**An Innovative Biocatalyst for Continuous 2G Ethanol Production from Xylo-Oligomers by** *Saccharomyces cerevisiae* **through Simultaneous Hydrolysis, Isomerization, and Fermentation (SHIF)**

**Thais S. Milessi-Esteves 1,†, Felipe A.S. Corradini 2,†, Willian Kopp 1, Teresa C. Zangirolami 1,2, Paulo W. Tardioli 1,2, Roberto C. Giordano 1,2 and Raquel L.C. Giordano 1,2,\***


Received: 3 February 2019; Accepted: 24 February 2019; Published: 1 March 2019

**Abstract:** Many approaches have been considered aimed at ethanol production from the hemicellulosic fraction of biomass. However, the industrial implementation of this process has been hindered by some bottlenecks, one of the most important being the ease of contamination of the bioreactor by bacteria that metabolize xylose. This work focuses on overcoming this problem through the fermentation of xylulose (the xylose isomer) by native *Saccharomyces cerevisiae* using xylo-oligomers as substrate. A new concept of biocatalyst is proposed, containing xylanases and xylose isomerase (XI) covalently immobilized on chitosan, and co-encapsulated with industrial baker's yeast in Ca-alginate gel spherical particles. Xylo-oligomers are hydrolyzed, xylose is isomerized, and finally xylulose is fermented to ethanol, all taking place simultaneously, in a process called simultaneous hydrolysis, isomerization, and fermentation (SHIF). Among several tested xylanases, Multifect CX XL A03139 was selected to compose the biocatalyst bead. Influences of pH, Ca2+, and Mg2+ concentrations on the isomerization step were assessed. Experiments of SHIF using birchwood xylan resulted in an ethanol yield of 0.39 g/g, (76% of the theoretical), selectivity of 3.12 gethanol/gxylitol, and ethanol productivity of 0.26 g/L/h.

**Keywords:** 2G ethanol; hemicellulose usage; *S. cerevisiae*; enzyme immobilization; cell immobilization; SHIF

#### **1. Introduction**

Biofuels will have a significant role in the energetic matrix of the low-carbon economy, helping to meet the goals established at Conference of the Parties (COP 21) [1,2]. Among biofuels, bioethanol production from lignocellulosic materials has been intensively studied once it was shown that these byproducts had high availability, had a low cost, and did not compete with the production of food [3]. Lignocellulosic raw materials are mainly composed of cellulose and hemicellulose (up to 70%), which are polysaccharides that, after a hydrolysis step, generate fermentable sugars, mostly xylose from hemicellulose and glucose from cellulose [4]. The use of these two polysaccharides is important for the economic feasibility of the biofuel production process.

Some microorganisms that naturally ferment pentoses to ethanol have been tested for industrial use, such as *Scheffersomyces stipitis* and *Pachysolen tannophilus* [5,6]. However, these microorganisms have a low tolerance to ethanol and slow fermentation rates and are inhibited by compounds generated during the biomass pretreatment step, such as furfural [7].

*Saccharomyces cerevisiae* is the most common microorganism used for ethanol production from hexoses, due to its high rate of fermentation and superior ethanol yield. In addition, this yeast exhibits unbeatable tolerance to ethanol, to inhibitors, and to high concentrations of sugar [8,9]. However, in its wild form *S. cerevisiae* is unable to efficiently metabolize D-xylose.

The genetic modification of *S. cerevisiae* aimed at xylose fermentation has been extensively studied [8,10–12]. However, the low specific growth rate, high xylitol production, reduced yeast tolerance, and possible genetic instability are still hindrances for the application of recombinant strains on an industrial scale [7].

In spite of the inability of *S. cerevisiae* to metabolize xylose, it is capable of fermenting its isomer, xylulose, to ethanol. Hence, an alternative for the utilization of the hemicellulose fraction for bioethanol production would be to isomerize xylose to xylulose ex vivo, followed by fermentation by *S. cerevisiae* [7]. The enzyme xylose isomerase (XI) (EC 5.3.1.5) is widely used in the industry for the production of fructose syrup from corn starch and also catalyzes the reversible isomerization of xylose to xylulose [13]. Although the xylose/xylulose chemical equilibrium is unfavorable (3.5:1 at 60 ◦C) [14], the reaction can be displaced by the simultaneous isomerization and fermentation (SIF) process, where the continuous conversion of xylulose to ethanol might allow the complete depletion of the available xylose [15].

The use of catalysts with immobilized enzymes may be crucial for the application of multi-enzymatic processes on an industrial scale. This approach allows the continuous operation of the reactor and facilitates the product recovery as well as the use of high loads of cells and enzymes [16]. The production of an active and stable enzyme derivative using a non-expensive support is also an important issue in enzyme immobilization [17]. The literature reports successful applications of immobilized enzymes on an industrial scale [18,19] and immobilized XI is one of the most successful and established examples [13]. Silva et al. [7] developed a biocatalyst containing chitosan-immobilized XI, co-immobilized with *S. cerevisiae* in calcium alginate gel. Calcium alginate gel was chosen for being a natural polymer widely studied as a support for the immobilization of viable cells [20]. However, this system showed to be susceptible to contamination by xylose-consuming bacteria. High concentrations of xylose in the medium disfavored the *S. cerevisiae* population, due to its low uptake rates of xylulose.

An alternative to tackling the contamination problem is to use a cultivation medium containing non-readily fermentable substrates, such as xylo-oligomers obtained by the solubilization of hemicellulose under mild conditions [21]. The hetero-polysaccharides that compose hemicellulose are polymers with about 100 units of monomers, mainly xylose, and their solubility depends on the number of monomeric units in the chain [22]. Thus, the extraction of xylan in the form of large oligomers must be carried out under conditions that allow a sufficient number of glycosidic bonds to be broken, so that soluble polymers with lower molecular weight (xylo-oligomers) are released.

Xylanases (β-1,4-D-xylanase) are enzymes that catalyze the hydrolysis of the glycosidic bonds between xylose units. The enzymatic complex is commonly composed of endoxylanases, exoxylanases, β-D-xylosidases as well as accessory enzymes such as glucuronidase and arabinofuranosidase that act on the ramifications of the xylan chain [23]. The addition of these enzymes to the biocatalyst proposed by Silva et al. [7] would allow the feeding of xylo-oligomers to the bioreactor. This substrate might decrease the probability of contamination during the operation of the bioreactor for long periods, which are typical in industry. Preliminary results showed the technical viability of this process [24].

The development of viable processes to increase ethanol yields from lignocellulosic materials is crucial, despite the challenges that still remain for the production of 2G ethanol from xylose. Considering the higher production cost of 2G ethanol (compared to 1G ethanol), the use of pentoses as raw material could make its production more profitable and might overcome the costs of 2G ethanol extra steps [25]. The integration of several biocatalytic transformations in a multi-enzymatic cascade system is particularly appealing to the development of cleaner and more efficient biochemical processes. Multi-enzymatic cascade reactions offer advantages such as lower demand of time, reduced costs, easier recovery of products, completion of reversible reactions as well as concentrations of inhibitory compounds restrained to a minimum [26].

In this context, the simultaneous hydrolysis, isomerization, and fermentation (SHIF) process stands out for 2G ethanol production since unmodified *S. cerevisiae* remains the preferred microorganism in industry, due to its robustness, high ethanol tolerance, and production rates. The use of wild strains to produce ethanol from xylose is an important issue in countries like Brazil, where biosafety regulations are strict [15]. In addition, one advantage of this approach is that XI, along with amylases and proteases, is among the most widely and cheaply available commercial enzymes [27]. The present work reports the results of using this new biocatalyst for the simultaneous hydrolysis, isomerization, and fermentation of xylan derived from the hemicellulose fraction of biomass, aimed at the production of ethanol (Figure 1).

**Figure 1.** Xylan biomass simultaneous hydrolysis, isomerization, and fermentation (SHIF). Biocatalyst composed of xylanases, xylose isomerase, and co-immobilized *S. cerevisiae*.

#### **2. Results**

#### *2.1. Application of the New Biocatalyst in the SHIF Process*

The biocatalyst is designed for the industrial production of second-generation ethanol in continuous, fixed-bed reactors through long-term operation, by applying simultaneous hydrolysis, isomerization, and fermentation (SHIF) of the hemicellulosic fraction of biomass. First, xylanase and xylose isomerase were covalently immobilized on chitosan. The obtained XI derivative presented an activity of 252.5 ± 1.6 IU/g (immobilization yield of 93% and recovered activity of 91%), whereas the Accellerase derivative exhibited 346.3 ± 9.2 IU/g (immobilization yield of 54% and recovered activity of 12%). Both derivatives were co-encapsulated with *S. cerevisiae* in Ca-alginate gel and this biocatalyst was used to produce ethanol from commercial birchwood xylan.

Birchwood xylan, which is a heteropolymer composed of long chains, was first hydrolyzed to smaller xylo-oligomers by the action of recombinant endoxylanase of *Bacillus subtilis* (XynA) in order to increase the concentration of xylo-oligomers with smaller chains that may diffuse into the catalyst beads [28]. This step was carried out to make xylan more similar to lignocellulosic hydrolysates obtained from the pretreatment of biomass (data not shown). The composition of the substrate after xylan pre-hydrolysis is shown in Table 1.

