Bioprocess Optimisation for High Cell Density Endoinulinase Production from Recombinant Aspergillus niger

Endoinulinase gene was expressed in recombinant Aspergillus niger for selective and high-level expression using an exponential fed-batch fermentation. The effects of the growth rate (μ), glucose feed concentration, nitrogen concentration and fungal morphology on enzyme production were evaluated. A recombinant endoinulinase with a molecular weight of 66 kDa was secreted. Endoinulinase production was growth associated at μ> 0.04 h−1, which is characteristic of the constitutive gpd promoter used for the enzyme production. The highest volumetric activity (670 U/ml) was achieved at a growth rate of 93% of μmax (0.07 h−1), while enzyme activity (506 U/ml) and biomass substrate yield (0.043 gbiomassDW/gglucose) significantly decreased at low μ (0.04 h−1). Increasing the feed concentration resulted in high biomass concentrations and viscosity, which necessitated high agitation to enhance the mixing efficiency and oxygen. However, the high agitation and low DO levels (ca. 8% of saturation) led to pellet disruption and growth in dispersed morphology. Enzyme production profiles, product (Yp/s) and biomass (Yx/s) yield coefficients were not affected by feed concentration and morphological change. The gradual increase in the concentration of nitrogen sources showed that, a nitrogen limited culture was not suitable for endoinulinase production in recombinant A. niger. Moreover, the increase in enzyme volumetric activity was still directly related to an increase in biomass concentration. An increase in nitrogen concentration, from 3.8 to 12 g/L, resulted in volumetric activity increase from 393 to 670 U/ml, but the Yp/s (10053 U/gglucose) and Yx/s (0.049 gbiomasDWs/gglucose) did not significantly change. The data demonstrated the potential of recombinant A. niger and high cell density fermentation for the development of large-scale endoinulinase production system.


Introduction
Therefore, there is a need for the development and optimisation of bioprocess parameters for the use of A. niger in the production of recombinant endoinulinase using a cheaper and easily assimilable carbon source. The aim of the study was to develop and assess the potential of a recombinant A. niger strain for endoinulinase production with a glucose-limited fed-batch exponential fermentation strategy. A. niger D15 (uridine auxotrophic (pyrG), proteasedeficient (prtT), nonacidifying (phmA)) mutant was transformed with pGT (bla gpd P -glaA T ) vector containing Inu A gene encoding endoinulinase. The transcription was under the control of the glyceraldehyde-3-phosphate dehydrogenase promoter (gpd P ) of A. niger and glucoamylase terminator (glaA T ) of Aspergillus awamori. Enzyme secretion was directed by the native gene's secretion signal. The effects of the bioprocess parameters, glucose feed concentration, nitrogen source concentration and growth rate, on biomass growth and enzyme production in fed-batch culture, were investigated. The study aimed to gain insight on the factors influencing and challenges related to recombinant endoinulinase production from A. niger during high cell density fermentation.

Media and Cultivation Conditions
All chemicals were of analytical grade and unless stated otherwise, sourced from Merck (Darmstadt, Germany). The E. coli DH5α strains were cultivated at 37°C in Terrific Broth and on Luria Bertani agar containing 100 μg ampicillin/ml for selective pressure (20).
The A. niger D15 parental strain was cultivated at 30°C in minimal media containing 5 g/L yeast extract, 0.4 g/L MgSO 4 ·7H 2 O, 2 g/L casamino acids, 20 ml 50×AspA (300 g/l NaNO 3 , 26 g/L KCl, 76 g/L KH 2 PO 4 , pH 6), 0.01 M uridine and 1 ml/L 1000×trace elements (21). Transformants were selected for on minimal medium lacking uridine. Media were inoculated to a concentration of 1×10 6 spores per ml unless stated otherwise. The A. niger D15 transformants were initially cultivated in 20 ml double-strength minimal media (2×MM) containing 10% glucose for screening purposes (enzyme activity determination). Cultivation took place in 125-ml Erlenmeyer flasks on a rotary shaker at 200 rpm at 30°C for 3 days. Supernatants were obtained by centrifugation at 12 000g for 10 min at room temperature and stored at 4°C for further analysis.

