Brewer’s spent grain biotransformation to produce lignocellulolytic enzymes and polyhydroxyalkanoates in a two-stage valorization scheme

Lignocellulolytic enzymes from low-cost sources are gaining attention as a tool to reduce production costs. Such enzymes can be obtained sustainably by diverse fungal strains via solid-state fermentation (SSF) of lignocellulosic-derived residues as substrates. Besides, these enzymes allow hydrolyzing the same residue, releasing fermentable sugars that can be transformed into value-added products. This study shows a two-stage valorization approach for the lignocellulosic leftover brewer’s spent grain (BSG): first, by producing lignocellulolytic enzymes through the SSF of BSG using three fungal strains and, second, by using the self-produced enzymes to hydrolyze the same BSG and obtaining sugar-rich hydrolysates that serve as an alternative carbon source for polyhydroxyalkanoates (PHA) production. From the evaluated set, Aspergillus niger and Thermoascus aurantiacus produced the highest xylanase activities compared with Trichoderma reesei (268 ± 24, 241 ± 10, and 150 ± 24 U per gram of dry BSG, respectively). Also, A. niger extracts resulted in the most effective for releasing sugars from BSG, obtaining up to 0.56 g per gram of dry BSG after 24 h without any pretreatment needed. Thus, the sugar-rich hydrolysate obtained with A. niger was used as a source for producing PHA by using two bacterial strains, namely, Burkholderia cepacia and Cupriavidus necator. Maximum PHA yield was achieved by using C. necator after 48 h with 9.0 ± 0.44 mg PHA·g−1 dry BSG. These results show the significant potential of BSG as raw material for obtaining value-added bioproducts and the importance of multiple valorization schemes to improve the feasibility of similar residue-based systems.


