Scalable methanol-free production of recombinant glucuronoyl esterase in Pichia pastoris

Objective Glucuronoyl esterase (GE) is an emerging enzyme that improves fractionation of lignin-carbohydrate complexes. However, the commercial availability of GE is limited, which hinders the research of GE-based bioprocesses for its industrial application in lignocellulose biorefineries. This study evaluated a workable, cost-effective, and commercially scalable production strategy to improve the ease of GE-based research. This strategy consisted of a constitutive and methanol-free enzyme production step coupled with a two-step filtration process. The aim was to determine if this strategy can yield copious amounts of GE, by secretion into the extracellular medium with an acceptable purity that could allow its direct application. This approach was further validated for cellobiose dehydrogenase, another emerging lignocellulose degrading enzyme which is scarcely available at high cost. Results The secreted recombinant enzymes were functionally produced in excess of levels previously reported for constitutive production (1489–2780 mg L−1), and were secreted at moderate to high percentages of the total extracellular protein (51–94%). The constant glycerol feed, implemented during fed-batch fermentation, lead to a decline in growth rate and plateaued productivity. Tangential flow ultrafiltration was used to concentrate cell-free enzyme extracts 5–6-fold, reaching enzyme activity levels (1020–202 U L−1) that could allow their direct application.

Despite the industrial potential of GE, very limited recombinant production has been reported. Pure commercial preparations of GE do not exist. This is an impedance in the development of GE-based bioprocesses, since each study must first produce and purify the enzyme before it is possible to determine its lignocellulosic degrading capabilities. Some commercial enzyme cocktails exist with low GE side activity ( [14], this study), but these cocktails do not allow for the determination of specific catalytic roles of each enzyme. To harness the catalytic ability of GE, understand its specific role, and determine optimum dosages, pure enzyme preparations are required. Notably, the recent increase in publications which evaluate application of GE, show an urgent need for GE production [7][8][9][10]12].
Considering the above restriction in GE availability, this study sought to create a 'plug-and-play' system for production of large amounts of GE. The production Open Access BMC Research Notes *Correspondence: garcia@sun.ac.za; mdpgaparicio2012@gmail.com 1 Departments of Process Engineering, Stellenbosch University, Private Bag X1, Matieland 7602, South Africa Full list of author information is available at the end of the article process was designed with commercial and industrial foresight, seeking to keep the process as straightforward, safe, and cost effective as possible, while limiting any proprietary restrictions. The use of a constitutive, patentfree, P GAP -Pichia pastoris (glyceraldehyde-3-phosphate dehydrogenase promoter) expression system, to produce Hypocrea jecorina GE at bioreactor scale was investigated. Both the enzyme production itself as well as downstream processing of the enzyme into a stable product were considered (Additional file 1: Fig. S1). The feasibility of this approach was further tested for another emerging enzyme of limited commercial availability, Neurospora crassa cellobiose dehydrogenase (CDH). CDH finds application in multiple areas including bioremediation, textile, biomedicine, biosensors and biofuels [15]. CDH could improve the saccharification of pre-treated lignocellulose through reduction of end-product inhibition and catalytic activation of lytic polysaccharide monooxygenase [16]. This type of studies is necessary to facilitate research of the functionality and application of GE and other emerging enzymes such as CDH.

Construction of the production strains
The P. pastoris type strain DSMZ 70382 (CBS704) (DSMZ German Collection of Microorganisms and Cell Cultures) was selected as the expression host. The commercial pJexpress 905 (pJ905) vector (ATUM, USA) containing the P GAP and alcohol oxidase I (AOX1) terminator, was used to generate constructs for GE (AAP57749.1) and CDH (XM_951498.2) expression and secretion. Details of cloning, transformation, confirmation and screening procedures are described in [17]. The transformant showing the best production during screening in shake flasks for each enzyme was selected for production at bioreactor scale.