DNA Manipulations and Gene Amplification by PCR
Standard protocols were followed for all DNA manipulations and E. coli transformations [20]. The A. niger ATCC10864 strain was cultivated in minimal media for 72 h. Mycelia were harvested, frozen under liquid nitrogen and the DNA isolated [21]. The InuA was amplified from the genome using the polymerase chain reaction (PCR) with oligonucleotide primers listed in Table 1. TaKaRa Extaq Polymerase (TaKaRa Bio Inc., Otso Japan) was used for amplification of the genes with the reaction set up in accordance with the supplier's specifications in the Perkin Elmer GeneAmp® PCR system 2400 (Perkin Elmer, USA). The InuA was cloned into the NotI site of plasmid pGT (21) to obtain pGT-InuA under the transcriptional control of the glyceraldehyde-3-phosphate dehydrogenase promoter (gpd P ) of A. niger and glucoamylase terminator (glaA T ) of A. awamori. The pyrG marker gene had been retrieved from pBS-pyrGamdS [22] via PCR and was cloned into the EcoRI site on plasmid pUC18, generating pUC-pyrG. Spheroplasts were prepared from the A. niger D15 (cspA1, pyrG1, prtT13, phmA, a non-acidifying mutant of AB1.13, ATCC 9029) strain using lyzing enzymes (Sigma-Aldrich, Steinheim, Germany) in accordance with [23]. The pGT and pGT-InuA vectors ( Fig. 1) were respectively co-transformed with pUC-pyrG to A. niger D15 to generate the A. niger D15[pGT-control] and D15[InuA] strains.

Sodium Dodecyl Sulphate Polyacrylamide Gel Electrophoresis
The proteins in the supernatant samples (20 μL) were separated using sodium dodecyl sulphate polyacrylamide gel electrophoresis (12% SDS-PAGE) as described by Sambrook et al. (1989). Electrophoresis was carried out at 100 V at ambient temperature and the proteins visualised using the silver staining method [24]. The broad-range Page Ruler Prestained SM0671 Protein Ladder (Fermentas, Shenzhen, China) was used as a molecular mass marker.

Pre-inoculum Preparation and Cultivation Medium for the Fermentations
Stock cultures of the strains were stored at −80°C in 30% (v/v) glycerol as cryoprotectant. A. niger spore production was performed in spore plates containing 18 g/L agar, 2 g/L peptone, 1 g/L yeast extract, 10 g/L glucose and 2 g/L casamino acids with nitrates, at 30°C. The densely conidiated culture was harvested with saline solution (9 g/L NaCl) after 5 days. The minimal medium (MM) without uridine [22] was used for the batch phase and was composed of 20 g/L glucose, 1 g/L casamino acids, 1 g/L peptone, 1 g/L yeast extract and 1.8 g/L MgSO4·7H 2 O. A pre-inoculum was prepared in a 1-L Erlenmeyer flask containing 400 ml of the medium by inoculating with a spore concentration of Aspergillus of 1 × 10 6 spores per 1 ml. The flask was incubated at 30°C on a rotary shaker (Innova 2300, New Brunswick Scientific, Edison, USA) at 180 rpm for 12 h.

Bioreactor Operating Conditions and Growth Medium
Enzyme production was carried out in batch cultures in 14 L BioFlo 110 bioreactors (New Brunswick Scientific company, Inc., USA) with a 10-L working volume and equipped with a polarographic DOT probe and a glass pH electrode (Mettler, Toledo, Sandton, South Africa). A batch culture with a total volume of 4 L of minimal medium (MM) without uridine [22] was used for the enzyme production. Batch culture fermentation were carried out with an initial total nitrogen concentration in the range of 3.8-18 g/L. The preinoculum was subsequently transferred directly from the flask into the bioreactor. The cultivation temperature and pH were 30°C and pH 5.5, respectively. The pH was maintained through the cultivation period using 25% w/v NH 4 OH. Furthermore, a constant aeration rate of 0.8 v/v/m was maintained in the bioreactor during the cultivation. The dissolved oxygen was maintained above 30% saturation through a control loop that linked the agitation to the dissolved oxygen tension. The agitation speed was cascaded between 250 and 400 rpm, with 400 rpm set as a maximum to limit biomass degradation [16]. Foaming was controlled by addition of 0.1% (v/v) of 30% antifoam (Sigma-Aldrich, Kempton Park, South Africa). The fermentation broth was sampled during the culture at specified intervals, vacuum filtered with a Buchner funnel and the enzyme activity in the broth determined before storage at 4°C. The biomass was subsequently washed with distilled water, air dried at room temperature and the dry weight determined. The cultures were performed in triplicates and mean values with the standard deviation of the biomass and activity reported.