Introduction
Commercial lignocellulolytic enzymes are mainly produced through submerged fermentation (SmF) systems. Typically, such enzymatic solutions are obtained starting from high-purity substrates implying a significant increase in the total processing costs [1]. In this context, the need for more economical and sustainable substrates, along with the development of more efficient processes for producing these enzymes, has gained attention recently. One of the most attractive bio-based materials with potential for being a source of enzymes is the lignocellulosic materials [2]. Lignocellulosic-derived wastes are abundant carbon sources produced in many industrial processes, reaching up to 10 11 tons of biomass leftovers per year globally [2]. Although most of the lignocellulosicderived wastes are either used as an energy source (by burning) or directly disposed into the environment, their characteristics make them potential raw materials for the development of different biotechnological applications [3]. In general, these wastes are attractive due to their renewable nature, abundance, and low cost, constituting one of the most studied materials in the biorefinery sector for producing value-added products [3,4].
However, the biodegradability of lignocellulosic biomass depends on the particular composition of its organic components: cellulose (9%-80%), hemicellulose (10%-50%), and lignin (5%-35%) [5,6]. Hence, to efficiently process lignocellulosic materials, a hydrolysis step is usually needed. Hydrolysis allows transforming cellulose and hemicellulose into fermentable sugars that can be used to produce value-added products [3,5]. Although this process can be carried out by using chemical hydrolysis or other physicochemical treatments, the current trend is to use lignocellulolytic enzymes such as cellulases and xylanases [7]. In the hydrolysis process, enzymes act as catalysts accelerating various reaction steps. However, differently than their chemical counterparts, enzymatic processes tend to work under mild conditions making them more attractive from diverse points of view [6,8].
Based on this context, different authors have proposed the solid-state fermentation (SSF) as a potential alternative technology to SmF for producing lignocellulolytic enzymes more sustainably and economically [9,10]. Compared with SmF, SSF usually achieves higher productivities and yields, it requires lower water and energy consumption, and it limits the production of waste streams [11,12]. For instance, Castilho et al. [13] have analyzed both, SSF and SmF, on the annual production of 100 m 3 of Penicillium restrictum lipase concentrate economically. The study concluded that SmF requires a total capital investment about 78% higher than the SSF process, and the unitary product obtained from the SSF process was more economical than the actual selling price (47% lower). Such an outcome suggests that the economic differences between the two approaches are mainly influenced by the raw material cost. Nevertheless, one of the most interesting traits of SSF resides in the ability to process a wide range of solid organic residues as raw materials to obtain an extensive set of marketable bioproducts [12]. Thus, apart from industrial enzymes, SSF has been successfully used to produce other value-added products such as biosurfactants [14], base chemicals [15], or biopesticides [16], starting from solid organic wastes, showing that SSF is a versatile and fruitful technology.
On the other hand, the brewery industry is one of the most productive industries of beverages in the world [17], and it generates large quantities of brewer's spent grain (BSG), a solid leftover obtained after the fermentation process which is primarily used as animal feed [18]. This lignocellulosicderived waste has significant amounts of proteins and fibers that could be used as a source for producing value-added products [19].
Although SSF has been previously used to obtain different lignocellulolytic enzymes using lignocellulosic materials [9,20], just a few reports show BSG as raw material to obtain fermentable sugars [21][22][23] from those enzymes. Furthermore, many of the reports focus on the enzyme production, but not on the application of such enzymes to obtain other value-added products. In this sense, it is evident that the enzymes are not final products per se, but intermediates required to produce other marketable products [24]. Thus, determining both the efficiency of the produced enzymes and their application on the production of final value-added products becomes of prime importance [24].
One of the fields requiring an efficient process of lignocellulosic biomass is the biosynthesis of biodegradable bioplastics [25]. Among them, the polyhydroxyalkanoates (PHA) production from lignocellulosic-derived materials has gained remarked attention as an alternative to producing these bioplastics economically and sustainably [25]. PHA resemble several properties of the traditional petroleum-based plastics while being non-toxic, biocompatible, and biodegradable. Such characteristics make these compounds highly-valued as potential substitutes of petroleum-based plastics [25]. Typically, PHA are produced via SmF [26,27]. However, one of the main constraints of these processes is the high cost of the substrates, representing between 40 and 60% of the total fermentation cost [28]. Hence, Brigham et al. [29] have proposed coupling the PHA production to the food industry, building the PHA production plants next to large landfills of food waste or agro-industrial facilities. This way, transportation costs could be reduced, process efficiency is improved, and the valorization of the produced leftovers in such industries is promoted. Having this in mind, the lignocellulolytic enzymes produced from the SSF of BSG could serve to hydrolyze the same leftover sustainably as previously done with other lignocellulosic wastes [30]. Then, the released fermentable sugars could be used as a source for PHA production by specific microorganisms, consequently obtaining value-added products from a lowcost substrate, achieving a two-stage valorization of the residue, and presenting an advance on the development of this bio-based industry.
This study aimed to assess three different fungal strains, namely, Aspergillus niger, Trichoderma reesei, a n d T h e r m o a s c u s a u r a n t i a c u s , f o r p r o d u c i n g lignocellulolytic enzymes (mainly cellulases and xylanases) from the leftover BSG via SSF in a batch-SSF system. Besides, after an initial assessment for determining the maximum enzyme activity from the evaluated group, the produced enzymes have been tested to perform the enzymatic hydrolysis of BSG. Finally, the sugar-rich hydrolysates which resulted from the hydrolysis step were evaluated as a source for the PHA production via SmF using the PHA producers Burkholderia cepacia and Cupriavidus necator. Thus, proposing a two-stage valorization scheme for BSG by producing enzymes and valuable bioplastics.

Strains and inoculum preparation
Aspergillus niger (ATCC 16888) and Trichoderma reesei (ATCC 26921) were obtained from "Colección Española de Cultivos Tipo" (CECT, Spain). Thermoascus aurantiacus (ATCC 26904) from American Type Culture Collection (ATCC, Virginia). B. cepacia (CCM 2656) was purchased from the Czech Collection of Microorganisms, Brno, Czech Republic, and Cupriavidus necator (DSM428) from the German Collection of Microorganisms and Cell Cultures (DSMZ, Braunschweig, Germany). Strains were maintained at − 80°C in cryovials (Brand, Germany). Inoculum preparation for the fungal strains consisted of adding one pearl into a 100-mL Erlenmeyer flask containing a glucose-starch media (GS) (peptone 5 g·L −1 , starch 20 g·L −1 , and glucose 20 g·L −1 ). Flasks were placed in an orbital shaker at 120 rpm for 96 h, and the temperature was set at 30°C for A. niger and T. reesei and at 45°C T. aurantiacus.
For B. cepacia and C. necator, the same procedure was performed but using LB (Lysogeny Broth) medium. Flasks were placed in an orbital shaker at 120 rpm, 30°C, keeping an aerobic environment for 24 h. All reagents and materials had been previously sterilized at 121°C for 15 min.

