D-lactic acid production from hydrothermally pretreated, alkali delignified and enzymatically saccharified rockrose with the metabolic engineered Escherichia coli strain JU15

Rockrose lignocellulosic residues (RR) were selectively fractionated for hemicellulose separation using autohydrolysis, followed by an alkaline treatment to solubilize the lignin. The cellulose-enriched solids were used to study the effect of solid loading (SL: 2–10%) and enzyme dosage (ED: 6.34–23.66 FPU/g dry biomass) on saccharification using a Doehlert experimental design, followed by fermentation with the metabolic engineered Escherichia coli strain JU15 to produce D-lactic acid (DLA). Pretreatment increased glucan content and enzymatic digestibility up to 84%. A significant positive effect of SL and ED was found for glucose production, but SL negatively impacted glucose yield. DLA concentrations and productivity varied from 8.85 to 32.98 g/L and 1.11 to 2.17 g/(Lh), respectively. Overall process efficiency strongly depended on saccharification yield and varied from 33 to 71%. These results indicate that sequential autohydrolysis, delignification, and fermentation of RR may be a potential relevant strategy for D-lactic production in the biorefinery framework.


Introduction
Lactic acid is an enantiomeric GRAS (Generally Recognized as Safe) chemical that is usually chemically synthesized as a racemic mixture of L-lactic acid and D-lactic acid. This organic acid has many applications in the food, cosmetic, medical, and pharmaceutical industries [1]. Most noteworthy, lactic acid is the precursor of polylactic acid (PLA), which is used for the manufacture of biodegradable bioplastics that can replace some of the current polyethylene terephthalate (PET) applications. According to the European Bioplastics Organisation, the global bioplastics production was 2.1 million tons in 2020 and is expected to reach close to 2.9 million tons in 2025 [2]. The copolymerisation of different contents of L-and D-lactate polymers improves the thermal and mechanical properties of PLA [1]. These characteristics are very important to promote the replacement of PET.
The production of lactic acid by fermentation using lactic acid bacteria (LAB) is one of the oldest biotechnological processes based on natural resources, e.g., milk [1], and from ancient times until the present day, many studies have been made for the optimization of this bioprocess. An extensive revision is made by Eş et al. [3], where the diverse strategies applied in the production of lactic acid to improve yields and productivity are presented and discussed. These include the utilization of co-cultures, genetic and metabolic engineering, and microbial immobilization techniques. Among the remaining hurdles for producing lactic acid through fermentation is the need for cheap cultivation media, namely cheap carbon and energy sources, as well as minimising the use of complex nutrients such as yeast extract. The selection of the microorganism is particularly important in terms of both high lactic acid production capacity and high optical purity. The production of optically pure D-or L-lactic acid is one of the advantages of a fermentation process as opposed to chemical synthesis [4]. Utrilla and collaborators [5,6] developed the genetically engineered Escherichia coli strain JU15 with the ability to produce D-lactic acid (DLA) with high purity (up to 99.84%) yield and productivity [7]. This bacterial strain can achieve a DLA yield of 0.95 g/g sugar (includes consumption of xylose, glucose, and arabinose) and a productivity of 0.79 gL −1 h −1 in a batch fermentation process with a mineral medium.
Suitable fermentation substrates can be derived from sugar-rich renewable starchy biomass [8][9][10], or preferably from cheaper, residual biomass such as lignocellulosic wastes [11,12], e.g., sugarcane [13,14]. In fact, some renewable lignocellulosic industrial residues are ideal biorefinery feedstock, due to environmental, logistic, and economic reasons, as they are already concentrated in industrial facilities [15]. A relevant type of these industrial residues, still poorly explored in the biorefinery framework, is the residual biomass from the commercial distilleries producing essential oils for the perfume and cosmetic industry, such as lavender, cedars, or rockrose (Cistus ladanifer), among many other herbaceous and woody feedstocks.
Specifically, rockrose is an endemic shrub in Mediterranean-type climates, used for the production of several specialty compounds, whose residual biomass is currently only used in low-value energetic applications by simple combustion. Previous studies have shown that rockrose residues have the potential to be used in biorefinery applications, namely for the production of added-value products such as oligosaccharides [16] and phenolic compounds [17,18], [16,17]. This is accomplished by a properly design fractionation strategy [18] using different pre-treatments that include a hydrothermal process (autohydrolysis), which allows the selective production of the added-value hemicellulosic oligosaccharides, and an alkali processing, which enables a signification delignification and the recovery of the phenolics from lignin. This strategy yields a cellulose-rich solid stream that needs to be valorised. However, the fractionation processes were optimised aiming improved product yields, and not for improving cellulose digestibility, conversely to what is usually reported in literature. As such, it is important to evaluate if this strategy is compatible to the use of the cellulose-rich solid in a subsequent fermentation process within the biorefinery biochemical platform.
The main goal of this work is to maximise the production of DLA by E. coli strain JU15 grown in a rockrose-based medium through a sequential hydrolysis and fermentation process. The optimisation is carried out using a Doehlert experimental design for two factors to study the effects of solid loading (2-10%) and enzyme dosage on both saccharification and DLA fermentation of rockrose cellulose-rich solids.