Exponential Feeding
An exponential feeding strategy was employed to evaluate the effects of growth rate on biomass and enzyme concentrations. Feeding was initiated at the end of the batch phase following depletion of the carbon source (glucose). Residual glucose was tested with glucose test strips (Accu-Chek®). The prediction of biomass (X t ) produced at a time t, using an exponential growth Eq. 1 [25], enabled estimation of the mass of glucose (S t ), the primary growth-limiting nutrient, fed during the fed-batch phase. This subsequently enabled the regulation of a pre-determined growth rate, where the glucose concentration in the reactor is assumed to be zero during the feeding phase [10]. Furthermore, the amount of glucose (S t ) to maintain a specific cell biomass (X t ) was determined according to Eq. 2 [26,27]: where X t , X 0 and μ are the biomass at time t, biomass at the end of the batch culture phase and the growth rate, respectively.
where S t , V 0 , X 0 , μ set , S 0 and Y x/s are the mass of glucose at time t, volume of broth at the end of the batch culture, biomass at the end of batch culture phase, the desired growth rate, mass of glucose at time 0 and biomass substrate yield coefficient, respectively. The feeding rate was determined according to Eq. 3 where F t , F 0 and μ are the feed rate (g/h) at time t, feed rate at the end of the batch culture and growth rate, respectively.

Enzyme Assay
Endoinulinase activity was determined based on the method of [15]. The broth culture from the fermentation was filter sterilised for use in the enzyme assays. A 100 μL solution of 25% (w/v) inulin (Novozymes), 750 μL of 0.1 M sodium acetate buffer (pH 5) and 100 μL of the crude enzyme were mixed and incubated at 50°C for an hour. The reaction was terminated by placing in boiling water for 5 min. The solution was centrifuged and analysed for reducing sugars with the DNS assay. One unit of endoinulinase activity was defined as the amount of enzyme required to produce 1 μmol of reducing sugar per minute under assay conditions. Positive control assay was performed with a commercial inulinases (Megazyme). The A. niger D15 [pGT-control] was used as negative control strain.

Results and Discussion
Strain Development Figure 2 is an illustration of an SDS-PAGE for the selection of the positive transformant strains used in the fed-batch fermentation for endoinulinase (InuA) production. Endoinulinase with a molecular mass of ca. 66 kDa [28] was identified relative to the molecular mass marker. The presence of the enzyme was further confirmed by the absence of a corresponding band of 66 kDa from the negative control lane 4 ( Fig. 2). Therefore, the recombinant A. niger overexpressed and secreted the endoinulinase extracellularly.