Substrate
Brewer's spent grain was provided by Companyia Cervesera del Montseny (Catalunya, Spain). It was dried at 60°C overnight, and then it was stored at room temperature until further use. Preparation of the BSG for solid-state fermentation consisted of adding 2 × 2 cm sponge pieces as a bulking agent (3% w/w) as suggested by [31]. The sponge was selected as an alternative to lignocellulosic bulking agents, due to its ability to increase the porosity of the mixture without providing a biodegradable carbon source. Then, distilled water was added to the mixture to reach 78 ± 2% of moisture content (MC) (similar level to the original moisture content of the received BSG). The prepared BSG was sterilized at 121°C for 15 min, and after cooling, it was inoculated with 10% (v/w) of each of the evaluated fungi strains.

SSF experiments for enzyme production
SSF experiments were carried out (triplicates) in 0.5-L glass Erlenmeyer where 100 g of the prepared substrate (Section 2.2) was added. The flasks were placed in a temperature-controlled water bath, and each reactor was connected to an air mass flowmeter that provided a controlled flow such that the air entered at the bottom of the solid bed as detailed elsewhere [32]. The system was sealed such that the exhausted gases were conducted to an oxygen sensor, connected to an online acquisition system (Arduino-based). With this data, a respirometric monitoring was made as suggested by Ponsá et al. [32] by computing the specific oxygen uptake rate (sOUR). Experiments were followed up to 216 h. The temperature for the experiments was set at 37°C for A. niger and T. reesei as recommended by Díaz-Godínez et al. [10] and at 45°C for T. aurantiacus as proposed by dos Santos et al. [33]. Besides, the specific airflow rate (sAFR) was set at 0.05 L h −1 g −1 of dry matter of the BSG used in the experiments (DM), to provide enough oxygen to the solid substrate, and therefore, avoiding oxygen limitations in the system [34]. Experiments with the fungal strains were conducted in different runs (not simultaneously) due to capacity limitations of the respirometric system.

Enzyme extraction from SSF
For each strain, enzymes were extracted from the fermented BSG in 24-h intervals. Extraction consisted of putting 10 g of solid sample in a 250-mL flask and mixing with citrate buffer (pH 4.8) in a 1:15 ratio. The flasks were shaken for 30 min at 25°C. Then, the mixture was centrifuged at 5000 rpm for 10 min at 4°C, and the supernatant was filtered using a 0.45-μm membrane filter. The obtained filtrate was used as a crude enzymatic extract for the hydrolysis experiments and to determine the enzyme activity.

BSG hydrolysis with enzymatic extracts
These experiments consisted of mixing 2 g of dried and sterile BSG with each of the following solutions: the crude enzymatic extract obtained after the SSF process (Section 2.4), a commercial enzymatic cocktail Viscozyme L (Novozyme Inc., Copenhagen, Denmark) (1% dose), or a citrate buffer 0.05 M as a blank control. In all cases, a 5% (w/v) BSG:extract ratio was used (40 mL of citrate buffer in the case of the blank control). Hydrolysis was followed by measuring the reducing sugar content in the media up to 48 h.

PHA production in liquid culture
Sugar-rich hydrolysates obtained from the hydrolysis of BSG (Section 2.5) were used to produce PHA via SmF. First, 20 mL of the hydrolysates was placed into 100-mL Erlenmeyer flasks, and they were complemented by adding two mineral media as described by Kucera et al. [35]. Both substrates and materials have been sterilized at 121°C for 15 min. After cooling, the liquid substrate was inoculated by using 5% (v/v) of B. cepacia or C. necator. Fermentation was conducted at 30°C and 120 rpm in an orbital shaker assuring an aerobic environment. The system was monitored up to 72 h as suggested by Kucera et al. [35], and the cells were harvested after 48 h for C. necator and after 72 h for B. cepacia for PHA analysis.