As expected, there was no xylose production since XynA is a strict endoxylanase [28]. The solubilized fraction corresponded to 67% (w/w) of the offered xylan. According to Gray et al. [22], the solubility of the xylan oligomers depends on the degree of polymerization of each compound. Under the used conditions, 33% (w/w) of birchwood xylan is insoluble. Therefore, the substrate obtained for the SHIF process had 73 g/L of xylo-oligomers.


**Table 1.** Characterization of SHIF substrate: birchwood xylan, 108 g/L after 24 h hydrolysis by endoxylanase XynA (150 IU/gxylan) at 50 ◦C and pH 5.6.

For the SHIF assays, the offered enzyme activity in the reactor was 1.7 × 104 IU/Lreactor for xylanase (Accellerase XY) and 3.7 × <sup>10</sup><sup>4</sup> IU/Lreactor of xylose isomerase. Accellerase XY was used due to the presence of β-xylosidase, which is necessary to xylose formation. Commercial baker's yeast (Itaiquara®) concentration was 50 gdry mass/Lreactor at the beginning of the SHIF assays. Results in Figure 2 show the production of ethanol through SHIF using the developed biocatalyst. Ethanol production was higher compared to xylitol, presenting a selectivity of 2.61 (2.2 g/L ethanol and 0.84 g/L xylitol). Ethanol productivity of 0.092 g/L/h and yield (YP/S) of 0.160 gethanol/gpotentialxylose (32% of theoretical, calculated on the basis of potential xylose in the xylan) were achieved at the end of the SHIF run.

**Figure 2.** (**a**) SHIF of pre-hydrolyzed birchwood xylan using a biocatalyst containing 5 w % of Accellerase XY, 15 w % of XI, and 10 w % dry mass yeast (1.7 <sup>×</sup> 104 IU/L, 3.7 <sup>×</sup> 104 IU/L, and 50 g/L, respectively), pH 5.6, 150 rpm stirring, and 32 ◦C; (**b**) control experiment using xylose as substrate (SIF): pH (-), xylose (), ethanol (Δ), and xylitol (-).

Figure 2 shows that ethanol is produced from xylo-oligomers. The slower velocity of SHIF compared to the control experiment (where the substrate is xylose, resulting in a simultaneous isomerization, and fermentation (SIF) process) was expected since xylose concentration in SHIF depends on the velocity of hydrolysis of the xylo-oligomers. However, a decrease in the xylose consumption was observed after eight hours of SHIF (Figure 2a), which indicates that the xylose isomerization was impaired. There are two possible reasons for this occurrence: XI is not catalyzing isomerization of xylose in the required velocity; or the yeast is not consuming the generated xylulose, which would be accumulating and consequently stopping the isomerization due to the xylose:xylulose equilibrium ratio. Since the yeast remained viable (initial and final cell viability unchanged: 96%) and there was no accumulation of xylulose in the medium, the accumulation of xylose seemed to be related to the isomerization step.

In the control experiment, using only xylose as substrate, XI catalyzed the isomerization reaction effectively, almost until depletion of the pentose (60 g/L of xylose were consumed in 12 h, a productivity of 1.2 g/L/h, with YP/S of 0.303 g/g and selectivity of 2.3 with respect to xylitol). Thus, an increase of the isomerization velocity must be sought. Some of the possible causes for the hindrance of the isomerization step were then investigated. High calcium concentrations are known to inhibit XI action [29], and the hydrolysis reaction could be demanding a higher release of the calcium ion (increased hydrolysis of CaCO3 to control the pH). However, even with a higher release of calcium, the results shown in Figure 2 indicate that the hydrolysis step led to a drop of pH to approximately 5.1. XI shows maximum activity at pH 8, being highly sensitive to a drop of pH to this range [15]. On the other hand, it is known that the magnesium ion is an important cofactor for XI, as an activator of this enzyme [30]. The influence of pH and Ca2+ and Mg2+ ions in the isomerization step was then investigated.

#### *2.2. Influence of pH, Ca2+, and Mg2+ on XI Activity*

In order to investigate the influence of calcium and magnesium ions on XI, the activity of this enzyme to catalyze fructose–glucose isomerization was measured at different pHs (5.0 to 8.0) in the presence of different concentrations of Ca2+ and Mg2+ ions. The standard medium for assessing activity was 2 M fructose, pH 8.0 (50 mM tris-maleate buffer) supplemented with 50 mM of MgSO4 and 2.5 mM of CoCl2 at 60 ◦C. Both Co2+ and Mg2+ are essential for the activity of XI, however they play differentiated roles. Mg2+ is superior to Co2+ as an activator, while the latter is responsible for the stabilization of the enzyme and maintenance of its conformation, especially the quaternary structure [31]. Table 2 shows the measured activities of XI in each condition studied, referred to the test performed with the standard medium at pH 8.0 as 100%.

**Table 2.** Influence of Mg2+ and Ca2+ on XI activity at different pHs (isomerization of fructose 2M, 60 ◦C). The activity measured in standard medium (pH 8.0, 50 mM MgSO4, 2.5 mM CoCl2) was taken as 100%.


\* [CaCl2] = 4 g/L (same as used during SHIF supplementation).

The Ca2+ ion proved to be an inhibitor of this enzyme, since a significant decrease in the XI activity occurred when the Ca2+ concentration in the medium increased. The Mg2+ ion, in turn, was able to activate the enzyme, increasing its activity in 16.2% at pH 8.0 and 163% at pH 5.0, both in calcium-free medium. This cofactor was still able to bypass the inhibition caused by calcium, since it reactivated the enzyme in the presence of this ion, increasing its catalytic activity in all studied pHs.

Xylose isomerization catalyzed by XI is initiated by opening the sugar ring, followed by isomerization through the exchange of hydride and finally stabilization of the product by ring closure [32]. Although there is no relationship between the presence of magnesium and the ring opening step, this cation is essential for the isomerization [33]. According to Kasumi et al. [34], the reaction mechanism demands the formation of a binary divalent enzyme–cation complex, since the substrate will bind only to the active site of this complex. Xylose isomerase has two active sites, each containing two divalent cations [33]. Thus, the presence of higher concentrations of magnesium in the reaction medium, improving the probability of the presence of this ion in the active sites of the enzyme, would increase the rates of the isomerization reaction.

Although calcium is a divalent cation and belongs to the same family as magnesium (same configuration in the valence layer) in the periodic table, Ca2+ has a larger ionic radius than Mg2+. This fact could be the reason why Mg2+ is an activator of the enzyme while Ca2+ inhibits XI, that is, the difference in their atomic radii would cause a different interaction with the active site of the enzyme.

In addition to the significant influence of Ca2+ and Mg2+, XI showed great sensitivity to pH, losing activity significantly at pH 5.0. The sensitivity of XI to pH was previously observed by Milessi et al. [15], who emphasized the importance of pH control during the simultaneous isomerization and fermentation (SIF) of xylose. However, the activation provided by Mg2+ is potentiated at lower pHs. When Ca2+ (4 g/L) was added, the magnesium ion was able to recover XI activity more effectively at pH 5.0 than at pH 8.0 (Table 2, media 4 and 7).

Data presented in Table 2 also prove that XI activity was greatly reduced at the SHIF pH range (5.0 to 6.0). The quaternary structure of this enzyme is composed of four subunits that are delicately folded and associated with noncovalent links and without interchain disulfide bonds [31,32]. At pH 8.0, the enzyme structure is composed of all four subunits combined, resulting in the maximum catalytic activity. However, as the pH of the reaction medium lowers, the enzyme is more likely to suffer structure distortions, unfolding and dissociating its tetrameric structure. The presence of Ca2+ ions at the low pH of the reaction medium results in a combined effect, acting both on the 3D structure and on the active site. Therefore, the performance of the SHIF process will certainly benefit from a pH control system (pH 5.6). Unfortunately, the substitution of calcium chloride by magnesium chloride in the coagulation solution during sodium alginate gelation was not possible since the resulting beads were not stable. For this reason, SHIF experiments with pH control and excess of magnesium were run, in order to minimize the inhibition of XI caused by the hydrolysis reactions, which release acids from the structures of the xylo-oligomers [35].

#### *2.3. Xylanase Selection*

Xylan hydrolysis has to occur efficiently to enable the cascade SHIF process. Hydrolysis cannot be the rate-determining step of these reactions in series: due to the unfavorable equilibrium of xylose to xylulose isomerization, the supply of xylose must not control the reaction [7]. Considering that the composition of each xylanase complex influences the hydrolysis efficiency, different xylanases were evaluated with the purpose of selecting the most efficient for the SHIF process.