Effects of Glucose Feed Concentration on DO Levels and Mixing as well as Subsequent Impact on Biomass and Enzyme Production
Biomass production, enzyme activity, DO and agitation speed profiles, during endoinulinase production at four different glucose feed concentrations, are demonstrated in Fig. 3. The maximum specific growth rate (μ max ) and biomass yield on substrate (Y x/s ) estimated in the batch culture prior to exponential feeding were 0.075 h −1 and 0.49 g biomassDW /g glucose (data not shown), respectively. Fed-batch fermentations were carried out at a constant exponential In contrast, fed-batch fermentation with a concentrated feed (100-300 g/L) resulted in a drastic change in the fungal morphology from pellet to mycelia form. The DO levels and agitation remained constant at ca. 30% of saturation (Fig. 3C) and 250 rpm (Fig. 3D), respectively, during the first 12 h in the stationary phase of the culture. However, these changed drastically (Fig. 3C, D, respectively) when the culture entered the exponential growth phase. The DO levels dropped rapidly reaching a low of 8% of saturation and remained at this level throughout the fermentation period (Fig. 3C), due to the exponential increase in biomass concentration (Fig. 3A) and apparent increase in oxygen demand from the increased biomass concentration [29][30][31]. Accordingly, the agitation speed increased, reaching the maximum setspeed limit of 400 rpm, because of the system attempting to maintain the DO setpoint of 30%  (Fig. 3D). Consequently, gradual pellet fragmentation to mycelial form was observed following the continued low DO levels of 8% of saturation (Fig. 3C) and agitation at 400 rpm (Fig. 3D). The experimental data illustrated the importance of bioprocess parameters on fungal morphology and the complex interdependence of the former in controlling the morphology [32]. Therefore, the culture conditions can be controlled to induce fungal growth in pellet morphology and subsequently minimise the viscosity limitations associated with fungal growth in mycelial form [14,16,33]. High cell densities are responsible for both increased broth viscosity [12,14] and low DO levels in the culture, due to rapid oxygen uptake [16]. The internal resistance, because of the former, results in inefficient mixing, oxygen transfer as well as nutrient diffusion, and subsequent pellet disintegration [31,32]. Changes in the culture conditions, associated with inefficient mixing, such as pH has been hypothesised to contribute to pellet disintegration due the impact on electrostatic forces that contribute to the pellet integrity [34]. Broth viscosity, at high cell densities, also increases the probability of collision and friction between the pellets which in turn weaken the hydrophobic and electrostatic interactions that keep the pellet structure intact [34,35]. Although increasing agitation is necessary to improve mixing efficiency, oxygen and nutrient diffusion, at high biomass concentrations [29,30], high agitation speeds cause shear forces that cause pellet disintegration and growth in dispersed morphology which further affect broth viscosity [14,31]. Sporh and co-workers [36] illustrated that agitation speeds of at least 400 rpm resulted in fungal pellet degradation.  (15,100,200 and 300 g/L) were evaluated at a growth rates close to μ max . The nitrogen-nutrient concentration was 3.8 g/L. Agitation speed was cascaded between 250 and 400 rpm. The feeding start and end point(s) are indicated by an arrow pointing upwards and downwards, respectively. The data is presented as mean ± standard deviation of triplicates runs The change in fungal morphology, however, did not affect both biomass growth and enzyme production, which contrasts with what has been reported for recombinant A. niger. In addition, the feed concentration did not have a significant impact on the biomass yield on substrate (Y x/s ) or enzyme yield on substrate (Y p/s ), and the specific enzyme productivity (Y p/s ) ( Table 2). However, increasing the feed concentration did significantly increase the volumetric productivity (Q p ) of the system. For instance, an increase from 15 to 100 g/L resulted in a Q p increase from 7980 to 9831 U/L/h. The control of bioprocess conditions to support a specific fungal morphology, in recombinant enzyme production using recombinant Aspergillus sp., may also contribute to productivity improvements [16,29]. However, the experimental data for the recombinant endoinulinase production in A. niger is contrary to what has been reported previously, regarding enzyme production under critical dissolved oxygen levels and different fungal morphology.
Lopez and co-workers [16] reported that Aspergillus terreus grew in large fluffy pellet morphology at DO levels of 80% saturation and agitation speeds less than 300 rpm and that agitation speeds > 300rpm resulted in growth in small pellets and a significant reduction in lovastatin productivity. Haack and co-workers [29] reported that A. oryzae growth and lipase production were inhibited by low oxygen availability as a result of increased biomass concentration, morphology change from pellets to mycelium, during the feeding phase of an exponential fed-batch culture. The insignificant impact of the fungal morphology on the endoinulinase productivity in recombinant A. niger could be attributed to the point of enzyme synthesis. Haack and co-workers reported that lipase production was localised from the hyphal tips of A. oryzae, and morphology change from pellets to dispersed morphology reduced the density of active hyphal tips thereby reducing lipase productivity.
In contrast, phytase production from Aspergillus ficuum [33] and fructofuranosidase production from A. niger [14] were enhanced by fungal growth in small pellets. Driouch and coworkers [12] further illustrated that there was no significant difference in biomass yields for pellets of different diameters. High DO levels were not a necessity for recombinant endoinulinase production that was growth-associated [12,14], and therefore productivity was not significantly different between pellet and hyphal growth despite the poor aeration and oxygen transfer of the latter.