Materials and methods
A general scheme for the followed methodology is presented in supplementary material ( Figure S1).

Rockrose residue source and feedstock preparation
The rockrose (Cistus ladanifer) residues, kindly provided by Quinta Essência Lda. (Portel, Portugal), were the solid residues of 2-to 4-year-old plants processed to extract the essential oils by steam distillation. This biomass was dried at room temperature and then ground in a knife mill to pass a 6-mm screen (Retsch, Haan, Germany). The milled biomass was subjected to ethanol and water extraction as described in Alves-Ferreira et al. [16]. Extractive free biomass, hereinafter denominated as RR (feedstock), were dried at 40 °C, homogenised in a defined lot, and stored at room temperature.

Autohydrolysis
In order to selectively hydrolyse hemicellulose, the feedstock was pretreated using liquid hot water (autohydrolysis process, AH) in a 600-mL stainless steel Parr reactor (Parr Instruments Company, Moline, IL, USA), under non-isothermal conditions up to 205 °C with a liquid-tosolid ratio of 6:1 (dry basis) as optimized before [17]. The temperature, agitation (150 rpm), and pressure were controlled by a Parr PID controller (model 4842). Upon reaction completion, the solid fraction, mainly containing cellulose and lignin, was separated by filtration, dried at 45 °C to reach a moisture content below 10%, and stored in plastic bags until further use. This material was denominated RR-AH.

Delignification
Delignification experiments were carried out by mixing RR-AH with a 4% sodium hydroxide solution in a liquidto-solid ratio of 10:1 (dry basis) in capped Schott-Duran flasks. The flasks were autoclaved (Felisa FE-399, Guadalajara Jalisco, Mexico) for 2 h at 130 °C (and 1.6 atm), after which the temperature was rapidly decreased by liberating the vapor. Solid and liquid fractions were separated by filtration and the solid fraction was thoroughly washed four times (two times with deionised water, followed by a third wash with dilute sulfuric acid solution to neutralise the pH of the sludge, and a final wash with water). The resultant delignified sample, hereinafter denominated as RR-DL, was dried at room temperature and stored in plastic bags, also at room temperature, until required.

Optimisation of the sequential saccharification and fermentation process
RR-DL was subjected to enzymatic saccharification according to Doehlert experimental design (see below), followed by fermentation without any separation of solids and liquid, and maintaining the same vessel used in saccharification.