Reducing sugars quantification
The reducing sugars content in the BSG was determined by using the supernatant obtained after a solid-liquid extraction of the solid material. The extraction consisted of adding distilled water in a 1:10 (w:v) ratio at 30°C. The mixture was mixed for 25 min at 180 rpm. The obtained supernatant was used to quantify the reducing sugars following the DNS method [38]. Briefly, 1 mL of sample was taken in a test tube, and 3 mL of DNS reagent was added. The solution was placed in a bath with boiling water for 5 min. After cooling, its absorbance was measured at 540 nm. The reducing sugars were calculated from a calibration curve of known concentrations using glucose as standard.

Xylanase activity
Xylanase activity was determined (triplicates) by using the supernatant obtained after a solid-liquid extraction of the solid material following the method presented by Ang et al. [39]. The extraction was conducted using citrate buffer (pH 4.8, 0.05 M) at 25°C, 120 rpm, and a 1:15 (w:v) ratio. The final product, xylose, was measured with the DNS method. The substrate was xylan from beechwood (Apollo Scientific, England) and the enzymatic reaction ran for 20 min at 50°C. One unit of xylanase activity (U) was expressed as 1 mol of xylose released from xylan from beechwood per minute. Xylanase activity was expressed as U·g −1 DM.

Cellulase activity
Total cellulase activity was measured (triplicate) using filter paper assay (FPase), as recommended by IUPAC, according to Ghose [40]. The extraction ratio for cellulases activity was the same as performed for xylanases. Final reducing sugars were measured using the DNS method. The substrate was 1 × 6 cm Whatman filter paper in 0.05 M citrate buffer (pH 4.8), and the enzymatic reaction was run for 1 h at 50°C. One unit of FPase (FPU) was expressed as the amount of enzyme that releases 1 mol of reducing sugars from Whatman filter paper per minute. Cellulase activity was expressed as FPU·g −1 DM.

Biomass and PHA extraction from liquid culture
The biomass produced during the bacterial fermentation was quantified by centrifuging 10 mL of the fermented culture at 5000 rpm for 5 min at 10°C. The cells were washed with distilled water and centrifuged again at 5000 rpm for 5 min at 10°C. The biomass pellet was dried at 60°C for 48 h and then weighed to obtain the cell dry weight (CDW) and used to determine the PHA content. PHA extraction was performed as described by Brandl et al. [41]. Briefly, 8-15 mg of the dry pellet was placed in a 2-mL vial, and 1 mL of chloroform and 0.8 mL of methanol-sulfuric solution were added. Benzoic acid (0.1 g·mL −1 ) was also added as the internal standard. Vials were adequately sealed and placed in a thermostatic bloc at 94°C for 3 h. After cooling, the vial content was transferred into 4-mL vials to be mixed with 0.5 mL NaOH 0.05 M through inversion for 5 min. The organic phase produced was used to quantify the PHA content of the biomass.

PHA quantification
PHAs were quantified by GC-FID using the extracts obtained from the biomass (Section 2.7.5). The GC system (Agilent 7820A) consisted of a flame ionization detector with an HP-Innowax column (30 m × 0.53 mm × 1 μm). The injection port was set at 250°C in splitless mode, and oven temperature was set at 70°C for 2 min, then temperature reached 190°C at 10°C min −1 and hold for 7 min. The detector temperature was set at 300°C. Identification and quantification were performed using calibration curves (internal standard) by comparing retention times of analytical grade standards (Sigma-Aldrich) P(3HB-co-3 HV-12% mol PHV) processed using the same conditions than the samples.

Inhibitory compounds quantification
Inhibitory compound content on the liquid hydrolysates was determined by HPLC using a modified method from [42]. Briefly, the HPLC system consisted of an Agilent 1920 Infinity UHPLC equipped with a UV-Vis Diode Array Detector G4212A and a Nucleosil 120C18 (3μm × 125mm × 4mm) column. The mobile phase (1 mL min −1 ) was a mixture of 0.05% H 3 PO 4 and acetonitrile: H 2 O (90:10). A gradient allowed changing 100% of 0.05% H 3 PO 4 until 100% acetonitrile: H 2 O after 38 min. Then this condition was held for 10 min. Quantification was performed by comparing samples from the analytical standards of the selected inhibitory compounds at the same conditions by using external standard calibrations.