Bioreactor cultivations
The inoculum preparation and fermentation procedure were conducted as described in Invitrogen guidelines [18,19]. Fermentations were conducted in 14 L New Brunswick BioFlo 110 bioreactors, using BioCommand ® software (Version 3.30 Plus, New Brunswick Scientific Co. Inc.). The fermentation conditions were identical for all fermentations: 30 °C, pH 5 (combination glass pH probe Mettler Toledo), dissolved oxygen (DO) maintained at 30% (polarographic DOT probe, Mettler Toledo), and an aeration rate of 1 vvm. The fed-batch stage was initiated after depletion of glycerol, where 4 L of 50% (w v −1 ) glycerol feed, supplemented with PTM 1 trace salts solution (1.2% v v −1 ), was fed at a constant rate of 72.6 mL h −1 .
The fed-batch stage was concluded after approximately 48 h, when the feed was depleted.
Samples taken (10 mL) during fermentation were analysed for biomass (dry cell weight, g DCW L −1 ) enzyme activity (U L −1 ) as described in [17,20,21], and glycerol concentration [22] (during fed-batch stage). The final samples were used to determine total protein concentrations [bicinchoninic acid (BCA ™ ) microassay (Sigma-Aldrich, USA)] and target protein concentrations [densitometry of tris-tricine sodium dodecyl sulphate polyacrylamide gel electrophoresis (SDS-PAGE) [23] image using ImageJ ® software]. Maximum growth rate (μ max ) was graphically calculated during the exponential growth phase of the batch phase. All trends were fitted to a linear regression with R 2 values of above 0.98.

Concentration of crude enzyme extracts
The culture was harvested immediately after the fedbatch phase to minimize protease degradation of the enzyme product. The total culture was centrifuged (8000 rpm; 4 °C; 10 min) and subjected to a two-step filtration process. The Pellicon 2.0 tangential flow filtration apparatus (Merck, South Africa) was used as per the manual instructions. The feed and retentate pressure were maintained at a maximum of 10 and 2 bar, respectively. The transmembrane pressure (TMP) was 2.5 bar. The supernatant was first filtered through a 0.22 μm filter cassette (Durapore ® PVDF, Merck Millipore), then concentrated by ultrafiltration using a 5 kDa filter cassette (Biomax ™ 5, Merck Millipore). Once the volume of the retentate/feeding reservoir was decreased tenfold, the process was complete. Samples of each permeate and retentate were analysed for volumetric activity. Volumes of the feed, permeate and retentate of each filtration process were measured.

Glycerol fed-batch fermentation kinetics, yields and productivity for GE and CDH
The growth and enzyme production characteristics of the created recombinant strains were evaluated using bioreactor cultivations. GE side-activity of three commercial enzyme cocktails, namely Celluclast ® 1.5 L (Novozymes, 27.47 U L −1 ), Depol ™ 740 L A (Biocatalysts, 10.70 U L −1 ) and Depol ™ 740 L B (Biocatalysts, 7.72 U L −1 ) were tested to serve as a comparative baseline. The bioreactor cultivation for expression of GE returned a final volumetric activity yield of 238.17 U L −1 , and a final recombinant protein titre of 2778.01 mg L −1 (Fig. 1a; Table 1), well in excess of previous attempts involving methanol induction of GE [5,24]. Notably, the volumetric activity was well above the side activities of the commercial enzyme cocktails tested here (7.72-27.47 U L −1 ). Bioreactor cultivation for CDH production returned a final mean volumetric activity of 329.49 U L −1 and a final mean recombinant protein concentration of 1489.30 mg L −1 ( Fig. 1b; Table 1). Although the protein concentration of secreted CDH reported here (1489.30 mg L −1 ) is an improvement on previous attempts at methanol-induced expression, the volumetric activity and specific activity is lower than previously reported [21,[25][26][27]. Lowered specific activity has been observed in recombinant CDH production by P. pastoris due to a sub-stoichiometric occupation of catalytic sites within the flavin adenine dinucleotide (FAD) cofactor, as well as hyper-glycosylation [28]. Further, hyper-glycosylation may have affected the specific activity of both target enzymes, since the reported protein sizes (Fig. 1c; CDH: 127 kDa; GE: 78 kDa) are larger than the expected, in silico protein sizes (ExPASy: CDH: 88.46 kDa; GE: 81.29 kDa) [2,29].
The general trends observed for the fermentations are as follows (Fig. 1a, b). During the batch phase, an initial lag phase of less than 6 h was observed, corresponding to minimal volumetric activity, followed by an exponential growth phase, where linear increases in volumetric activity were observed until the end of the batch phase (between 20 and 22 h). During the glycerol fedbatch stage, there was a brief exponential growth phase of approximately 6 h, followed by an extended growth phase. During this continuous growth phase, the agitation of the bioreactor remained at the maximum (1000 rpm), and oxygen sparging was required throughout to maintain sufficient availability of DO (30%) in the culture.
A decrease in specific growth rate, modelled as a power regression (GE: 20979x −3.63 ; R 2 = 0.9871; CDH: 885.76x −2.852 ; R 2 = 0.9882), was observed during the fedbatch process (Additional file 2: Fig. S2), during which the levels of glycerol remained below detectable HPLC levels. This is attributed to the fact that the amount of biomass in the fermentation vessel increased with time (Additional file 2: Fig. S2), while the amount of glycerol fed into the bioreactor remained constant, effectively decreasing the ratio of available carbon source per gram of biomass. This growth rate decline has been suggested to be the concluding stage of an ideal growth rate time course for a number of secreted protein fermentations in P. pastoris [30,33,34].
Specific productivity describes the efficiency of the process by quantifying the amount of recombinant enzyme secreted per gram of biomass per hour. Maximum values of specific productivity (q p.max ) of protein secretion were reached during the batch phase  GE and CDH, respectively; Table 1). During the fedbatch stage, specific productivity remained constant at a lowered value close to the mean specific productivity value (q p.mean ) (0.52 and 0.43 mg recombinant protein.biomass −1 h −1 for GE and CDH, respectively), resulting in GE and CDH accumulation throughout the fed-batch stage (Fig. 1a, b). This has been observed in previous constant fed-batch cultivations of recombinant P GAP -P. pastoris [33,35]. The productivity could be increased by manipulating the growth rate such that an extended growth phase at μ max is maintained before the end-phase decline in growth rate [31,[33][34][35]. This is achieved by an increasing (linear or exponential) glycerol feeding rate during the fed-batch stage. The optimal duration of the phase at which μ max is maintained differs between recombinant strains and should be optimised on a case-by-case basis.