Enzymatic saccharification
A Doehlert experimental design was used to establish the effects of the solid loading (X 1 ) between 2 and 10% (w/v) and enzyme dosage (X 2 ) between 5 and 25 FPU/g dry biomass. Table 1 shows the combination resultant from five levels for solid loading combined with three levels for enzyme dosage that enables the estimation of curvature effects for each independent variable. To reduce mixing difficulties during saccharification, sterilised 200-mL caped mini-reactors fitted with a pegmixer [7,19,20] were used. To each mini-reactor, the following were added: appropriate amounts of solids, sterile water, the stock solution of sodium citrate buffer to obtain a final concentration of 50 mM, and pH was adjusted to 4.8 with 5 M HCl, when needed. After mixing to obtain a homogenous mixture, the appropriate volume of cellulase, Accellerase 1000 (Genencor Inc.), was added and the total volume adjusted to 200 mL. Kanamycin was added as a preservative to a final concentration of 30 mg/L. The saccharification was performed at 50 °C and 150 rpm for 24 h. The enzymatic preparation of Accellerase 1000 was characterized for filter paper (45 FPU/mL) and β-glucosidase (320 IU/ mL) activities as described below (Sect. 2.5.2); hence, no additional β-glucosidase was added for the saccharification assays. Samples of 1.5 mL were taken each 3 h until 12 h and then at 24 h, and centrifuged immediately (13,000 g, 5 min, room temperature; Eppendorf centrifuge 5410, Hamburg, Germany). The supernatant was frozen for posterior glucose quantification. The experiments were performed, at least, in duplicate.

Fermentation of enzymatic hydrolysates
The fermentation of enzymatic hydrolysates was performed using the lactogenic E. coli strain JU15, prepared as indicated above, and fermentations were performed in the same vessels used for saccharification. After 24 h of enzymatic saccharification, and before inoculation, the pH was adjusted to 7 with 6 N KOH and supplemented with all components used for inoculum preparation, except sodium citrate, since the saccharification was performed in citrate buffer, and glucose that was replaced by the sugar released during the saccharification step. The mini-reactors for DLA production were inoculated by re-suspending a cell pellet of inoculum centrifuged at 4,700 g (8 min, room temperature) (Sorvall-ST16R Thermo Scientific, Germany) to start the cultures with ~ 3 OD 600 (ca. 1.11 g DCW L −1 ). Absorbance was measured in NanoDrop 2000c (Thermo Scientific, USA). Finally, the cultures were incubated at 37 °C and 60 rpm (in the mini-reactors fitted with the peg-mixer), and fermentations were controlled at pH 7.0 by the automatic addition of 2 N KOH. Samples were taken every 2 h until the end of fermentation and processed as described in the saccharification section. The supernatant from the beginning (t 0 ) and end (t f ) of the fermentation was analyzed for glucose, xylose, and organic acids by HPLC.

Analytical procedures
Solid biomass samples were characterized according to NREL's laboratory analytical procedures [22]. Sugars and organic acids were determined using a Waters HPLC system (Milford, MA) comprised of a 600E quaternary bomb, 717 automatic injector, and 2410 refraction index detector, and using an Aminex HPX-87H column (Bio-Rad, Hercules, CA). Sulfuric acid 5 mM was used as mobile phase at a flow rate of 0.5 mL/min and 50 °C [6,23].
The quantification of glucose in the enzymatic hydrolysates samples was measured with a biochemical analyzer (YSI model 2700, YSI Inc., Yellow Springs, OH, USA).

Analytical evaluation of cellulose digestibility
Analytical enzymatic digestibility assays were performed for RR, RR-AH, and RR-DL, according to NREL technical protocol [24] in order establish their differential cellulose digestibility. Cellulase and β-glucosidase enzymes were purchased from Sigma-Aldrich (Celluclast 1.5L with 66 FPU/ mL and Novozyme 188 with 814 UI/mL, respectively). Biomass containing 0.1 g glucan was mixed with cellulase (60 FPU/g glucan) and β-glucosidase (64 UI/g glucan) in 50 mM citrate buffer pH 4.8, and 100 μL of 2% sodium azide in a total volume of 10 mL. Samples were incubated for 72 h in an orbital shaker (TEQ/OSFT-LS, Portugal) at 150 rpm and 50 °C. The experiments and proper blanks were performed, at least, in duplicate.

Enzymatic activity characterisation
Filter paper (FPase) activity was measured according to Ghose [25]. Beta glucosidase activity was measured by an adapted protocol based on Berghem's procedure [26]; enzyme solution was mixed with 1 mL of 5 mM p-nitrophenyl-β-d-glucopyranoside, prepared in 50 mM sodium citrate buffer (pH 4.8) at 50 °C for 10 min. The reaction was stopped by the addition of 2 mL of 1 M Na 2 CO 3 plus 10 mL of water. The amount of liberated p-nitrophenol was measured spectrophotometrically at 400 nm. The blank reaction was made by substitution of p-nitrophenyl-β-dglucopyranoside solution by citrate buffer. The calibration curve was made with different volumes of 1 mM p-nitrophenol in 50 mM citrate buffer. One unit of activity (IU) was defined as the amount of the enzyme that catalyzes the release of 1 μmol of p-nitrophenol per minute.

