellular mechanisms thatcan lead to increased cellulase production and activity.Using two sets of experiments, we first sought tocharacterize the mechanistic changes associated withmeasured differences in cellulase and endoglucanaseactivity. We then sought to delineate the effects of aerationand stir speed as culture parameters that influence thecellulase and endoglucanase activity of T. fusca. Mechanis-tically, we found that (1) mRNA transcript levels of

Appl Microbiol Biotechnol (2010) 85:965–974 971

cellulase-related genes were not closely associated withmeasured differences in specific cellulase and endogluca-nase activity and (2) cellular energetics (intracellular ATP)correlated more closely to changes in specific cellulaseactivity. In terms of culture system parameters, it was foundthat increasing the aeration rate and stir speed (to yielda high oxygen transfer coefficient) led to the highest cellu-lase activity but did not appear to affect endoglucanaseactivity.

Initially, it was observed that T. fusca exhibits markedlydifferent growth behaviors when grown in a flask or abioreactor. We found that cell yields were significantlydifferent for growth in a shaken flask as compared to abioreactor, so subsequent characterizations of cellulaseactivity were compared both as overall activity and specificactivity (normalized to cell mass). Initial comparisons oftotal cellulase activity, endoglucanase activity, and cellgrowth indicated that there were clearly different pheno-types due to differences of the culture conditions with flaskcultures having higher total cellulase and endoglucanaseactivity but bioreactor cultures having higher specificcellulase and specific endoglucanase activity.

Mechanistically, the expectation was that mRNA transcriptlevels (a per cell measurement) should be most closely relatedto the specific cellulase activity during exponential growth.

The mRNA transcript measurements did not appear tostrongly correlate with the measured specific cellulase orspecific endoglucanase activity. In both cases, mRNAtranscript levels were statistically significantly higher in theflasks than in the bioreactor, but specific cellulase and specificendoglucanase activites were higher in the bioreactor. Theonly genes that showed statistically increased mRNAtranscript levels in the bioreactor (as compared to growth inflasks) were two exocellulases (Tfu_0620 and Tfu_1959), onetransporter (Tfu_0934), and one mannanase (Tfu_0900). Oneconclusion could be that these genes, especially the twoexocellulases (Tfu_0620 and Tfu_1959), play a dominantrole in the overall cellulase activity. Interestingly, these twogenes represent two types of exoglucanase. One, Tfu_0620(Cel6B), hydrolyzes cellulose from the non-reducing end ofthe cellulose chain while the other Tfu_1959 (Cel48A) actsfrom the reducing end (Irwin et al. 2000). This combinationof exoglucanase activity can be very effective (synergistic) atincreasing the efficiency of cellulose hydrolysis and couldpotentially explain the higher observed total specific cellu-lase activity.

An alternative interpretation could be that the specificcellulase and specific endoglucanase activity are onlytangentially associated with mRNA transcript levels, andsubsequent biological steps (such as translation) may be

Fig. 4 The cell growth,maximum cellulase andendoglucanase activity of T.fusca at different aeration rateswith a constant stir speed of200 rpm for growth in abioreactor. a Cell growth,b maximum cellulose activity(U/mL), endoglucanase activity(U/mL), and Kla (h−1)

Fig. 5 The cell growth,maximum cellulase, andendoglucanase activity of T.fusca at different stir speedswith aeration of 1 vvm forgrowth in a bioreactor. a Cellgrowth, b maximum celluloseactivity (U/mL), endoglucanaseactivity (U/mL) and Kla (h−1)

972 Appl Microbiol Biotechnol (2010) 85:965–974

more dominant in dictating enzyme activity. By measuringthe intracellular ATP levels, we found that there wasapproximately a threefold increase in available ATP duringexponential growth in the bioreactor cultures as comparedto the flask cultures. While not conclusive, this result doessupport the idea that cellular energetics may have a largeinfluence on cellulase and endoglucanase activity more sothan mRNA transcription levels.