Statistical analysis
Statistical differences of the experiments were evaluated by using one-way ANOVA (p < 0.05) using the Tukey test. Experiments were conducted in triplicates, and values were presented as mean ± standard deviation. Data were analyzed using SPSS 21. Table 1 summarizes the characterization made on the used BSG. As it can be seen, this substrate contains a significant amount of hemicellulose and cellulose, making this leftover a potential source of sugars after a hydrolysis process. Besides, lignin content is particularly low, which could be considered a positive aspect since this would limit the adverse effects of lignin on the availability of hemicellulose and cellulose [43]. In general, these values are in accordance with those previously found by other authors such as del Río et al. [44] but slightly different compared with Mussatto et al. [43] and Paz et al. [23] wherein they found higher lignin contents for BSG.

Substrate characterization
Besides, in other studies such as in Russ et al. [45], cellulose, hemicellulose, and lignin of the used BSG were 23-25%, 30-35%, and 7-8% respectively, while in Mussatto and Roberto [46] they were 16.8%, 28.4%, and 27.8% respectively. Thus, compared with the characterization presented here, some differences can be attributed to the inherent dependence on factors such as the quality of the used cereal, or the specific conditions of the beer processing [47]. Hence, it is expected that BSG could be a potential source of fermentable sugars after the hydrolysis of its cellulose and hemicellulose fractions.

Fungal strain assessment for enzyme production
The initial experiments consisted of assessing the ability of each of the selected strains to produce xylanases and cellulases through SSF. Table 2 summarizes the main results obtained for each fungal strain. As observed in Table 2, it was found that A. niger reached the maximum xylanase activity of the evaluated set (268 ± 24 U·g −1 DM) after 168 h of fermentation. However, this strain presented a second peak of xylanase activity (217 ± 30 U·g −1 DM) after 48 h of fermentation with no significant difference compared with the first one (p 0.086) (Fig. 1). It can be observed that, from the productivity standpoint, the peak at 48 h (4.5 ± 0.6 U·g −1 DM·h −1 ) was better than the 168 h peak (1.6 ± 0.1 U·g −1 DM·h −1 ), making the first more interesting to produce xylanases. In terms of cellulase activity, A. niger reached up to 1.9 ± 0.1 FPU·g −1 DM after 48 h of fermentation. Regarding T. aurantiacus, the maximum xylanase activity was reached after 72 h (241 ± 10 U·g −1 DM) and the maximum cellulases activity (3.0 ± 0.2 FPU·g −1 DM) after 168 h. Similarly, with Trichoderma reesei, the maximum xylanase activity was reached after 168 h (150 ± 24 U·g −1 DM) and the maximum cellulases activity (3.0 ± 0.1 FPU·g −1 DM) after 120 h of fermentation.
Comparing the enzyme activities found here against a commercial multi-enzyme complex (Viscozyme L) with a standard dose for enzymatic hydrolysis (1%) [35,48], the produced extracts reached similar activity levels ( Table 2). As observed, A. niger (p 0.986) and T. aurantiacus (p 0.584) extracts reached the same xylanase activity levels than Viscozyme L (1%) (p 0.986). Only T. reesei was significantly different (p 0.008), but in terms of cellulases, all extracts resulted similarly to Viscozyme L (1%) (p 0.183). Although the obtained cellulase activities were not as high as in other studies using similar feedstocks, they were in the same order of magnitude of Viscozyme L (1%). Such a result suggests that there was enough potential for using these enzymes for the hydrolysis processes. This aspect could be highly dependent on another factor such as the substrate selection. For instance, Dhillon et al. [49] have used apple pomace as a substrate with A. niger reaching cellulase activities as high as 384 FPU· g −1 DM, while authors such as Ortiz et al. [50] using wheat bran and T. reesei have reached activities around 96 FPU· g −1 DM. Also, Paz et al. [23] reached cellulase activities of 14.8 FPU·g −1 DM using BSG as substrate, while Leite et al. [21] reached between 50 and 60 FPU·g −1 DM using A. niger and A. ibericus with the same leftover. Although variability among substrates is significant, the obtained results were at the same levels as previous reports using BSG such in Leite et al. [21].  Similarly, by using other substrates, the xylanase activity levels could be as high as 1000 U·g −1 DM [49], but typical values have been found between 200 and 800 U· g −1 DM [51] as occurred here. Also, with BSG as a substrate, Leite et al. [21] reached xylanase activities between 250 and 310 U·g −1 DM with A. niger and A. ibericus. These values are in concordance with the results obtained in this study using the same fungal strain. Moreover, the xylanase activity obtained by Leite et al. [21] with BSG resulted similarly to those obtained in this study with T. aurantiacus.
G i v e n t h e s e r e s u l t s , i t c o u l d b e st a t e d t h a t T. aurantiacus and A. niger were the fungal strains with higher potential to produce lignocellulolytic enzymes from BSG from the evaluated group. Figure 1 details the time-course of the xylanases and cellulase activities from the SSF of the BSG of the three fungal strains. With A. niger (Fig. 1a), the maximum enzyme activities were obtained after the peak of maximum microbiological activity (indirectly measured as sOUR) (32 h), and the maximum cellulases activity coincided with the first peak of xylanases activity. Similarly, with T. aurantiacus (Fig. 1b), it can be observed that, as occurred with A. niger, the peaks of activities have occurred some hours after the peak of maximum sOUR, which suggests that there could exist a relationship among enzyme activities and the respirometric parameter as it was found in other SSF processes [52]. Hence, sOUR could be potentially used in further development stages as a monitoring and control parameter for producing these enzymes [53].
From Fig. 1, it can also be seen that, in the evaluated period, xylanases activity had a pseudo-cyclic behavior for both strains. For A. niger, there were two characteristic peaks separated by 120 h, and the levels achieved in both points were significantly higher than in the rest of the fermentation. Similar behavior can be found when using T. aurantiacus. This time, peaks were split by 144 h, and there were no significant differences among them (p 0.585). Finally, for T. reesei, the same pseudo-cyclic behavior was found (Fig. 1c). However, in this case, the xylanase activity levels were significantly lower, and the activity of the second peak was higher than the first one, leading to a decrease in the productivity of the process. During the fermentation of the three fungal strains, no significant change in the MC was detected (MC changes were below 3%), assuring that neither drying of the bed nor the leaching of the products affects the process. According to these results, fermentation time for maximum productivity corresponds to 48 h for A . n i g e r and 72 h for T. aurantiacus and T. reesei. Thus, the extracts obtained from these fermentations were selected for further hydrolysis tests.