The xylanase family is strongly related to the profile of products generated in the process [36]. A xylanase capable of depolymerizing xylan into xylose efficiently is required to ensure that the SHIF process will proceed as expected. Thus, besides Accellerase XY A03304, used in previous SHIF tests, two additional xylanases were evaluated: recombinant *B. subtilis* endoxylanase (XynA) and Multifect CX XL A03139.Hydrolysis profiles and xylooligosaccharide (XOS) composition are reported in Table 3 and Figure 3, respectively.


**Table 3.** Composition of xylooligosaccharides (XOS) after enzymatic hydrolysis ofbirchwood xylan (25.4 g/L, 3.8 IU/mL, 24 h of reaction at 50 ◦C, pH 5.6).

**Figure 3.** Hydrolysis of birchwood xylan soluble fraction (25,4 g/L) by the studied xylanases (150 IU/gxylan = 3810 IU/L) at 50 ◦C, 24 h, pH 5.6. XynA (Δ); Multifect (-), and Accellerase (). Bars are standard errors of triplicates.

Figure 3 shows that Multifect stands out, with a xylan conversion of 78.7%. Moreover, the higher xylose concentration achieved with this enzyme at the end of the experiments indicates that this xylanase complex has a more stable β-xylosidase enzyme, responsible for catalyzing the hydrolysis of xylobiose, the essential final step for the complete xylan hydrolysis. Hence, this xylanase seems to be the most suitable for the production of xylose in the SHIF process, among the studied enzymes.

Accellerase has the highest enzymatic activity under standard conditions. However, in long-term reaction it was able to convert only 58.8% of the available xylan. Several factors may have contributed to Accellerase's inferior performance, such as the amount of each enzyme in the complex, thermal inactivation, substrate affinity, and inhibitory effects.

XynA was already known to have a strictly endoxylanase action, lacking β-xylosidase activity and consequently not producing xylose when hydrolyzing xylan [28]. Accordingly, it presented the lowest conversion (44.8%), probably due to the absence of debranching enzymes.

None of the tested xylanases reached 100% of xylan conversion. Indeed, the incapacity of xylanases to completely hydrolyze xylan has been previously reported. Akpinar et al. [37] observed a yield of 13.8% for tobacco xylan using *Aspergillus niger* xylanase (200 IU/g) at 50 ◦C after 24 h. Aragon et al. [38] achieved 13% of conversion in the hydrolysis of birch xylan (18 g/L) using *Aspergillus versicolor* endoxylanase immobilized on agarose-glyoxyl at 25 ◦C and pH 5.0. In fact, since xylan is not a linear polymer of pure xylose, its complete depolymerization requires the use of a varied pool of enzymes [21,23,39]. In this context, the xylanase Multifect CX XL A03139 was selected to be co-immobilized with XI and *S. cerevisia*e in the SHIF process.

#### *2.4. SHIF Assay with pH Control and Excess of Mg2+*

In order to overcome the possible inhibition of Ca2+ in XI activity, a SHIF assay was performed with pH control and excess of magnesium. Beads without CaCO3 in its composition were prepared, since this salt is only necessary to sustain the pH at the desired range. Moreover, an isomerization free of CaCO3 would contribute to reduce the undesired Ca2+ effects. After all these modifications, the obtained derivative of xylanase Multifect presented 330.2 ± 8.1 IU/g (immobilization yield of 96% and recovered activity of 77%). For the SHIF experiment, the medium supplemented with 100 mM MgSO4 (24.6 g/L) and 4 g/L of CaCl2 (to maintain the integrity of the beads) was added together with the beads to the pH-stat stirred reactor. It is important to note that the moderate agitation used during the process did not affect the integrity of the beads. According to Rahim et al. [40], damages to Ca-alginate immobilized biocatalysts due to stirring are usually observed above 200 rpm. In fact, Carvalho et al. [41], in experiments carried out at 300 rpm, noticed a 30% reduction in the size of Ca-alginate beads during experiments with immobilized *Candida guilliermondii*. Accordingly, in the

present work, an agitation of 150 rpm was employed and the structural characteristics of the biocatalyst beads were preserved.

The obtained results, showed in Figure 4, indicated a higher ethanol production in SHIF using pH control and excess of Mg2+ (3.1 g/L of ethanol) compared to the value of 2.2 g/L, which was achieved under the original SHIF conditions (Figure 2). Ethanol yield (0.39 g/g, 76% of the theoretical), selectivity (3.12), and productivity (0.26 g/L/h) were also improved.

**Figure 4.** SHIF of previously hydrolyzed birchwood xylan with pH control and excess of Mg2+ using immobilized biocatalyst containing5w% Multifect xylanase derivative, 15 w % XI derivative, and 10 w % yeast (1.7 x 10<sup>4</sup> IU/L, 3.7 x 104 IU/L, 50 g/L), initial pH 5.6, 150 rpm stirring and 32 ◦C; pH (-), xylose (), ethanol (Δ), and xylitol (-).

To the best of our knowledge, these are the highest yield and productivity reported in the literature for ethanol production from xylan through the simultaneous hydrolysis, isomerization, and fermentation (SHIF) process. Only a few works have studied ethanol production from pentoses using ex vivo isomerization and native *S. cerevisiae*, due to the differences in optimal pH and temperature ranges for each step (and none of them with co-immobilized enzymes/cells). The inclusion of the xylan hydrolysis step should not be a problem in relation to the temperature and pH of the process. Xylanases have the highest catalytic pH and temperature at approximately 5.5 and 50 ◦C, respectively, whereas XI optimal conditions are pH 7.0–8.0 and 70 ◦C [7]. Alcoholic fermentation, on the other hand, operates at pH 5.0 and 30 ◦C. Due to these facts, the process integration for ethanol production is still a challenge. Nakata et al. [42] studied ethanol production of hot-compressed water pretreated Japanese beech using soluble β-xylosidase, XI, and *S. cerevisiae*. It should be stressed that in this work the absence of exo- and endoxylanases would be a restraint to the saccharification step. The best results reported (0.62 g/L of ethanol, corresponding to 13% of theoretical yield) were achieved at pH 5.0, 30 ◦C after 100 h. Hence, the immobilized biocatalyst containing enzymes and yeast co-encapsulated reported in the present work was significantly more efficient than using enzymes and microorganisms in their soluble form, leading to a better yield and productivity.

Although there are only a few studies addressing the SHIF of xylan, the simultaneous isomerization and fermentation (SIF) of xylose has been more frequently reported. Rao et al. [27] studied the xylose SIF in the presence of 0.05 M borax to shift equilibrium concentration of xylulose/xylose and improve the isomerization step. However, although the isomerization was enhanced, only half of the available xylose was consumed. Lastick et al. [43] observed an ethanol titer of 2.1% (w/v) from the SIF of 6% xylose using XI and *Schizosaccharomyces pombe* (Y-164). Silva et al. [7] studied the SIF of 65 g/L of xylose at 30 ◦C, using a biocatalyst containing 32.5 x 103 IU/L of xylose isomerase and 20 g/L of yeast co-immobilized in Ca-alginate gel, and reported an ethanol productivity of 0.25 g/L/h. However, the isomerization step became a limiting factor, due to the decrease of the pH from 5.3 (initial) to 4.8 (final). Milessi et al. [15] incorporated CaCO3 into SIF beads to control the pH

of the process. The biocatalysts were prepared with 20% chitosan-immobilized XI and 10% fresh yeast. An ethanol yield of 0.35 g/g (70% of the theoretical yield) and 2 g/L/h productivity was observed. However, the long time needed by *S. cerevisiae* to ferment xylulose makes the SIF process susceptible to contamination by bacteria capable of metabolizing the xylose.

In this context, the proposed SHIF process appears as a promising approach for 2G ethanol production from hemicellulose. Process conditions and enzyme loads in the biocatalyst can be optimized to achieve higher yields and productivity as well as to overcome the difference between optimal pH ranges for each step of the process. Despite the improvement achieved after pH control and supplementation with excess of Mg2+, a small accumulation of xylose was still observed, which suggests that the isomerization step may be still limiting the process. There are other factors that might be affecting the isomerization and/or the fermentation steps, such as the presence of xylooligosaccharides (2–8 xylose units) or other intermediate products released during the xylan hydrolysis, which are not present in the SIF process. The understanding of the influence of these compounds as well as the optimization of the biocatalyst composition regarding the balance of the enzyme pool are important issues to be addressed in order to improve ethanol production rates.

In general, the SHIF process using co-immobilized enzymes and cells stands out for 2G ethanol production. Besides presenting the advantages of multi-enzymatic cascade reactions, it also enables an easy recovery of the biocatalyst, which could be applied in continuous or repeated batch ethanol production runs using a medium that inhibits contamination. In addition, it builds on the advantage of using the same native yeast, already employed in 1G ethanol industry, which simplifies the operation of the fermentation unit. The low genetic stability of recombinant microorganisms together with the strict Brazilian biosafety regulations for genetically modified organisms (GMO) application in the industrial environment make the 2G ethanol production process based on a native yeast an attractive alternative.