A secondary consequence of conducting measurementsof mRNA transcript levels was the ability to comment onpotential transcriptional regulatory mechanisms specificallyrelated to the previously identified regulatory gene, celR(Tfu_0938) (Spiridonov and Wilson 1999). CelR binds to a14-bp inverted repeat (5′-TGGGAGCGCTCCCA), andsome genes with imperfect palindromes (for instanceTfu_2923) are also thought to be regulated by CelR(Lykidis et al. 2007). Overall, CelR is thought to regulatethe following genes: cellulase-related (Tfu_0620, Tfu_0901,Tfu_0934, Tfu_1074, Tfu_1627, Tfu_1665, Tfu_1959,Tfu_2176, and Tfu_2923 that has the imperfect palindromesequence) and three non-cellulase related genes (Lykidis etal. 2007). Results for exponential growth showed noinconsistency with the idea that CelR could act as aregulator for the eight cellulase-related genes that have theperfect 14-bp repeat. Results from stationary phase showeda discrepancy in transcript levels between four of thesegenes (Tfu_0620, Tfu_0934, Tfu_1959, and Tfu_1627) andcelR levels indicating that the transcriptional regulatorymechanism controlling theses genes may be more complexduring stationary phase (not solely controlled by celR or theeffect of celR is altered). With regard to the imperfectregulatory sequence associated with Tfu_2923, it has beensuggested that celR acts to repress expression of Tfu_2923(Lykidis et al. 2007). Our results do not support thissuggestion, and the regulatory mechanisms for Tfu_2923and other cellulase-related genes remains an area requiringfurther detailed research.

Since T. fusca is aerobic and filamentous, we focused onaeration rate and stir speed (associated shear stress canaffect the formation of mycelia) when studying the culturesystem parameters that can influence cellulase activity. Aseries of bioreactor experiments studied the affect ofaeration rate and stir speed on cell growth, overall cellulaseactivity, and endoglucanase activity. It was found that, at aconstant stir speed (200 rpm), increasing the aeration rate(up to the 1 vvm tested in this study) resulted in an increasein the overall cellulase activity (combination of exocellu-lase and endoglucanase activity) but did not appreciablyaffect cell growth. At a constant high aeration rate (1 vvm)with various stir speeds, there was a tradeoff betweenoverall cellulase activity and endoglucanase activity (over-all cellulase activity increased with increased stir speedwhereas endoglucanase activity decreased with increased

stir speed). In addition, it was found that high stir speeds(above 200 rpm) were detrimental to cell growth.

Growth and cellulase assay results indicated that overallcellulase activity and endoglucanase activity were affecteddifferently by changes in culture parameters. Specifically,endoglucanase activity was much more sensitive to theagitation by stirring than exoglucanase (as evidence by thedifference in endoglucanase activity compared to overallcellulase activity), and exoglucanase activity appeared tocompensate for the reduction of endoglucanase to increaseoverall cellulase activity at the high stir speed.

Overall, we found that the culture parameters of aerationrate and stir speed strongly influence the growth character-istics and expression and activity of different cellulases inT. fusca. The specific cellulase and specific endoglucanaseactivity was found to be higher in a stirred, aeratedbioreactor than in a shaken flask. mRNA transcript levelsindicate that two exocellulases (Tfu_0620 and Tfu_1959)may play dominant roles in cellulase activity, but there isalso evidence that cellulase activity may be more closelyrelated to the energetics of a post-transcriptional event suchas protein translation. Measurement of intracellular ATPlevels show a large difference in available ATP withbioreactor cultures having nearly three times more ATPavailable. This large energetic difference likely plays animportant role in allowing increased protein synthesis(cellulase synthesis) in the cells grown in the bioreactor,thereby potentially explaining the measured differences inspecific cellulase and specific endoglucanase activity thatwere accounted for by mRNA transcript levels at exponen-tial phase. This is also consistent with our results showingthat the highest enzyme activities were found at the highestaeration rates where oxidative respiration could generatethe maximal amount of ATP. This perspective was onlypossible by utilizing an aerobic cellulolytic microbe andindicates that overall cellular energetics may be the mostdominant controlling factor in cellulase activity.

References

Ai YC, Wilson DB (2002) Mutation and expression of N233C–D506Cof cellulase Cel6B from Thermobifida fusca in Escherichia coli.Enzyme Microb Technol 30:804–808

Bachmann SL, Mccarthy AJ (1991) Purification and cooperativeactivity of enzymes constituting the xylan-degrading systemof Thermomonospora fusca. Appl Environ Microbiol 57:2121–2130

Bandyopadhyay B, Humphrey AE, Taguchi H (1967) Dynamicmeasurement of the volumetric oxygen transfer coefficient infermentation systems. Biotechnol Bioeng 9:533–544

Beki E, Nagy S, Vanderleyden J, Jager S, Kiss L, Fulop L, Hornok L,Kukolya J (2003) Cloning and heterologous expression of a beta-D-mannosidase (EC 3.2.1.25)-encoding gene from Thermobifidafusca TM51. Appl Environ Microbiol 69:1944–1952