Calculations
The solid yield obtained after each treatment, Y s , was calculated according to the following equation: The glucan recovery in the pretreated solids is given by: where G n and G nF are the percentages of glucan in the pretreated solid residue and the feedstock, respectively. For the other structural components, e.g., xylan or Klason lignin, equivalent equations were used.
Glucan saccharification yield (GSY) was calculated by the following equation: The potential glucose concentration is being calculated by multiplying the biomass concentration at the beginning of enzymatic saccharification by its glucan content and 1.11 that converts cellulose to glucose equivalent.
Lactic fermentation Yield (Y PS ) is given by Eq. 4: where [lactic acid] m is the maximum concentration of lactic acid obtained during fermentation and [glucose] 0 and Dry biomass after treatment Dry biomass before treatment × 100 (2) × 100 [glucose] m are the glucose concentration at the beginning and the maximum point of the fermentation process.
Overall process efficiency (OPE) was calculated through Finally, lactic acid volumetric productivity (Q P ) was calculated by the equation: where [lactic acid] m is the maximum concentration of D-lactic acid and t is the time when the maximum concentration of D-lactic acid was achieved.

Modelling and statistical analysis
The responses of the Doehlert experimental design experiments were modeled based on a second-order polynomial equation (Eq. 7): where Y is the response, X are the independent variables, and the subscripts 1 and 2 are related to solid loading and enzyme dosage, respectively. β 0 is the regression coefficient at center point; β 1 and β 2 are the linear coefficients of the solid loading and enzyme dosage, respectively; β 12 is the second-order interaction coefficient between the two independent variables; β 11 and β 22 are their quadratic coefficients; and ε are independent random errors, assumed to be normally and independently distributed.
The linear multiple regression of Eq. 7 and its analysis of variance (ANOVA) were carried out using Microsoft® Excel 2016 regression tool pack. The best enzymatic saccharification conditions were determined by using the Microsoft Excel® 2016 Solver tool based on the best-fit equations. Coded representation of the variables was used for all calculation purposes.

Effect of autohydrolysis and delignification on the chemical composition of biomass
The chemical composition of lignocellulosic biomass is important to envisage its potential to produce bioproducts. As shown in Table 2, RR has a very interesting composition for the biochemical upgrade within the biorefinery framework, as it is constituted by more than 50% of polysaccharides, most noteworthy cellulose that accounts for 60% of the potential sugars. Nevertheless, as a woody material, there is also a relevant content of lignin (27.71%), which may hinder the enzymatic recovery of glucose. This composition regarding total sugars and Klason lignin content is similar to the values described before [27,28] and in the same magnitude as wheat straw [29], olive tree pruning [30], and bamboo [31] To reach an efficient cellulose hydrolysis, it is usually necessary to fractionate the lignocellulosic biomass, removing both hemicellulose and lignin. In the present study, the autohydrolysis pretreatment was used as a process to recover the added-value hemicellulosic oligosaccharides as previously optimized [16], originating a solid with increased content of glucan and lignin reaching 34.64 and 46.83 g/100 g of pretreated material (RR-AH), respectively. This effect is quite similar to the previously applied pre-treatments with dilute acid hydrolysis [28] and steam explosion [32] that also preferably solubilise hemicellulose and generated a solid residue with increased glucan and Klason lignin content. Although autohydrolysis induced to a lower lignin increase regarding feedstock, e.g., when compared with steam explosion [32] and similar to dilute acid hydrolysis [28], the lignin content of RR-AH is quite high. The increase of glucan and lignin content after autohydrolysis treatment was also observed with other types of biomass such as rice [33], wheat [29,34], and rapeseed straw [31].
As such, a subsequent delignification process was then applied. An alkaline pretreatment, based on the use of sodium hydroxide, was chosen, as it is well known that this process can effectively solubilise lignin [30,31] for many feedstock. Specifically, a previous study made on RR-AH by Alves-Ferreira et al. [18] showed that varying sodium hydroxide dosage (2 and 4%) and time (1 and 2 h) at 130 °C (1.6 atm) enables a high delignification, especially for the highest NaOH dosage, and as such, this was chosen for the present study. The solid yield obtained for this second pre-treatment step (approx. 40%) indicates that a significant part of the lignin was solubilised to the liquid fraction. Klason lignin content decreased from 46.83 (RR-AH) to 15.26 g/100 g of pretreated biomass, thus enabling a lignin content decreased almost sevenfold concerning the initial raw material (from 27.71 to 3.99% in respect to RR) that corresponds to a solubilisation of 87% of the initial Klason lignin. Equivalent lignin solubilisation was also obtained with other biomasses after hydrothermal treatments, [33,34] but this study surpasses the delignification efficiency (65%) reported previously for a similar material [32]. As such, the two-step fractionated solid presents a very high content of polysaccharides (88.76%), without significant losses of the initial glucan present in the RR-AH.