#### **3. Materials and Methods**

#### *3.1. Materials*

GENSWEET® SGI (3400 IU/mL, 127 mgprotein/mL, DuPont™ Genencor®, Palo Alto, CA, USA), an enzymatic extract of commercial xylose isomerase (EC 5.3.1.5) from *Streptomyces rubiginosus*, and the commercial enzyme preparations Accellerase XY (3670 IU/mL, 9.8 mgprotein/mL) and Multifect CX XL A03139 (785 IU/mL, 35 mgprotein/mL) were kindly donated by DuPont™ Genencor® (Palo Alto, CA, USA). The *Bacillus subtilis* recombinant endoxylanase (502 IU/mL, 9.2 mgprotein/mL) was donated by Verdartis (Ribeirão Preto, SP, Brazil). Powdered chitosan (85.2% deacetylation degree) was supplied by Polymar Ind. Ltda (Fortaleza, CE, Brazil) and 25% glutaraldehyde solution was purchased from Vetec (Duque de Caxias, RJ, Brazil). Birchwood and beechwood xylans were from Sigma-Aldrich (St. Louis, MO, USA). A *Saccharomyces cerevisiae* industrial strain (purchased from Itaiquara®, Tapiratiba, SP, Brazil) was used in all SHIF experiments. All other reagents were of analytical grade.

#### *3.2. Biocatalyst Production*

#### 3.2.1. Preparation of Chitosan-Glutaraldehyde Beads

Chitosan gel (2% or 4%, w/v) was prepared as described by Budriene et al. [44], using the coagulation of the chitosan-acetic acid solution in 0.5 M KOH. Activation of the support was carried out by the addition of glutaraldehyde (5%, v/v) in a suspension of chitosan at pH 7.0 (100 mM phosphate buffer, 1:10 msupport/vsuspension). After 60 min stirring at 25 ◦C, the support was filtered under vacuum, washed first with distilled water until neutrality, and then with ultrapure water.

#### 3.2.2. Enzyme Immobilization

Xylose isomerase immobilization was carried out onto 2% (w/v) chitosan-glutaraldahyde according to Silva et al. [7]. The enzyme solution was prepared in 50 mM Tris-maleate buffer (pH 7.0) containing 5 mM MgSO4.7H2O and 2.5 mM CoCl2.6H2O, in order to provide 50 mgprotein/gsupport. The support was added to the enzyme solution at a ratio of 1:10 (w/v). After 20 h of immobilization at 25 ◦C under 150 rpm stirring, sodium borohydride was added (1 mg/mL) and the suspension was kept under gentle agitation for 30 min in an ice bath [45]. The derivatives were filtered and washed under vacuum, first with 200 mM Tris-maleate buffer (pH 7.0), then with ultrapure water, and finally with 50 mM Tris-maleate buffer (pH 7.0), in order to remove borohydride and adsorbed enzyme.

Xylanase complexes were immobilized onto 4% (w/v) chitosan-glutaraldehyde according to Milessi et al. [28]. The immobilizations were performed at pH 7.0 (100 mM phosphate buffer), 25 ◦C, and under constant stirring. A load of 20 mgprotein/gsupport was offered, maintaining 1:10 (w/v) ratio of mass of support to volume of enzymatic solution. After the completion of immobilization, sodium borohydride was added (1 mg/mL) and the reduction reaction proceeded for 30 min at 4 ◦C. The derivatives were extensively washed with 50 mM citrate buffer pH 4.8 and stored until use.

#### 3.2.3. Biocatalyst Co-immobilization

The biocatalyst preparation was carried out through Ca-alginate gel entrapment according to the methodology described in Silva et al. [7]. The industrial strain, supplied as freshly compressed yeast cells, was used as purchased, without previous propagation or activation [46]. A solution of sodium alginate (1% w/v), immobilized XI (15% w/v), immobilized xylanase (5% w/v), and fresh yeast (10% w/v) was gently dropped into a 0.25 M CaCl2/0.25 M MgCl2 solution. Spherical particles (Ø = 1–1.5 mm) were produced using a pneumatic extruder [47]. The procedure was carried out in a laminar flow chamber (Airstream, ESCO, Horsham, PA, USA) and the sodium alginate and coagulation solutions were previously sterilized at 121 ◦C for 20 min. After immobilization, the beads were cured in a refrigerator for 12–16 h in cure solution (4 g/L of MgSO4, 10 g/L of KH2PO4, 3 g/L of urea, 0.2 g/L of CoCl2, and 4 g/L of CaCl2.2H2O).

#### *3.3. Simultaneous Hydrolysis, Isomerization, and Fermentation (SHIF) of Birchwood Xylan*

First, the birchwood xylan substrate was hydrolyzed in smaller xylo-oligomers by the action of recombinant endoxylanase of *B. subtilis* (XynA) in order to make xylan more similar to lignocellulosic hydrolysates, since after a pretreatment step the xylo-oligomers present in the medium are smaller than those in commercial xylan [28]. Xylan pre-hydrolysis was carried out by XynA (150 IU/gxylan, 3.8 IU/mL) immobilized on chitosan-glutaraldehyde (35.0 ± 0.8 IU/g) for 24 h and 50 ◦C under 150 rpm stirring. At the end, the immobilized enzyme was recovered by filtration and the dried xylan mass retained on the filter was quantified (xylan insoluble fraction). The pH of the xylan solubilized fraction (SHIF substrate) was adjusted to 5.6 with HCl or NaOH 1M and the medium was sterilized by filtration through a 0.22 μm membrane. SHIF experiments were carried out in a shaker incubator (32 ◦C and 150 rpm), using sealed tubes with a total reaction volume of 2.4 mL (bead ratio of 1:1, 1.2 g of beads and 1.2 mL of medium, bead density was 1 g/cm3). The composition of SHIF medium was 108 g/L of birchwood xylan supplemented with MgSO4 (4 g/L), KH2PO4 (10 g/L), urea (3 g/L), CoCl2.6H2O (0.2 g/L), and CaCl2.2H2O (4 g/L). Samples were collected at regular intervals for determination of pH, substrate consumption, and product formation.

#### *3.4. Influence of pH, Ca2+, and Mg2+ on XI Activity*

In order to study the influence of Ca2+ and Mg2+ ions on the catalytic activity of XI, the enzyme activity was measured at different pHs (5.0 to 8.0) and different ion concentrations. Nine medium compositions were tested for each studied pH. The standard medium was constituted of 2 M fructose at pH 8.0 (50 mM Tris-maleate buffer), 50 mM MgSO4, and 2.5 mM CoCl2.

#### *3.5. Enzymatic Xylan Hydrolysis for Xylanase Selection*

Xylan hydrolysis was carried out using the soluble fraction of the birchwood xylan obtained by adding 8 g of commercial birchwood xylan in 100 mL of 50 mM citrate buffer, pH 5.5 at 50 ◦C. After 1 h at 150 rpm stirring, the solution was centrifuged for 20 min at 9500× *g* and 5 ◦C. The supernatant was then recovered for further use at the concentration of 25 g/L of xylan. It was offered 150 IU/gxylan (3.8 IU/mLreactor). The reaction was conducted at 50 ◦C under mechanical stirring for 24 h.

#### *3.6. Analytical Methods*

#### 3.6.1. Xylanase Activity

Xylanase activity was determined according to International Union of Pure and Applied Chemistry (IUPAC) [48] by calculating the initial velocity of xylan hydrolysis catalyzed by a known amount of enzyme. The standard substrate was birchwood xylan (1% w/v) in 50 mM citrate buffer pH 5.5. Enzyme was added to the reaction medium and incubated at 50 ◦C for 10 min under 250 rpm stirring. Aliquots were withdrawn at 2 min intervals, and the released reducing sugars were quantified by the dinitrosalicylic (DNS) acid method [49]. One unit of activity (IU) was defined as the amount of enzyme required to release 1 μmol of xylose per minute under the assayed conditions.

#### 3.6.2. Xylose Isomerase Activity

Xylose isomerase activity was determined according to Giordano et al. [13], by measuring the initial velocity of fructose isomerization to glucose, under the following conditions: 2 M fructose solution prepared in 50 mM Tris-maleate buffer containing 50 mM MgSO4.7H2O and 2.5 mM CoC12.6H2O, at pH 7.0 and 60 ◦C. The glucose concentration was determined colorimetrically using the commercial enzyme kit containing glucose oxidase and peroxidase (GOD-PAP®, Bioclin, Belo Horizonte, Mg, Brazil). One international unit (IU) of xylose isomerase was defined as the amount of enzyme that released 1 μmol of glucose per minute under the assayed conditions.

#### 3.6.3. Substrate and Product Quantification

The concentrations of XOs, xylose, xylulose, xylitol, and ethanol were determined by high performance liquid chromatography (HPLC), equipped with a Waters Sugar-Pak I column (Milford, MA, USA) (300 × 6.5 mm) coupled to a refractive index detector (W410 Waters) (Milford, MA, USA). Ultrapure water was used as eluent at a flow rate of 0.5 mL/min. The column temperature was 80 ◦C, the detector was set at 40 ◦C, and the injected volume was 20 μL. Before the analysis, the samples were filtered using a 0.22 μm filter.