Appl Microbiol Biotechnol (2010) 85:965–974 973

Bevillard E, Goodheart DB, Karnati SKR, Bayer EA, Lamed R, MironJ, Nelson KE, Morrison M (2004) Ruminococcus albus 8 mutantsdefective in cellulose degradation are deficient in two processiveendocellulases, Cel48A and Cel9B, both of which possess anovel modular architecture. J Bacteriol 186:136–145

Bradford MM (1976) Rapid and sensitive method for quantitation ofmicrogram quantities of protein utilizing principal of protein-dyebinding. Anal Biochem 72:248–254

Chen SL, Wilson DB (2007) Proteomic and transcriptomic analysis ofextracellular proteins and mRNA levels in Thermobifida fuscagrown on cellobiose and glucose. J Bacteriol 189:6260–6265

Demain AL, Newcomb M, Wu JHD (2005) Cellulase, clostridia, andethanol. Microbiol Mol Biol Rev 69:124–154

Ferchak JD, Pye EK (1983) Effect of glucose and other sugars on thebeta-1, 4-glucosidase activity of Thermomonospora fusca.Biotechnol Bioeng 25:2855–2864

Gusek TW, Johnson RD, Tyn MT, Kinsella JE (1991) Effect ofagitational shear on growth and protease production by Thermo-monospora fusca. Biotechnol Bioeng 37:371–374

Irwin DC, Zhang S, Wilson DB (2000) Cloning, expression andcharacterization of a Family 48 exocellulase, Cel48A, fromThermobifida fusca. Eur J Biochem 267:4988–4997

Jung H, Wilson DB, Walker LP (2003) Binding and reversibility ofThermobifida fusca Cel5A, Cel6B, and Cel48A and theirrespective catalytic domains to bacterial microcrystalline cellulose.Biotechnol Bioeng 84:151–159

Kim JH, Irwin D, Wilson DB (2004) Purification and characterizationof Thermobifida fusca xylanase 10B. Can J Microbiol 50:835–843

Lao GF, Wilson DB (1996) Cloning, sequencing, and expression of aThermomonospora fusca protease gene in Streptomyces lividans.Appl Environ Microbiol 62:4256–4259

Lin E, Wilson DB (1988) Transcription of the CelE gene inThermomonospora fusca. J Bacteriol 170:3838–3842

Lykidis A, Mavromatis K, Ivanova N, Anderson I, Land M, DiBartoloG, Martinez M, Lapidus A, Lucas S, Copeland A, Richardson P,Wilson DB, Kyrpides N (2007) Genome sequence and analysisof the soil cellulolytic actinomycete Thermobifida fusca YX. JBacteriol 189:2477–2486

OSullivan AC (1997) Cellulose: the structure slowly unravels.Cellulose 4:173–207

Saloheimo M, Lehtovaara P, Penttila M, Teeri TT, Stahlberg J,Johansson G, Pettersson G, Claeyssens M, Tomme P, KnowlesJKC (1988) EGIII, a new endoglucanase from Trichoderma reesei:the characterization of both gene and enzyme. Gene 63:11–21

Spiridonov NA, Wilson DB (1998) Regulation of biosynthesis ofindividual cellulases in Thermomonospora fusca. J Bacteriol180:3529–3532

Spiridonov NA, Wilson DB (1999) Characterization and cloning ofCelR, a transcriptional regulator of cellulase genes fromThermomonospora fusca. J Biol Chem 274:13127–13132

Spiridonov NA, Wilson DB (2001) Cloning and biochemicalcharacterization of BglC, a beta-glucosidase from the cellulolyticactinomycete Thermobifida fusca. Curr Microbiol 42:295–301

Watson DL,Wilson DB,Walker LP (2002) Synergism in binary mixturesof Thermobifida fusca cellulases Cel6B, Cel9A, and Cel5A onBMCC and Avicel. Appl Biochem Biotechnol 101:97–111

Wilson DB (2004) Studies of Thermobifida fusca plant cell walldegrading enzymes. Chem Rec 4:72–82

Xiao Z, Storms R, Tsang A (2004) Microplate-based filter paper assayto measure total cellulase activity. Biotechnol Bioeng 88:832–837

Zhang YHP, Himmel ME, Mielenz JR (2006) Outlook for cellulaseimprovement: screening and selection strategies. Biotechnol Adv24:452–481

Zhou WL, Irwin DC, Escovar-Kousen J, Wilson DB (2004) Kineticstudies of Thermobifida fusca Cel9A active site mutant enzymes.Biochemistry 43:9655–9663

974 Appl Microbiol Biotechnol (2010) 85:965–974