Effect of autohydrolysis and delignification on biomass digestibility
RR presented a higher saccharification yield than nontreated rockrose studied before [28,32] using the same enzymes, which only reached 0.9 and 1.2%, respectively. This could be explained either by the fact that in the cited reports, 5% of solid loading and lower content of enzymes (15 FPU and 15 CBU) were used or by the different complexities of the biomass itself since the plants used before were older (10 years) than the ones used in the present work (2-4 years). In this work, the digestibility of the feedstock and the effect of sequential autohydrolysis followed by alkali treatment on the digestibility of rockrose biomass were evaluated using low(er) solid loadings and high(er) enzyme dosages conditions to determine the maximum extent of digestibility possible [24]. Autohydrolysis pre-treatment increased 1.8-fold the glucan saccharification yield (from 22.59 to 40.67%) and the following delignification further more than duplicated it, reaching 84.25%, which is among the highest reported in the literature for rockrose-derived materials [27,28]. As indicated before, the autohydrolysis solubilised part of hemicellulose and lignin leaving more accessible cellulose fraction to enzymes action. The subsequent alkaline treatment led to a decrease of hemicellulose and lignin content, exposing the cellulosic fraction even more and disrupting even more its structure. This enabled easier access of the enzyme to the glucan fraction, with the increase of both glucan saccharification yield and glucose concentration ( Table 3). The improvement of the digestibility is even more evident when analyzing the value for glucose yield based on dry processed biomass that also reflects the effect of the higher glucan content achieved as a consequence of the treatments. It is also interesting to observe the glucose yield as a function of the initial biomass. In this case, an almost threefold increase can be observed.
A similar trend was observed for the enzymatic xylan saccharification yield, with delignification inducing an even higher increase (approx. eightfold, data not shown). However, final xylose concentration and xylose yield based on dry processed biomass or initial biomass decrease significantly, because of the concomitant xylan solubilization during the autohydrolysis and delignification process.
Hydrothermal pre-treatment followed by sodium hydroxide post-treatment applied to other biomass gave similar results of cellulose enzymatic hydrolysis yield. For instance, studies performed with wheat straw [34] [34] showed increased enzymatic hydrolysis yield from 36 to 84% (with 6% NaOH) and 60 to 82% (2% NaOH) for the treatment with autohydrolysis and after post-alkali treatment, respectively. For rapeseed straw, a doubling of cellulose saccharification yield was also observed after the sequential fractionation of autohydrolysis followed by NaOH (2%) treatment [31]. Steam-explode bamboo [35] also presented improved enzymatic hydrolysis (from 6 to 32%) and then to 74% followed by alkali treatment. Compared to the reported results, the improved enzymatic hydrolysis performance achieved for rockrose is particularly relevant, specially taking into consideration that the processes were optimized for products recovery rather than to improve cellulose digestibility.