#### **4. Conclusions**

A new biocatalyst composed of co-immobilized xylanases, xylose isomerase, and unmodified *S. cerevisiae* was able to produce 2G ethanol from birchwood xylan in a one-spot multi-enzymatic simultaneous hydrolysis, isomerization, and fermentation (SHIF) process. Although more studies are required to increase ethanol productivity, the SHIF process showed to be promising from the point of view of its technical viability. The SHIF process brings advantages to industrial applications for easing the integration of 1G and 2G ethanol production processes (because both are based on the same native yeast strains), while reducing the contamination risk due to the use of xylo-oligomers as substrate instead of xylose.

**Author Contributions:** T.S.M. and F.A.S.C. performed the experiments, analyzed the data, and wrote the paper; W.K. performed the experiments; P.W.T and T.C.Z. analyzed the data and revised the paper; R.C.G. and R.L.C.G. supervised the work and reviewed the final manuscript.

**Funding:** This research was funded by the São Paulo Research Foundation (FAPESP), grants #2008/56246-0 and #2016/10636-8, the Brazilian National Council for Scientific and Technological Development (CNPq) grant #140982/2013-2, and in part by the Coordenação de Aperfeiçoamento de Pessoal de Nível Superior - Brazil (CAPES), Finance Code 001.

**Acknowledgments:** The authors thank DuPontTM Genencor® (USA) for the donation of xylose isomerase, Accellerase, and Multifect xylanases; and Verdartis for the donation of xylanase from *B. subtilis*.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Article* **Highly Selective Oxidation of 5-Hydroxymethylfurfural to 5-Hydroxymethyl-2-Furancarboxylic Acid by a Robust Whole-Cell Biocatalyst**

#### **Ran Cang 1,**†**, Li-Qun Shen 1,**†**, Guang Yang 1, Zhi-Dong Zhang 3, He Huang 1,2,\* and Zhi-Gang Zhang 1,\***


Received: 16 May 2019; Accepted: 11 June 2019; Published: 12 June 2019

**Abstract:** Value-added utilization of biomass-derived 5-hydroxymethylfurfural (HMF) to produce useful derivatives is of great interest. In this work, extremely radiation resistant *Deinococcus wulumuqiensis* R12 was explored for the first time as a new robust biocatalyst for selective oxidation of HMF to 5-hydroxymethylfuroic acid (HMFCA). Its resting cells exhibited excellent catalytic performance in a broad range ofpH and temperature values, and extremely high tolerance to HMF and the HMFCA product. An excellent yield of HMFCA (up to 90%) was achieved when the substrate concentration was set to 300 mM under the optimized reaction conditions. In addition, 511 mM of product was obtained within 20 h by employing a fed-batch strategy, affording a productivity of 44 g/L per day. Of significant synthetic interest was the finding that the *D. wulumuqiensis* R12 cells were able to catalyze the selective oxidation of other structurally diverse aldehydes to their corresponding acids with good yield and high selectivity, indicating broad substrate scope and potential widespread applications in biotechnology and organic chemistry.

**Keywords:** biocatalysis; extremophile; 5-hydroxymethylfurfural; 5-hydroxymethylfuroic acid; platform chemicals; whole cells

#### **1. Introduction**

The production of bio-fuels and chemicals from carbon-neutral and renewable biomass is attracting increasing interest [1–5]. Biomass is regarded as a sustainable resource from which some platform chemicals can be manufactured [6,7]. 5-hydroxymethylfurfural (HMF), derived from lignocellulosic materials via dehydration of carbohydrates, is one of the most important platform chemicals [8–10]. It has been listed as one of "Top 10+4" bio-based chemicals by the U.S. Department of Energy (DOE) [11], being applied in the synthesis of a variety of value-added pharmaceutical and biomaterial intermediates [12]. Due to its high reactivity, HMF is a versatile molecule that can be converted into various useful furan derivatives [12–14]. Its structure comprises a furan ring, an aldehyde group and a hydroxymethyl group which can be subjected to upgrading processes by selective redox reactions, leading to 5-hydroxymethylfuroic acid (HMFCA), 2,5-diformylfuran (DFF), 5-formylfuroic acid (FFCA), 2,5-furandicarboxylic acid (FDCA), maleic anhydride (MA) and 2,5-bis

(hydroxymethyl) furan (BHMF) (Scheme 1). Among these HMF derivatives, the completely oxidized product FDCA displays very promising application potential and may serve as a "greener" substitute for terephthalate in the manufacture of polyester and polyamide materials [15,16]. HMFCA is the oxidation product of the aldehyde group in HMF and a promising starting material for the synthesis of various polyesters [17]. It was reported that HMFCA can also be used as an antitumor agent and interleukin inhibitor [18,19].

**Scheme 1.** Catalytic biotransformation of 5-hydroxymethylfurfural (HMF) into high value derivatives.

In order to form HMFCA, selective oxidation of the aldehyde group in HMF is required, while the alcohol group is left intact. Chemoselective oxidation methods are mainly used in the synthesis of HMFCA from HMF, in which noble metal catalysts are generally used [12,20–22]. Recently, HMF was selectively oxidized to HMFCA by an immobilized molybdenum complex in toluene within 3 h, with a yield of approximately 87% [23]. Han et al. reported a selective and mild photocatalytic method for HMFCA synthesis from HMF under ultraviolet and visible light conditions with a yield of 90–95% [24]. In addition, the conversion of HMF to HMFCA via the Cannizzaro reaction is of great value [25,26]. However, the maximal selectivity of HMFCA was 50% due to the formation of an equimolar by-product.

Biocatalytic oxidation of HMF to HMFCA represents a promising alternative to chemical methods [14,27]. Biocatalysis offers many advantages, such as mild, environmentally friendly reaction conditions and often excellent selectivity, as well as high efficiency. However, compared to chemical methods, there are only a few reports on biotransformation of HMF to selectively form HMFCA in the literature [28–32]. In seminal work, Sheldon et al. reported the chloroperoxidase-catalyzed oxidation of HMF affording HMFCA with a selectivity of 25–40% [31]. Krystof et al. reported lipase-mediated and peracid-assisted oxidation of the HMF process to produce HMFCA [32]. Recently, Li and co-workers made use of a molybdenum-dependent enzyme—xanthine oxidase from *Escherichia coli*—for the biocatalytic oxidation of HMF to form HMFCA, with 94% yield and 99% selectivity [29].

Relative to the use of isolated enzymes, we believe that, in HMF oxidation, whole-cell biocatalysts have advantages. They are not only inexpensive and relatively stable, but they also do not require cofactor regeneration [27,33]. Biocatalysis is more efficient when recombinant whole cells that overexpress the enzyme(s) important for catalysis are used [34]. However, employing whole-cell biocatalysts for HMF oxidation is still challenging due to the well-known toxicity of HMF to microbial cells [30]. In addition, due to the variety of enzymes in microbial cells many side reactions are likely to occur during the process of HMF oxidation with formation of HMFCA [28]. Hence, exploring highly tolerant and selective microbial strains is crucial for the biotransformation of HMF into value-added derivative. To our knowledge, there are only a few studies on whole-cell-catalyzed selective oxidation of HMF to form HMFCA in the literature [28,30]; processes that are accompanied by a certain amount of HMF derivatives as byproducts. It was reported that some *Pseudomonas* strains have an HMF degradation pathway, in which HMF is converted to HMFCA as an intermediate [35–37]. A careful

literature search did not reveal any studies describing the use of this system for the production of HMFCA. In 2010, Koopman et al. reported the production of 2,5-furandicarboxylic acid (FDCA) from HMF by using recombinant *P. putida* S12\_hmfH. As part of this biotransformation, HMFCA hardly accumulated, leading to a mixture of other metabolites [38,39]. Therefore, in the challenging quest to obtain large amounts of pure HMFCA, the use of the *Pseudomonas* strain metabolic pathway is not feasible. Moreover, long standing issues still exist, such as low substrate loading, substrate toxicity and insufficient selectivity, etc. Therefore, searching for new and robust biocatalytic systems with high selectivity is a demanding task.

Extremophiles are organisms that have evolved to thrive under one or more extreme adverse environmental conditions where other organisms cannot survive [40,41]. They are regarded as an ideal and valuable source of biocatalysts, allowing biotransformation under relatively harsh industrial conditions [42–44]. Nevertheless, employing whole-cells or isolated enzymes derived from extremophiles for biocatalysis in a general manner is just beginning to be implemented experimentally. Recently, a *Deinococcus sp*, designated as *Deinococcus wulumuqiensis* R12, was isolated from radiation-polluted soil [45,46]. Previous studies showed that it is phylogenetically more closely related to a prototype strain of the *Deinococcus* genus, namely *Deinococcus radiodurans* R1 [47]. It was found that this strain was capable of producing carotenoids with good yield, and related biosynthesis genes were subsequently cloned and heterogeneously expressed in *E. coli.* by Xu et al. [48]. Furthermore, its whole genome was sequenced by Huang et al. [49]. Recently, genes encoding heat shock proteins from *D. wulumuqiensis* R12 were introduced into *Clostridium acetobutylicum* ATCC824 in order to improve the robustness and butanol titers of host cells [50]. Considering the robustness of *D. wulumuqiensis* R12, it would be of great interest to explore the catalytic properties of its whole cells in biotransformation or bioconversion.

