**4. Discussion**

The biodegradation of contaminants and the production of renewable energies are among the most important challenges of modern society. Bioethanol is a sustainable fuel that can be obtained by the yeas<sup>t</sup> fermentation of sugars [42]. When using sugars from lignocellulosic by-products instead of sugars from food crops, the product is called second-generation bioethanol. However, the hydrolysis of these polymers generates, in addition to fermentable sugars, compounds that act as inhibitors of the yeas<sup>t</sup> catalyzing further fermentation. The formation of furfural from pentoses was described many years ago [43]. Furfurals, and specially HMF and its derivatives, are feedstock for the synthesis of numerous valuable products. Therefore, their production in a sustainable manner is receiving an increasing interest [44]. On the other hand, from the bioethanol-production point of view, it is a challenge eliminating these inhibitors leaving sugars intact, with the purpose of increasing efficiency in second-generation ethanol production. Furfural and 5-hydroxymethyl furfural (HMF) are not the only inhibitory compounds present in the hydrolysates, but many others are not so well characterized, such as methanol or acetate [39]. The selection of yeas<sup>t</sup> resistant to aldehydes may partially circumvent the problem. In fact, yeasts resistant up to 40 mM furfural and 80 mM HMF have been described recently [45,46]. Nevertheless, and taking into account the diversity of hydrolysates, the direct elimination of inhibitory compounds has to be taken into account as an alternative (bio)technology.

In this manuscript, we describe the selection of an adaptive mutant of *Pseudomonas pseudoalcaligenes* CECT 5344 able to assimilate furfural. The mutation was found to be located in a regulatory gene of the AraC/XylS family of activators [47]. In *E. coli, araC* is necessary for the assimilation of L-arabinose, a five carbon sugar [37]. Although speculative, perhaps it is not casual that the homologous gene in *P. pseudoalcaligenes* CECT 5344 could recognize as an activator the five-carbon dehydrated and oxidized pentose FA. This hypothesis has to be tested experimentally in the future. AraC homologous are proteins approximately 300 aa long, whose C-terminal segmen<sup>t</sup> is the HTH domain interacting with DNA. The mutation in the adapted strain was detected just in this domain. The substitution of a

hydrophobic amino acid by a basic one (L262R) introduces a positive charge in the protein that seems to be essential for interacting with the negatively-charged DNA molecule. The less-conserved N-terminal domain of the AraC family is presumed to contain binding sites for specific activator molecules that confer specificity to each member [47].

Adaptive mutations frequently target regulatory genes [34], although there are interesting examples in which the genetic adaptation lies in catabolic genes. For example, in *P. putida* KT2440, a single point mutation was detected causing the suppression of a frameshift mutation in the transporter (*galT*), thus allowing the evolved strain to grow in gallic acid [48]. The *hmf* locus in *P. pseudoalcaligenes* contains two di fferent modules, a catabolic one homologous to *C. baisilensis* HMF14 (plus *benE*) and another one homologous to *P. putida* Fu1 containing the regulatory gene *araC* (Figure 1B). As far as we know, no gene homologous to *benE* has been described in the context of FA assimilation. *benE* is a member of the MSF-family of transporters, homologous to the benzoate transporters. Since both compounds, benzoate and furoate, are aromatic monocarboxylic acids, it can be speculated that BenE is a furoate transporter, although this hypothesis needs further experimental evidence. The *hmf* locus (Figure 1) is flanked by mobile genetic elements, thus suggesting that it has been horizontally transferred. Furthermore, this hypothesis is in agreemen<sup>t</sup> with the sequence composition of the locus. The average GC content of the genes flanking the *hmf* locus in *P. pseudoalcaligenes* (green genes in the Figure 4B) is around 62%, which is also the %GC content of the genome of *P. mendocina* ymp (Figure 4A, 62.8%) and *P. pseudoalcaligenes* CECT 5344 itself (62.34%). By contrast, the average composition (%GC) of the *hmf* operon (BN5\_2298 to BN5\_2305) is 66.5%, close to the composition of *Cupriavidus basislensis* (65.3%) and other betaproteobacteria. The other two genes present in the island are BN5\_2306 (*psfD*) and BN5\_2307 (*araC*), whose %GC are 67.69% and 57.16%, respectively. It is also remarkable that all genes present in the *hmf* locus (Figure 1B) are singletons in relation to most Pseudomonaceae, except *psfD* and *benE*. In the evolution of prokaryotic metabolic networks and their regulation, the number of transcriptional regulators grows faster than the metabolic genes [49]. The horizontally transferred *hmf* pathway homologous to that described in *C. basilensis* HMF14 does not include its dedicated transcriptional regulator, but it seems that it has been acquired in a separate module from a second donor strain. Curiously, the regulator was originally in an inactive form, but evolved to the active form under selective pressure (Figure 4).

Even though *P. pseudoalcalignes* CEC T 5344 R1D assimilates furfural very e fficiently, it could not be useful for the pre-treatment of lignocellulosic residues because it simultaneously assimilates glucose. By contrast, the *edd*− mutant assimilated furfural leaving the glucose intact (Figure 6). This is an important feature in comparison to the equivalent mutant in *P. putida* KT2440, that accumulates 6-phosphoogluconate from glucose [40]. On the other hand, *P. pseudoalcaligens* could not assimilate HMF (Figure 3), the other main inhibitory component of hydrolysates. In fact, the genome analyses anticipated this result due to the absence of homologous genes to *hmfH* and *hmfFG* genes. These genes in *C. basilensis* code for the enzymes catalyzing the oxidation of HMFA to FDCA and the decarboxylation of the latter to generate furoic acid, respectively. FA is the metabolite in which converges the assimilation of HMF and furfural in *C. basilensis* HMF14 [24]. To circumvent the problem, it would be convenient to have bacteria capable of eliminating HMF. We have isolated, by selective enrichment from ashes, some bacterial strains belonging to the genus *Pseudomonas*, which are able to assimilate HMF in addition to furfural, furoic acid or furfuryl alcohol [50]. These abilities could be used in mixed cultures, provided that they do not assimilate sugars, or by constructing the required traits using the available genetic modules. Even though *P. pseudomonas* CECT 5344 R1D fully transformed HMF into two new compounds, 5-Hydroxymethylfuranoic acid (HMFA) and 2,5-furandicarboxylic acid (FDCA), both of which are versatile chemical intermediates of high industrial potential [51,52]. Therefore, the *edd*− mutant of the evolved R1D strain of *P. pseudoalcaligenes* described in this manuscript may increase the productivity of second generation bioethanol both by eliminating yeasts' inhibitory chemicals and by producing value-added chemicals from biomass.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2073-4425/10/7/499/s1, Figure S1: Schematic construction of PGEM-TE *edd aacC1* plasmid. Figure S2. Schematic construction of PGEM-TE *araC aacC1* plasmid.

**Author Contributions:** Conceptualization, R.B. and M.I.I.; methodology, R.B. and M.I.I.; investigation, D.M. and M.I.I., writing, R.B.; visualization, R.B., D.M. and M.I.I.; supervision, R.B. and M.I.I.; project administration, R.B.; funding acquisition, R.B.

**Funding:** This research was funded by IB16062, Junta de Extremadura (Consejería de Economía e Infraestructuras), GR18031, Fondo Europeo de Desarrollo Regional (FEDER), European Union. The work of Daniel Macias was supported by a fellowship from Universidad de Extremadura (Accion III-41, UEX 2010-2014).

**Acknowledgments:** The technical support of Gloria Gutiérrez and Gracia Becerra are gratefully acknowledged.

**Conflicts of Interest:** The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.