Hydrolysis tests
In these experiments, the enzymatic extracts obtained from the SSF have been used to hydrolyze BSG to determine their efficiency. As seen in Fig. 2a, these tests have been conducted at 37 and 45°C [23,43] using the extracts from the three fungal strains, the commercial enzyme cocktail Viscozyme L (1%), and a blank control. From these tests, it was clear that there was a significant difference among the obtained extracts and the commercial enzymes during the first 12 h (p 000). Independently of the hydrolysis temperature, the commercial enzymes reached the highest reducing sugar release after 12 h reaching almost 35 g·L −1 . On the other hand, A. niger extract at 45°C showed the best performance among the evaluated extracts and temperatures, reaching 86-90% of the reducing sugar release obtained by the commercial enzymes at 45°C and 37°C, respectively, after 24 h. In general, these results suggest that working at 45°C is better than at 37°C for both the commercial enzymes and those produced from BSG.
On the other hand, after 12 h, a small decrease in reducing sugar release was detected. In this sense, it was found that after 24 h of hydrolysis, the extracts from BSG started to present fungal growth (samples were inoculated in Petri dishes, and fungal growth was confirmed), suggesting that these could be consuming part of the sugars released during the process. Thus, based on these results, the second set of experiments was conducted at 45°C using only A. niger extracts in an extended period to identify the effect of this reduction due to the potential fungal growth. As Fig. 2b details, the maximum reducing sugar release in this scenario was obtained after 24 h with 28.1 ± 1.0 g·L −1 , which was 36% lower than the maximum achieved by the commercial enzymes at the same conditions. Hence, the maximum sugars released corresponded to 0.56 g of reducing sugar g −1 DM. After reaching the maximum, a significant decrease in reducing sugar content occurred, reaching up to 51% after 48 h of hydrolysis.
Consequently, it is expected that, at the evaluated conditions, the hydrolysis could be conducted just for 24 h such that the potential loss of sugars can be minimized. Besides, due to the potential loss of sugars, future developments using such an enzymatic extract should include a purification step to limit this adverse effect, for instance, by using ultrafiltration processes with 30 kDa membranes capable of retaining the contaminants [54].
In general, the reducing sugar levels found here (28.1 ± 1.0 g·L −1 ) are similar to those found by other authors such as Mussatto and Roberto [46] that reached concentrations between 21 and 35 g·L −1 by using acid hydrolysis of the BSG. Paz et al. [23] using enzymatic hydrolysis by A. niger reached a total sugar content of almost 23 g·L −1 with raw BSG. Besides, the performance of BSG hydrolysate was competitive compared to other residues used to obtain fermentable sugars. For instance, Buzała et al. [55], using a commercial multi-enzyme preparation NS-22086, reached concentrations of 16.5 g·L −1 using poplar pulp, 14.5 g·L −1 for pine pulp, 16.9 g·L −1 for beech pulp, 15.9 g·L −1 for birch pulp, 17.7 g· L −1 for wheat straw pulp and 11.1 g·L −1 for hemp hard pulp. Similarly, Saucedo-Luna et al. [56], using a commercial multienzyme mixture from A. niger and T. reesei, reached 17.5 g· L −1 of fermentable sugars after 48 h of hydrolysis of agave bagasse. Also, Wei et al. [57], using the commercial enzymatic mixture Celluclast (Novozyme Inc., Denmark), reached 17.3 g·L −1 of fermentable sugars after 48 h of hydrolysis of rice hull. However, in many of these studies, chemical pretreatments have been performed before the enzymatic hydrolysis to enhance the sugar release, or synthetic substrates have been added as substrates, making these alternatives more expensive. Thus, by using BSG without any pretreatment appears as an interesting option for obtaining fermentable sugars economically and sustainably.