In this study, we report that the radiation resistant strain *D. wulumuqiensis* R12 that can indeed be used as a whole-cell biocatalyst in HMFCA synthesis by selective oxidation of HMF (Scheme 2). The catalytic properties of this strain were evaluated in the transformation of HMF, and the reaction conditions were optimized. In addition, the substrate scope of this new whole-cell biocatalyst was also investigated.

$$\begin{array}{ccccc} \mathsf{O} & \mathsf{O} & \mathsf{O} \\ \mathsf{H} & \mathsf{O} & \mathsf{O} \\ \mathsf{H} & \mathsf{O} & \mathsf{O} \\ \end{array} \quad \begin{array}{ccccc} \mathsf{O} & \mathsf{O} & \mathsf{Set} \\ \mathsf{H} & \mathsf{C} & \mathsf{S} \\ \end{array} \quad \begin{array}{ccccc} \mathsf{O}\_{\mathsf{O}} & \mathsf{O}\_{\mathsf{O}} & \mathsf{S} \\ \mathsf{O} & \mathsf{O} & \mathsf{O} \\ \end{array} \quad \begin{array}{ccccc} \mathsf{O}\_{\mathsf{O}} & \mathsf{O} & \mathsf{Set} \\ \mathsf{O} & \mathsf{O} & \mathsf{Set} \\ \end{array}$$

**Scheme 2.** Chemoselective oxidation of MF to 5-hydroxymethylfuroic acid (HMFCA) with whole *D. wulumuqiensis* R12 cells.

#### **2. Results and Discussion**

#### *2.1. Growing and Resting Deinococcus Cells as Catalysts in HMF Oxidation with Selective Formation of HMFCA*

Similar to the prototype strain of the *Deinococcus* genus, *D. radiodurans* R1, *D. wulumuqiensis* R12 is also well known for its excellent ability to resist extremely high doses of gamma and UV radiation [45]. In order to explore its potential applications in biocatalysis, growing and resting cells of this strain were applied as biocatalysts in the conversion of HMF to form HMFCA. As shown in Figure 1a, 100 mM of the HMF substrate were converted almost completely within 12 h using resting cells, whereas growing cells gave only a 32% yield at a prolonged reaction time of 36 h. Resting cells enabled a much higher yield with more than 98% of HMFCA and a trace amount of 2,5-bis (hydroxymethyl) furan (BHMF) as sole byproduct, indicating excellent chemoselectivity in this biocatalytic process. Increasing substrate concentration further did not affect the selectivity of the resting cells (Figure S2). Considering the reported degradation mechanism of HMF in microbial cells [51], it is reasonable to speculate that the intermediate HMF alcohol (from HMF reduction) was almost completely oxidized in a very

short time to form the final HMFCA, or the HMF substrate was oxidized directly—which constitutes a different mechanistic hypothesis. However, to validate this inference, more efforts need to be invested.

**Figure 1.** The influence of various factors on HMFCA synthesis by HMF. General conditions unless otherwise stated: 100 mM HMF, 0.12 g/mL microbial cells, 100 mM,pH 7.4, phosphate buffer, 850 rpm, 30 ◦C, 4h reaction time. (**a**) Resting (solid symbols) and growing cells (open symbols), wherein tryptone glucose yeast extract (TGY) culture was used when growing cells were employed under 200 rpm, 30 ◦C; (**b**) cell concentration; (**c**) temperature; and (**d**)pH values. Time courses of HMF biotransformation using resting (solid symbols) and growing cells (open symbols).

In addition, we were pleased to discover the performance of two other *Deinococcus* strains stored in our lab, *D. radiodurans* R1 and *Deinococcus xibeiensis* R13, which were also used in the biocatalytic oxidation of HMF to HMFCA. It was found that both radiation resistant strains selectively oxidized HMF with formation of HMFCA. The conversions achieved by *D. radiodurans* R1 and *D. xibeiensis* R13 were slightly lower than that of *D. wulumuqiensis* R12 under the same reaction conditions (Figure S3). These results suggest that the *D. wulumuqiensis* R12 cells act as a catalytic system with high activity and excellent chemoselectivity in the oxidation of HMF to HMFCA. Its catalytic properties were subsequently investigated in greater detail (Figure 1).

#### *2.2. E*ff*ect of Cell Dosage in the Reaction System for HMFCA Synthesis*

Figure 1b shows the influence of microbial cell dosage on HMFCA synthesis based on selective oxidation of HMF. The yield of HMFCA increased steadily from 18% to 99% with increasing cell dosage in the presence of 100 mM of HMF substrate. The maximal yield of 99% was achieved when the cell dosage reached 0.12 g/mL, and further increasing did not improve the HMFCA yield, indicating that the biocatalyst was potentially saturated by substrate under the given reaction conditions. Our results imply that the conversion of HMF to HMFCA correlates with the cell dosage employed in the biocatalytic system. A small amount of cell dosage was required to reach maximal conversion

when the substrate concentration decreased to 40 mM under the same reaction conditions (Figure S4). In addition, cell dosage had no significant effect on the selectivity of the reactions (>98%). A higher cell dosage may result in higher viscosity, however, which could impact mass transfer of the reaction mixture. Thus, the optimal cell dosage of 0.12 g/mL wet cells was used in subsequent experiments.

#### *2.3. E*ff*ect of Temperature and pH on HMFCA Synthesis*

The influence of temperature and pH on HMFCA synthesis in the whole-cell catalyzed oxidation of HMF was also studied. As shown in Figure 1c, the effect of reaction temperature on HMF selective oxidation was determined by performing the transformation at different temperatures. Remarkably, the microbial cell biocatalyst showed considerable activity at a broad temperature range, from 25 to 60 ◦C. The maximal substrate conversion of 79% was obtained at 35 ◦C after 4 h in the presence of 100 mM HMF substrate. In addition, even at 50 ◦C, 67% of the HMF substrate was converted to HMFCA, which is in accord with an early report that *D. wulumuqiensis* R12 has a broad growth temperature range [45]. Slightly decreased conversion is possibly due to the inactivation of the enzymes in the microbial cells at 60 ◦C. It should be mentioned that HMFCA was obtained as essentially the only oxidative product—with a yield of 99%—in the reaction within the temperature range of 25 ◦C to 60 ◦C, indicating excellent catalytic selectivity of the whole-cell biocatalyst. Considering the thermostability of cells and energy efficiency, a temperature of 35 ◦C was set for subsequent experiments.

In addition, we further studied the pH profile of the whole-cell catalyst in HMFCA synthesis via selective oxidation of HMF (Figure 1d). It was found that the microbial cells had a broadpH activity profile and exhibited a particularly good catalytic performance in the pH range of 5.0 to 10.0. The best yield of 81% was achieved in 100 mM phosphate buffer at a pH 7.0 after 4 h. Interestingly, a conversion percentage of 60% and 58% was obtained in phosphate buffer ofpH 5.0 and Gly-NaOH buffer ofpH 10.0, respectively, after a reaction time of 4 h. In addition, it appeared that the buffer types had a moderate influence on the conversion of HMF, as a yield of 54% was obtained in Tris-HCl buffer (pH 9.0), compared to 63% in Gly-NaOH buffer at the samepH . One should not be surprised that *D. wulumuqiensis* R12 cells are able to resist such harsh reaction conditions with extremepH values. In their studies, Wang et al. reported that the *D. wulumuqiensis* R12 strain is able to grow in a widepH range from 5.0 to 12.0 [45]. Compared to *Comamonas testosterone* SC1588, which has been applied in HMFCA synthesis from HMF [28], *D. wulumuqiensis* R12 cells showed higher tolerance to extremepH values. Therefore, the optimalpH value of 7.0 was selected for all subsequent experiments.

#### *2.4. Inhibitory and Toxic E*ff*ect of Substrate*

HMF is a well-known toxic inhibitor of microbial cells, inhibiting their growth and hindering their upgrading of HMF by whole-cell biocatalysis [52]. The catalytic performance of *D. wulumuqiensis* R12 cells towards HMF under varying concentrations was therefore tested. As shown in Figure 2a, HMFCA was synthesized in 99% yield within 12 h when the substrate concentration was 150 mM. However, the yield decreased slightly to 81% at the substrate concentration of 200 mM. The yield decreased gradually in the substrate concentration range of 250–1000 mM. Remarkably, 41% yield of HMFCA was obtained when the substrate concentration reached 500 mM, and a yield of 23% was observed at the substrate concentration of 1000 mM. The essentially complete selectivity remained almost constant at these varying substrate concentrations.