Effects of Nutrient Concentration of Biomass Yield and Enzyme Production
Nitrogen sources are vital nutritional components for biomass growth and biosynthesis pathways [37,38]. Moreover, nutrient composition is an important factor in bioprocess optimisation and plays a critical role in maintaining optimal bioprocess conditions for the maximum productivity [32]. Therefore, fed-batch fermentation was performed to evaluate the effect of nitrogen concentration, on fungal morphology, biomass yields and enzyme productivity. The nutrients were comprised of a cocktail of three organic nitrogen sources, which were yeast extract, peptone and casamino acids in proportions of 50, 25 and 25%, respectively [22]. Exponential fed-batch cultures were carried out at a μ of 0.07 h −1 , glucose feed concentration of 300 g/L and organic nitrogen cocktail concentrations of 3.8, 12 and 18 g/L in the batch culture medium. An increase of approximately 3-fold in the nitrogen cocktail, from 3.8 to 12 g/L, resulted in significant (p<0.05) increase in the final biomass concentration (Fig. 4A) and volumetric enzyme activity (Fig. 4B), from 18.36 to 34.4 g/L and 393 to 670 U/L, respectively. A further increase of the nutrients to 18 g/L did not result in a further increase in the biomass concentration and enzyme activity (Fig. 4A, B, respectively). The data thereby demonstrated that an increase in the feed concentration should be supplemented with a corresponding amount of nitrogen to ensure the culture has excess nitrogen and remains carbon limited. Although the increment in the nutrient concentration resulted in an increase in the final biomass concentration and enzyme volumetric activity, the Y x/s , Y p/x and Y p/s did not significantly (p>0.05) differ (Table 3) at the different nutrient concentrations, demonstrating that the enzyme yield on biomass and carbon source (glucose) did not change. Therefore, the organic nitrogen sources did not have a direct induction effect on recombinant endoinulinase production, confirming that enzyme production was growth associated. The data demonstrated that at a feed concentration of at least 100 g/L glucose, biomass growth and enzyme production were nitrogen limited at a concentration of 3.8 g/L nitrogen cocktail and conversely carbon limited at an excess nitrogen concentration of 12 g/L. The supplementation of the culture media with organic nitrogen sources of up to 12 g/L was sufficient to prevent nitrogen limitation without impacting the fungal morphology of A. niger D15 [InuA] in a negative way. Fu and co-workers [39] reported that peptone had a positive impact on pellet formation. However, excess nitrogen components in the medium can result in fungal growth in mycelial form and consequently cause viscosity limitations.