PHA production from sugar-rich hydrolysates
In these experiments, the BSG sugar-rich hydrolysates obtained from the enzymatic extracts of A. niger and the commercial enzyme cocktail Viscozyme L (1%) have been used to evaluate the PHA production in liquid cultures. The solution obtained after the hydrolysis of BSG with citrate buffer was also tested as control (blank). The PHA production was analyzed by quantifying the produced poly-3-hydroxybutyrate (P3HB), and the polyhydroxyvalerate (P3HV) as the most common forms of polyhydroxyalkanoates produced starting from sugars. In this case, the P3HV levels were negligible (data not shown), so the results only considered the produced P3HB.
In all cases, C. necator had slightly higher production per biomass than B. cepacia. This result suggests that C. necator is a better PHA producer than B. cepacia from hydrolyzed BSG. Previously, Rodrigues et al. [58] compared the PHA production using C. necator and B. cepacia from soybean as a substrate. In that study, C. necator resulted better than B. cepacia to produce PHA, reaching up to 0.84 g P3HB·L −1 . Also, the PHA content in the biomass was higher with C. necator, between 20 and 23%, compared with B. cepacia, which only reached 7-10%, suggesting that C. necator is better accumulating PHA from this kind of hydrolysates. Figure 3b summarizes the reducing sugars content of the BSG sugar-rich hydrolysates at time zero of the fermentation and the end of the fermentation. The reducing sugars content obtained after 24 h of hydrolysis with each enzymatic extract was considered as the initial value (corresponding to time zero of fermentation). As seen, after hydrolysis of the BSG, the commercial enzymes and A. niger extract released 38.9 g· L −1 and 21.8 g·L −1 of reducing sugars, respectively, in accordance with the results found before. As detailed, it was found that B. cepacia consumed more sugars than C. necator in all   . 3b). However, P3HB production with C. necator resulted better than using B. cepacia. Thus, even though reducing sugars consumption was higher for B. cepacia, that consumption was not reflected in higher P3HB production. A higher sugar consumption was observed in A. niger extracts than in the commercial enzyme's samples, and it could be attributed to the potential fungal growth observed before. Comparing A. niger extracts and the commercial enzymes as hydrolysis agents on the C. necator and B. cepacia P3HB production, A. niger samples obtained 9.0 ± 0.4 and 7.0 ± 0.6 mg P3HB·g −1 DM respectively, and the commercial enzymes samples 23.4 ± 1.8 and 19.7 ± 3.5 mg P3HB·g −1 DM respectively.
Additionally, A. niger extracts reached between 35 and 40% of P3HB production reached with the commercial enzymes with both strains. This result suggests that an optimization of the hydrolysis step to obtain higher sugar levels and, then, potentially increase the production of P3HB from the BSG is required. However, the obtained P3HB levels could also be attributed to the presence of inhibitory compounds produced during the hydrolysis step. In this sense, some authors [59] suggested that the formation of furan aldehydes as well as some soluble substances like hemicellulose-derived and cellulose-derived carbohydrates could interfere with the cellulolytic enzymes altering the hydrolysis step. Other compounds such as acetate, furfural, vanillin, and levulinic acid could also affect the bacterial fermentation, decreasing the PHA content [60]. Table 3 summarizes the concentration of some of the most common inhibitory compounds found in the hydrolysates used here (from Viscozyme L and A. niger).
In general, it could be stated that the commercial enzymes produced hydrolysates with a higher content of inhibitory compounds such as 5-HMF, furfuraldehyde, vanillic acid, or syringic acid compared with hydrolysates obtained from A. niger. Although the levels found in both hydrolysates allowed the growth of both bacterial strains used, these could be limiting the transformation of the available sugars into PHA. In future developments, a detoxification step could reduce the presence of such compounds, promoting a higher sugar consumption.
On the other hand, the results obtained here were in line with other reports using alternative raw material for producing PHA. For instance, Kucera et al. [35] reached 0.30 g PHB·g −1 CDW using B. cepacia and up to 0.88 g Fig. 3 a P3HB production in liquid culture using different enzymatic extracts and b initial and final reducing sugars content during P3HB production. CDW, cell dried weight; P3HB, poly-3-hydroxybutyrate.
Different capital letters indicate significant differences between the evaluated groups (p < 0.05) based on the Tukey test analysis Mean values are presented as the mean ± the standard deviation. "<" means below the detection limit of the method PHB·g −1 CDW with Burkholderia sacchari starting from spruce sawdust as a substrate. Also, Tripathi et al. [61] using sucrose as a carbon source obtained 0.35 g PHA·g −1 CDW with Pseudomonas aeruginosa. Thus, to the best of our knowledge, this is the first report showing a valorization strategy in two stages for BSG, by using a combined approach of obtaining self-produced enzymes (via SSF) and the further PHA production with the same BSG hydrolysates.