Concentration of crude enzyme extracts by tangential flow ultrafiltration (TFU)
The final secreted protein concentrations differed between enzymes, probably due to protein-specific factors associated with the secretion process. The heterologous proteins were secreted at moderate (CDH: 50.89%) to high (GE: 93.61%) percentages of the total extracellular protein, as determined by densitometry analyses (Fig. 1c). CDH and GE were, therefore, the major proteins present in the fermentation broth. While native, extracellular P. pastoris proteins were present in the fermentation broth (Fig. 1c), these were devoid of any contaminating enzymatic activity (as determined with the use of a negative control strain), which allows for simple and economical downstream processing, contributing to the affordability and accessibility of the process.
The harvested supernatant was concentrated using TFU ( Table 2). The microfiltration process returned satisfactory volumetric activity yields (GE: 86.84%, CDH:   Table 2), and the loss in activity units can be explained by the inherent loss of sample volume in the filtration apparatus itself. The total protein concentration in the enzyme extracts was increased 3.69-, 2.84-fold for GE and CDH, respectively. In terms of volumetric activity, GE and CDH were concentrated 4.95-, and 5.20-fold, respectively. The ultrafiltration process returned volumetric activity yields that were acceptable (49.57 and 50.00%), but lower than expected [26,[36][37][38] despite acceptable integrity of membrane (permeate with minimal volumetric activity 0.00-2.95 U Total L −1 ). It is therefore hypothesised that the recombinant protein is being retained or degraded in the filtration apparatus itself, as proteins can adsorb to membranes, often causing permanent fouling [39] and the formation of a protein monolayer on the surface of the membrane [40]. Despite these shortcomings, the final concentrated enzyme products could be used directly on lignocellulose according to enzyme dosages established in previous studies [2,41].

Limitations
This is the first report of GE secretion yields under constitutive expression, and the second report of bioreactor heterologous production [9]. This work demonstrates the feasibility of a constitutive, methanol-free P. pastoris expression system for production of emerging lignocellulose-degrading enzymes such as GE. The bioprocess reported here returned high protein yields of active GE and shows promise for different enzymes as well, including CDH (Additional file 1: Fig. S1). However, the constant specific productivity and the lack of an extended μ max phase during the fed-batch cultivation shows potential for improvement of secretion levels through manipulating the growth rate. TFU concentrated the enzyme extracts to suitable dosages for lignocellulose application studies but showed inefficient volumetric activity yields. This is significant due to the economic implications at manufacturing scale, therefore optimization of the ultrafiltration process is recommended. Future studies should also confirm the application of the concentrated recombinant enzymes on industrially relevant substrates.
Additional file 1: Figure S1. Illustration showing the strategy applied for the production of glucuronoyl esterase (GE) and cellobiose dehydrogenase (CDH) enzymes and the summary of main results.
Additional file 2: Figure S2. Exponential decline in specific growth rate during glycerol fed-batch stage of the bioreactor cultivation, with a constant glycerol feed of 72.6 mL/h. of (A): Glucuronoyl esterase; (B): Cellobiose dehydrogenase.