**Figure 2.** Effects of HMF concentration on (**a**) HMFCA synthesis and (**b**) cell viability. Reaction conditions: 0.12 g/mL microbial cells, in phosphate buffer (100 mM,pH 7.0) under 850 rpm and 35 ◦C. Fresh harvested cells incubated without HMF/HMFCA under identical conditions were used as a control. Reaction periods: 12 h for 100 mM 150 mM; 24 h for 200 mM, 250 mM; and 36 h for 300 mM, 350 m, 400 mM, 500 mM, 1000 mM. (**c**) Effects of the product concentration on cell viability. HMFCA was incubated for 24 h under the same conditions.

It is well known that the cytotoxicity of HMF to microbial cells is a key parameter in whole-cell biocatalytic conversion of HMF. As shown in Figure 2b, the effect of substrate concentration on cell viability was investigated by using a cell viability assay. The microbial cells were incubated with varying concentrations of HMF under the usual reaction conditions, and the cell viability was subsequently measured using an Annexin V-FITC/PI Apoptosis Detection Kit. Microbial cells incubated in buffer without substrate under the same reaction condition were used as a control. Good cell viability (94%) was unexpectedly obtained in the presence of 400 mM substrate, which was not significantly different than that of the control. This result suggests that the microbial cells can tolerate as much as 400 mM of HMF substrate without losing viability. Further increasing HMF concentration to 500 mM led only to a slight decrease in cell viability to 83%. When 1000 mM of HMF substrate was used, 73% cell viability was still observed, indicating that *D. wulumuqiensis* R12 is extremely tolerant to HMF substrate. However, the conversion of HMF in the oxidation reaction decreased dramatically with increased substrate concentration. Since detailed toxic mechanisms of HMF to microbial cells are not completely understood, nor the reason for the extreme resistance of *D. wulumuqiensis* R12 cells to harsh environmental factors, the present results are not considered surprising.

#### *2.5. Inhibitory and Toxic E*ff*ect of the HMFCA Product*

In the oxidation of HMF to HMFCA, the product is an acidic compound harboring a carboxylic and hydroxyl group. In our work, the pH of the reaction system decreased over the reaction time due to the accumulation of the HMFCA product. This situation could conceivably become critical when high concentrations of substrate are produced. Thus, it would be of great interest and importance to investigate the possible inhibition and toxicity of the product towards microbial cells. Microbial cells incubated in phosphate buffer without the product was used as the control for such investigations. Based on the high cell viability (>93%), which was comparable to that of the control, there appeared to be no significant toxicity towards the microbial cells at HMFCA product concentrations less than 400 mM, as shown in Figure 2c. However, a product amount of 500 mM resulted in slight toxicity, as the cell viability value decreased to 87%. A further increased product concentration of 1000 mM led to significant toxicity of the microbial cells, with a viability of 84%. To our surprise, the data showed that the product toxicity to the viability of microbial cells was not as high as expected, even at extremely high concentrations. Considering that *D. wulumuqiensis* R12 is a robust strain, able to grow in a broad range ofpH values, one should not be surprised that the microbial cells are highly resistant to HMFCA production, with excellent cell viability at extremely high product concentrations. Further product toxicity tests with still higher concentrations were not performed, as the conversion of HMF was already very low at the substrate concentration of 1000 mM.

#### *2.6. Manufacture of HMFCA Under Optimized Conditions*

Obtaining large amounts of HFMCA is highly desired in biocatalytic oxidation of HMF by whole-cell biocatalysis, with great potential applications in industrial production. Therefore, further enhancement of the catalytic performance of *D. wulumuqiensis* R12 was investigated by optimizing the biocatalytic parameters of the conversion process. It was found that the HMFCA product yields were affected significantly by increasing the substrate concentration. For example, due to the known negative effect of HMF, the yield of HMFCA decreased significantly when HMF concentrations were higher than 300 mM (Figure 2a). Based on the catalytic properties of this strain, increasing the dosage of microbial cells in the reaction may further enhance the HMFCA yield. Recently, Zhang et al. reported that improved synthesis of HFMCA from HMF was obtained by tuning the pH of the reaction mixture using NaOH solution during the catalytic process [28]. Thus, we decided to employ the same strategy. In addition, it has been reported that adding furfural and furfural alcohol as inducers during cultivation of microbial cells for biocatalysis can trigger the expression of the enzymes responsible for HMF oxidation, which can facilitate HMFCA production. Therefore, three strategies (increasing microbial cell dosage, using inducing cells and tuningpH of reaction mixture) were applied in subsequent studies.

Increasing the dosage of microbial cells proved to be effective for enhancing the yield of HMFCA (Figure S4a). For example, in the presence of 300 mM HMF, the yield of HMFCA increased considerably from 59% to 71% when the concentration of cells increased from 0.12 g/mL to 0.2 g/mL. A further increase in cells dosage was not performed considering the negative effect of higher viscosity in the reaction mixture. On the other hand, influencing the expression of cells by the use of furfural and furfural alcohol for enhancing the yield of HMFCA proved not to be effective (Figure S4b). The reason for this might be that enzymes in *D. wulumuqiensis* R12 responsible for HMF oxidation are expressed constitutively. Finally,pH tuning was found to be an effective method for improving the yield of HMFCA (Figure S4c). Thus, the pH of the reaction mixture was tuned to approximately 7.0 using a NaOH solution. Compared with the control withoutpH tuning, the HMFCA yield improved from 66% to 83% in the presence of 300 mM HMF substrate, and from 48% to 65% at an HMF concentration of 500 mM.

Therefore, both increasing the dosage of microbial cells and pH tuning was applied together for enhancing the production of HMFCA. As shown in Figure 3, at a high HMFCA concentration, a yield of 90% was achieved after 36 h when the substrate concentration was 300 mM. This demonstrates the considerable effectiveness of the combined strategy. In addition, the oxidative conversion of HMF to HMFCA reached 80% in the presence of 350 mM HMF after 48 h. When the concentration of HMF was set to 500 mM, 66% of the substrate was still converted after 48 h, but further prolonged reaction times did not lead to an increase in HMFCA yield.

**Figure 3.** Synthesis of HMFCA under optimized conditions. Reaction conditions: HMF of the designated concentration, 0.2 g/mL microbial cells, 5 mL phosphate buffer (100 mM,pH 7.0), 35 ◦C, and 850 rpm. Tuning the pH of the reaction system to approximately 7.0 occurred every 3 h in the first 12 h, and then every 12 h in the 36 h that followed.

Compared with previous reported biocatalysis results, the data obtained in this work proved to be more efficient and selective because of a higher substrate concentration and simpler catalytic process. As shown in Table 1, the substrate concentrations used in the reported biocatalytic routes were still very low and co-enzymes were usually required when isolated enzymes were applied. Although *C. testosterone* SC1588 cells also display good selectivity and high HMF tolerance, its catalytic performance is highly sensitive topH [28]. Thus, a considerable amount of histidine co-substrate is required for efficient selective oxidation of HMF. The extreme environment-derived *D. wulumuqiensis* R12 strain used in this work exhibited excellent resistance to highpH and temperatures, and proved to be a robust biocatalyst for HMFCA synthesis by way of selective oxidation of HMF.


**Table 1.** HMFCA synthesis via HMF oxidation by various biocatalytic systems.

#### *2.7. E*ffi*cient Synthesis of HMFCA by a Fed-Batch Strategy*

As mentioned above, excellent yields of HMFCA from selective oxidation of HMF were obtained under optimized conditions. It is highly desirable to manufacture HMFCA on a large scale in an effort to create a practical biocatalytic process. Thus, by applying a fed-batch strategy, in which HMF substrate was added continuously, the accumulation of high concentrations of product was achieved. Figure 4 shows the results of biocatalytic synthesis of high concentrations of HMFCA. It was found that 511 mM of product was produced within 20 h after three-batch feeding of HMF, affording a total yield of 85% and a productivity of approximately 44 g/L per day. Only 4 mM of BHMF was observed as the sole byproduct (<1%). Chemoselectivity towards the target product reached more than 99%. In addition, a decrease in yield of HMFCA in each batch feeding was observed, indicating possible substrate and/or product inhibition in the whole-cell biocatalyst. An attempt to improve the yield of HMFCA further in this fed-batch process was performed by prolonging the reaction time, but no significant improvement was observed (Data not shown).

**Figure 4.** Synthesis of HMFCA by a fed-batch method. Reaction conditions: 150 mM HMF, 0.2 g/mL microbial cells, 5 mL phosphate buffer (100 mM,pH 7.0), 35 ◦C, and 850 rpm. In each cycle of 5 h, 0.75 mmoL of HMF was added.