Effect of Growth Rate on Biomass Growth and Enzyme Production as well as Yields and Productivities
Exponential fed-batch fermentations at 93% and 53% of μ max , μ of 0.07 h −1 and 0.04 h −1 , respectively, were performed to determine the effect of the growth rate on biomass growth and enzyme production, with a high glucose concentration feed (300 g/L) and excess nitrogen (18 g/L) supplementation. The agitation was cascaded between 250 and 400 rpm to minimise biomass degradation. Exponential feeding resumed after depletion of 20 g/L glucose in the batch culture and was terminated when residual glucose started to accumulate and DO levels spiked from the 8%, maintained during the exponential growth phase, to ca. 20% level of saturation (Fig. 5D). The highest biomass concentration was obtained at the higher growth rate of 0.07 h −1 and was equivalent to 33.94 g/L, compared to 29.03 g/L obtained at the lower growth rate of 0.04 h −1 (Fig. 5A). The highest volumetric activity (680 U/ml) reported in this study was significantly lower than reported in a P. pastoris, a yeast host, which reported a volumetric activity of 4000 U/ml [15]. The enzyme yield by this host was 36 U/g biomass in comparison to 20 U/g biomass reported for A. niger. The activities of the A. niger D15 [InuA] Fig. 4 Biomass concentration (A), enzyme activity (B) and DO (C) using different concentration of the nitrogen sources. A cocktail of three nitrogen sources were used that consisted of yeast extract, peptone and casamino acids in proportions of 50, 25 and 25%, respectively. An exponential feeding rate of μ = 0.07 was used with a fixed glucose feed of 300 g/L. Agitation speed was cascaded between 250 and 400 rpm. The feeding start and end point(s) are indicated by an arrow pointing upwards and downwards, respectively. The data is presented as the mean ± standard deviation of triplicates runs a,b,c Similar letter superscripts represents values that are not statistically significant based on analysis of variance at a 95% confidence level strain from this study was, however, higher than those reported for Y. lipolytica and S. cerevisiae. An endoinulinase volumetric and specific activity of 93.8 and 16.7 U/mg, respectively, were recorded for Y. lipolytica [15]. Zhang and co-workers (2010) reported an activity of 34.2 U/ml for S. cerevisiae. The plots for the natural log of the total biomass ( Fig.  5C) showed that an exponential growth rate was maintained during the fed-batch phase and that the measured growth rate deviated by approximately 5% and 10% from μ set for μ of 0.07  showing that a constant exponential growth rate was maintained during the fed-batch fermentations. The end of the fermentation was accompanied by glucose accumulation and subsequent cessation of biomass growth and enzyme production (Fig. 5A, B) and an increase in the DO levels to ca. 30% of saturation (Fig. 5D). In addition, visual inspection indicated that the culture was characterised by highly fragmented fungal mycelia 1 components. Fungal fragmentation was preceded by drastic drop in DO level to 8% (Fig. 5D), as a result of an exponential increase in biomass concentration (Fig. 5A), and a subsequent increase in agitation speed, to maximum speed of 400 rpm (data not shown), to improve the mixing efficiency. Fragmentation started with gradual pellet disruption into mycelia and eventually degradation of the mycelia, at the end of fermentation (T 54 and 66 h for μ of 0.07 h −1 and 0.04 h −1 , respectively), and this coincided with growth cessation. Furthermore, the transition from pellet morphology to dispersed morphology, during the exponential growth phase, did not impact endoinulinase production (Fig. 5B). Comparisons of the yields and productivities (Table 4) between enzyme production at high growth rates (0.07 h −1 ) and low growth rates (0.04 h −1 ) showed that the biomass and enzyme yield coefficients (Y x/s and Y p/s was 0.049 and 10068, respectively) were significantly (p<0.05) higher at the high growth rates (0.07 h −1 ). The substrate specific consumption (Q s = 0.101) was also significantly higher at high growth rates (0.07 h −1 ) compared to a Q s of 0.072 at μ of 0.07 h −1 and 0.04 h −1 . The lower enzyme productivities and biomass yield at low specific growth rate could be attributed to a high maintenance energy requirement at low growth rates [40,41]. Slow-growing biomass has been reported to have high metabolic energy requirements for maintenance of cellular integrity and viability [42] at the expense of biomass growth. The growth-associated nature of endoinulinase production at higher growth rates, where maintenance energy did not have a significant impact, was attributed to the constitutive gpd promoter controlling endoinulinase expression from recombinant A. niger [9,43].

Conclusions
Viscosity challenges, associated with high cell densities during the fed-batch phase, limited the ability of the system to achieve enzyme volumetric activity similar to yeast systems. Therefore, strategies that improve the specific productivity of the A. niger system will enable attainment of improved volumetric activity, considering the culture has limitations in the maximum attainable biomass concentration. The optimisation of nitrogen components of the nutrients and growth rate enhanced the productivity of recombinant endoinulinase production from A. niger. Moreover, recombinant endoinulinase production from A. niger was not strictly growth associated and only growth associated at high growth rates. The data demonstrated the potential of recombinant A. niger and high cell density fermentation for the development of largescale endoinulinase production using a glucose-inducible promoter.

Consent for Publication
We declare that the information in this manuscript has not been published elsewhere nor is it under consideration by any other journal. Furthermore, it is the consensus of all authors to submit this manuscript for possible publication in ABB Competing interests The authors declare no competing interests.