Effect of solid loading and enzyme dosage on the saccharification of pretreated rockrose
To optimise the enzymatic saccharification of RR-DL and minimizing operation costs, the influence of solid loading (SL) and enzyme dosage (ED) was studied using a two-factor Doehlert experimental design. The kinetic performance of the enzymatic hydrolysis step is presented in Fig. 1A and B, for glucose concentration and glucan saccharification yield (GSY), respectively. As expected, it was found a decrease in saccharification yield and an improvement of glucose concentration as compared to the conditions used to determine the maximum extent of digestibility possible described above. Although the reaction time is lower in this assay, 24 h Table.3 Glucan digestibility of extracted Cistus ladanifer residue (RR) and solid residues after autohydrolysis (RR-AH) and subsequent alkali extraction (RR-DL) at 1% of glucan loading and 60 FPU of cellulase (Celluclast 1.5L) and 64 UI of β-glucosidase (Novozyme 88) per g glucan conversely to 72 h, it is possible to observe from figures that both glucose concentration and GSY have almost reached a plateau at 24 h of experiment, and hence the fermentation step can be started.
These results were modeled using an empirical secondorder polynomial equation (Eq. 1). The statistically significant (p < 0.05) regression parameters are presented in bold typeface in Table 4 and overall performance is graphically represented in Fig. 2. Both models, for glucose concentration and saccharification yield, presented a very good fitting (R 2 = 0.992 and 0.979, respectively), and are also statistically significant at the same p level. Figure 2A shows the response surface obtained for glucose concentration (Glc) that varied from 12.9, obtained with the smallest SL and ED, to 47.2 g/L, obtained with the highest values of SL and ED. This graph shows the greater effect of solid loading when compared to enzyme dosage, and the positive interaction between these 2 variables. According to the model, its response is significantly (p < 0.05) affected by solid loading with the highest impact (β 1 = 16.52) ( Table 4), followed by enzyme dosage, and also, with a positive impact (6.99). The second-order interaction coefficient between variables SL and ED (β 12 ) is also statistically significant, indicating a synergistic effect, but with a lower magnitude as compared to the direct effects of both variables. Figure 2B shows the response surface for glucan saccharification yield (GSY) that varied from 40.8 to 75.1%. Conversely to what was observed for glucose concentration, GSY presents a negative slope when solid loading increases and enzyme dosage decreases. According to the model, GSY response is significantly and positively affected (p < 0.05) by enzyme dosage (β 2 = 13.20), but negatively affected by solid loading (β 1 = − 8.68), as expected. Conversely to the glucose concentration, the second-order interaction coefficient   between the two variables is not statistically significant at p < 0.05. As a consequence, and as expected, the maximum GSY value was obtained with the lowest solid loading and the highest enzyme dosage. This is justified by the known inhibitory effects induced by high solid loadings [36]. Using the model data and the linear programming approach, the optimal condition estimated for the maximum glucose production would be obtainable by using 10% (w/v) for solid loading and 25 FPU/g for Accellerase enzyme dosage. Nevertheless, the estimated increase may not justify the higher usage of the enzyme at the tested time-frame (24 h). Applying the same approach to establish the highest saccharification yield, within the experimental design, yields that the optimal condition for saccharification yield can be obtained by using 2% (w/v) for solid loading and 25FPU/g DB for enzyme dosage. This would yield an 86.51% yield, but that would induce a significant lower glucose concentration, which is not desirable for the subsequent fermentation process, and fall short of the required performance for an industrial process.