Conclusions
A. niger produced the highest enzymatic activities from the SSF of BSG reaching up to 268 ± 24 U·g −1 DM, obtaining an enzymatic extract capable of producing significant fermentable sugars contents. Although A. niger reached similar xylanase and cellulase activities than Viscozyme L (1%), its efficiency to hydrolyze the BSG resulted significantly lower. However, the BSG hydrolysates obtained from A. niger enzyme extract have supported the P3HB production using C. necator and B. cepacia reaching 9.0 ± 0.4 and 7.0 ± 0.6 mg P3HB·g −1 DM, respectively. Future work will be needed to optimize the SSF and hydrolysis steps to increase the enzymatic activities as well as the release of sugars from BSG. Furthermore, adding a pretreatment/ detoxification strategy coupled with the hydrolysis step could serve to improve the overall performance of the proposed approach. These results suggest that BSG can be used as a low-cost raw material for obtaining both lignocellulolytic enzymes and PHA as value-added products resulting in a valorization approach in two stages.
Acknowledgment EM acknowledges the Juan de la Cierva research fellowship (FJCI-2015-24,616) from the Spanish Ministry of Economy and Competitiveness.
Availability of data and material It will be available upon request.
Authors' contributions Jordi Llimós: data collection, data analysis and wrote the paper. Oscar Martínez-Avila and Elisabet Marti and: data analysis and critical revision of the manuscript. Carlos Corchado-Lopo: data collection. Teresa Gea, Laia Llenas, and Sergio Ponsá: Financial support acquisition, study conception, and critical revision of the manuscript.
Funding information This work was supported by the Spanish Ministerio de Economía, Industria y Competitividad (CTM2016-81038-R), VALORA Project.