#### *2.8. Exploring the Substrate Scope of D. wulumuqiensis R12*

In order to examine the substrate spectrum of this novel whole-cell biocatalyst, a set of structurally unique aldehyde compounds was applied in the oxidation reactions catalyzed by *D. wulumuqiensis* R12 cells (Scheme 3). Considering possible solubility and toxicity effects, proper concentrations of these aldehyde compounds were applied. The results showed that the whole cells of the strain readily accept furfural as a substrate, thereby enabling an efficient synthesis of furoic acid (>99% yield). Furoic acid can be used in the pharmaceutical, agrochemical, and flavor industries [9]. In the case of aldehydes containing an additional hydroxyl group, the microbial cells proved to be strictly chemoselective for the aldehyde group, affording the corresponding carboxylic acids with good to excellent yields (Table 2). For example, oxidation of DFF and FFCA to FDCA was achieved with 100% and 63% yields, respectively. Due to solubility issues, higher DFF substrate concentrations were not applied. We also discovered that the aldehyde group of vanillin could be selectively oxidized by the microbial cells to vanillic acid. Vanillic acid has important applications in the pharmaceutical industry, and also as a monomer in polyester synthesis [54]. In addition, terephthalic acid and p-hydroxybenzoic acid were also prepared by selective oxidation of the corresponding aldehydes. In short, the results showed that *D. wulumuqiensis* R12 as a biocatalyst has an amazingly broad substrate spectrum.

**Scheme 3.** Aldehyde substrates used and products obtained in biocatalytic oxidation by *D. wulumuqiensis* R12 whole cells.


**Table 2.** Whole-cell biocatalytic oxidation of aldehydes.

Reaction conditions: 0.12 g/mL microbial cells, 5 mL of phosphate buffer (100 mM,pH 7.0), 850 rpm, 35 ◦C for 12 h.

#### **3. Materials and Methods**

#### *3.1. Chemicals and Strains*

Extremely radiation resistant strains *D. wulumuqiensis* R12 (DSM 28115T), *D. radiodurans* R1 (ATCC NO.13939), and *D. xibeiensis* R13 (NRBC 105666T) were acquired from Zhi-Dong Zhang at the Institute of Microbiology, Xinjiang Academy of Agricultural Sciences in China and stored in our laboratory. The GenBank accession number for the 16S rDNA sequence was KJ784486, while the whole genome sequence was APCS01000000.

HMF (97%) and HMFCA (97%) were purchased from Macklin Biochemical Co., Ltd (Shanghai, China). FDCA (98%), DFF (98%) and furfural (98%) were obtained from Aladdin Biochemical Technology Co., Ltd (Shanghai, China). BHMF (98%) was purchased from Ark Pharm, Inc (Arlington Heights, IL, USA). HMFCA (97%) was obtained from J&K Scientific Ltd (Beijing, China). Both furfuryl alcohol (99.5%) and furoic acid (98%) were obtained from TCI (Shanghai, China). Annexin V-FITC/PI Apoptosis Detection Kit was purchased from Nanjing KeyGen Biotech. Co. Ltd. (Jiangsu, China) for the cell viability assay.

#### *3.2. Cultivation of D. wulumuqiensis R12 Cells*

The glycerol stock of *D. wulumuqiensis* R12 was pre-cultivated at 30 ◦C, 200 rpm for 24 h in tryptone glucose yeast extract (TGY) medium containing 0.5% tryptone, 0.1% glucose, and 0.3% yeast extract. Then, 1% of the overnight preculture was transferred to fresh TGY medium. The culture was incubated at 30 ◦C, 200 rpm for 48 h and was centrifugated under 5000 rpm for 10 min to harvest cells. The cell pellet was washed twice with 100 mM phosphate buffer (pH 7.4) and resuspended in phosphate buffer with a final cell concentration of 0.12 g/mL (cell wet weight).

#### *3.3. General Procedure for the Biocatalytic Oxidation of Aldehyde Substrates*

Five milliliters of phosphate buffer (0.1 M,pH 7.0) containing the designated amounts of microbial cells (cell wet weight) and substrates was incubated at 35 ◦C and 850 rpm for a given reaction time. Aliquots of the reaction mixture were withdrawn at specified reaction times and diluted with the phosphate buffer prior to high-performance liquid chromatography (HPLC) assays. The conversion of HMF and other aldehydes by biocatalytic oxidation was defined as the percentage of the consumed substrate amount in the initial amount. The selectivity of the reaction was defined as the ratio of HMFCA product amount to the sum of all the products. The yield was defined as the percentage of the measured product amount in the theoretical product amount based on the initial amount of HMF.

$$\% \text{ yield } = \frac{\text{Actual yield}}{\text{Theoretical yield}} \times 100\%$$

#### *3.4. Analytical Method*

The reaction products were analyzed by HPLC following a previously reported method with slight modifications [28]. Briefly, a reverse-phase HPLC (Thermo Fisher ultimate 3000), equipped with Sepax GP-C18 column (4.6 mm × 250 mm, 5 μm), was used at 25 ◦C. The mobile phase was the gradient of acetonitrile in 20 mM KH2PO4 (pH 6.0) at a flow of 1.0 mL min<sup>−</sup>1, increasing from 10% to 24% within 7 min and from 24% to 10% within 3 min. The HPLC retention time of the HMFCA product and HMF were 2.90 min and 6.20 min, respectively. All experiments were performed in triplicate and mean values are presented. Data are expressed as the mean ± standard deviation. Duncan's multiple range test (using SPSS software 16.0, Chicago, IL, USA) was used to analyze the statistical significance of differences between the groups. A significance difference was judged to exist at a level of *p* < 0.05. HPLC runs are shown in Figure S1.

#### *3.5. Cell Viability Assay*

Cell viability assay experiments were performed using an Annexin V-FITC/PI Apoptosis Detection Kit and flow cytometry following the manufacturers' instructions. Cell viability was determined using ACEA NovoCyte Flow Cytometer (ACEA Biosciences, Inc., San Diego, CA, USA) with the excitation light and emission light wavelengths set at 488 nm and 530 nm, respectively. Data were collected and analyzed using NovoExpress software. The cell viability of *D. wulumuqiensis* R12 when using HMF as the substrate is presented as the percentage of living cells to the total amount of cells.

#### *3.6. Synthesis of HMFCA by the Substrate Fed-Batch Feeding Process*

Five milliliters of 100 mM phosphate buffer,pH 7.0, which contained 150 mM HMF substrate and 0.2 g/mL of microbial cells, was incubated at 35 ◦C and 850 rpm. After 5 h, 0.75 mmol of HMF was repeatedly added to the reaction mixture. During the whole biocatalytic process, the pH of the reaction mixture was adjusted to the range of 7.0–8.0 with NaOH solution and the concentration of substrate and products was analyzed by HPLC.

#### **4. Conclusions**

Herein, we successfully explored for the first time the use of a radiation resistant *D. wulumuqiensis* R12 strain as a whole-cell biocatalyst for the efficient synthesis of HMFCA from HMF. The whole cells of this strain proved to be highly tolerant to HMF and the product, HMFCA. The whole-cell system is an excellent biocatalyst for the selective oxidation of HMF. An excellent yield of HMFCA of up to 90% was achieved within 36 h in the presence of 300 mM HMF substrate under optimized conditions. A yield of 80% to 66% was obtained when the substrate concentration increased from 350 mM to 500 mM, while the selectivity towards HMFCA remained at approximately 98%. In addition, up to 511 mM of HMFCA was synthesized in 20 h via a fed-batch method, resulting in a productivity of 44 g/L per day. Thus, *D. wulumuqiensis* R12 cells are a promising catalyst in the biocatalytic process of HMF upgrading. Moreover, the cells were able to transform a set of structurally different aldehyde compounds into their corresponding carboxylic acids with good to excellent selectivity. Since the genome sequence of this strain has been sequenced, exploring the genes that encode the enzymes responsible for HMFCA synthesis from HMF has become feasible in future work. The catalytic properties of these microbial cells can also be further engineered by introduction of other oxidases to form a cell factory for HMF biotransformation. Furthermore, this strain may also have potential applications for the biodetoxification of lignocellulosic hydrolysates in the process of biofuel production. Discovery of *D. wulumuqiensis* R12 as an efficient biocatalyst broadens the toolbox of biocatalysts for the biotransformation of HMF into value-added derivatives and will further facilitate the utilization of biomass for the production of useful chemicals and biofuels.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2073-4344/9/6/526/s1, Figure S1: HPLC measurements: (**A**) HMF and its derivatives; (**B**) HMF substrate, HMFCA and BHMF products. Figure S2: Effects of reaction time on HMFCA synthesis. Figure S3: Biocatalytic oxidation of HMF to HMFCA by different *Deinococcus* strains. Figure S4: Effect of cell concentration on HMFCA synthesis. Figure S5: Optimizing reaction condition for HMFCA synthesis by using various strategies.

**Author Contributions:** Conceptualization, Z.-G.Z.; investigation, R.C., G.Y. and L.-Q.S.; funding acquisition, Z.-G.Z.; resources, Z.-G.Z.; writing—original draft preparation, Z.-G.Z.; writing—review and editing, H.H. and Z.-D.Z.

**Funding:** This research was funded by the National Natural Science Foundation of China (No. 21646014, 21776134 and 21776136), the program of Jiangsu Synergetic Innovation Center for Advanced Bio-Manufacture (XTE1851). and The APC was funded by the program of Jiangsu Synergetic Innovation Center for Advanced Bio-Manufacture (XTE1851).

**Acknowledgments:** We thank Professor Dr. Manfred T. Reetz (Max-Planck-Institut für Kohlenforschung, Germany) for critical reading of the manuscript and helpful comments.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

#### *Article*