Fermentation to lactic acid with the lactogenic E. coli strain JU15
After the pH correction of the enzymatic hydrolysis syrups to 7, the E.coli JU15 was inoculated to each assay microreactor. The fermentation was followed by the addition of KOH 2 N, which was used to neutralize the lactic acid produced by the conversion of sugars. Variation of added KOH to each run during the fermentation process and the concentration of sugars and the lactic acid at the beginning (t0) and end of fermentation (tf) are shown in the supplementary material. A continuous sugar conversion to D-lactic acid along the time terminates with the presence of a plateau indicating that lactic acid production has reached its maximum, or sugar depletion has occurred. The elapsed time between the beginning and the end of fermentation was used for the calculation of the volumetric productivity (Q P ), which are gathered with other kinetic and stoichiometric parameters in Table 5. Potassium hydroxide consumption along fermentation varied from 32 to 98 mmoles, depending on sugar conversion to lactic acid (supplementary material). Generally, it was possible to observe a very rapid consumption of KOH, indicating that the strain was preferably metabolizing the sugar mainly to lactic acid and not to cell mass. Volumetric productivity of D-lactic acid varied with different solid loadings, after the first 2 h of fermentation. Assays with equivalent sugar concentrations have similar volumetric productivity although slightly lower productivities were obtained in the presence of higher solid loadings. In fact, it is possible to see that the higher the solid loading, the more pronounced is this variation, being possible to see its effect by the time where the plateau is reached (Table 5). When the solid loading was higher than 8%, the presence of the solids has more impact in the DLA production, slowing the completeness of the fermentation process. Assays B and D, with 10 and 8% of SL, have a final fermentation time (t f ) of 24 and 14 h, respectively. These are the example of this situation meaning that at this high SL, a difference of solid (2%) has more impact than in presence of low SL as can be seen with assays C and E (supplementary material and Table 5).
Analysis of supplementary material and Table 5 shows that most of the glucose was consumed for DLA production, varying from 8.85 to 32.98 g/L. Albeit JU15 can consume xylose, during the fermentation, xylose apparently was not consumed. This could be explained in part by a diauxic effect. This hypothesis is corroborated by previous studies made by Utrilla et al. [7] with JU15 that observed depletion of pentoses (xylose and arabinose) after glucose consumption, i.e., that glucose represses the consumption of xylose. Also, inocula were developed using glucose as a carbon source; hence, it is likely that transport and xylose catabolism could be repressed in the E. coli cells used to ferment the sugars present in the syrups. Furthermore, acetate concentrations in all runs remained constant or slightly decreased at the end of fermentation (data not shown). D-lactic acid production was higher in experiments with higher glucose concentration at the beginning of fermentation. Fermentation yield was higher than 1 in some experiments (Table 5), which could be derived by the consumption of other minor sugars not quantified by HPLC, or, most probably, by the continuous production of glucose as Accellerase enzyme and solids were not removed. Most runs reached the maximum value of fermentation yield (1) with exception of runs F and G, which can be within the experimental error. Volumetric productivity (Table 5, Q P ) varied from 1.11 to 2.17 g DLA/(Lh). This was the fermentative parameter most affected by the presence of solids. The values of D-lactic acid yield and volumetric productivity of the present work fit and surpass the values indicated in Utrilla et al. [7] which obtained a fermentation yield of 0.98 and volumetric productivity of 0.5 g/Lh when fermented sugarcane bagasse hydrolysate with initial OD of 0.1 of JU15 in a pilot scale. Comparing with the fermentation of AV03 strain, a metabolic engineered strain from JU15 [7], in corn stover dilute acid hydrolysates, the present study reached equivalent values of fermentation yield, but again surpassed volumetric productivity (1.11 and 1.21 g/Lh, respectively) [7].
The present study shows that it is possible to obtain high concentrations of D-LA, from 8.85 to 32.98 g/L, using RR and lactogenic engineered E. coli JU15, with higher volumetric productivity for the range of the runs performed in the present study (1.11-2.17 g/Lh).
Overall process efficiency (OPE) ( Table 5) varied from 33 to 63%, with the solid loading negatively affecting saccharification yield and consequently OPE, as fermentation yield was always 1 for most of the assays.

Conclusions
This work aimed to produce DLA from rockrose residues. Sequential autohydrolysis and alkaline delignification removed hemicelluloses and lignin enabling their recovery in added-value products, and significantly simultaneously increasing enzymatic cellulose digestibility. Doehlert experimental design showed that solid loading and enzyme dosage positively affected glucose titers, being SL the most relevant. Glucan saccharification yield is positively influenced by ED, but SL showed a negative effect. Enzymatic hydrolysate fermentation with JU15 produced DLA from 9.91 to 32.96 g/L varying the productivity from 1.11 to 2.17 g/Lh. Fermentation was efficient reaching yields of 100% in almost all cases, with overall process efficiency being strongly dependent on saccharification yield.