Influence of furfural on the physiology of Acinetobacter baylyi ADP1

Abstract Pretreatment of lignocellulosic biomass produces growth inhibitory substances such as furfural which is toxic to microorganisms. Acinetobacter baylyi ADP1 cannot use furfural as a carbon source, instead it biotransforms this compound into difurfuryl ether using the reduced nicotinamide adenine dinucleotide (NADH)-dependent dehydrogenases AreB and FrmA during aerobic acetate catabolism. However, NADH consumption for furfural biotransformation compromises aerobic growth of A. baylyi ADP1. Depending on the growth phase, several genes related to acetate catabolism and oxidative phosphorylation changed their expression indicating that central metabolic pathways were affected by the presence of furfural. During the exponential growth phase, reactions involved in the formation of reduced nicotinamide adenine dinucleotide phosphate (NADPH) (icd gene) and NADH (sfcA gene) were preferred when furfural was present. Therefore a higher NADH and NADPH production might support furfural biotransformation and biomass production, respectively. In contrast, in the stationary growth phase genes of the glyoxylate shunt were overexpressed probably to save carbon compounds for biomass formation, and only NADH regeneration was appreciated. Finally, disruption of the frmA or areB gene in A. baylyi ADP1 led to a decrease in growth adaptation and in the capacity to biotransform furfural. The characterization of this physiological behavior clarifies the impact of furfural in Acinetobacter metabolism.


Introduction
The de v elopment of alternativ e ener gies to petr oleum is needed to avoid pollution and due to the e v entual exhaustion of this fossil resource (Isikgor andBecer 2015 , Zoghlami andPaës 2019 ).Lignocellulosic biomass can be an excellent option for it (Zoghlami and Paës 2019 ).Pentoses and hexoses contained in lignocellulosic biomass can be used as a carbon source for microorganisms to produce a variety of compounds such as biofuel, biochemicals and biomaterials (Isikgor and Becer 2015 ).Lignocellulosic biomass must be pretreated to release the fermentable sugars it contains.During one of the most common pr etr eatments, i.e. dilute acid hydr ol ysis at high temper atur es, aldehydes suc h as fur an-2-carbaldehyde (furfur al) and 5-(hydr oxymethyl)fur an-2carbaldeh yde (h ydroxymeth yl furfur al, HMF) ar e formed fr om xylose and glucose degr adation, r espectiv el y (P almqvist and Hahn-Hägerdal 2000a , Allen et al. 2010 ).These aldehydes inhibit the growth of those microorganisms emplo y ed in the fermentation of lignocellulosic biomass hydr ol ysates (P almqvist and Hahn-Hägerdal 2000b ).
Ov er all, aldehydes ar e known for their toxic effects on microorganisms by damaging proteins , lipids , and nucleic acids and hampering the production of targeted compounds (Zaldivar et al. 1999, Mills et al. 2009, J ay ak ody and Jin 2021 ).Fur an aldehydes can be found at concentrations ≤5 g L −1 in lignocellulosic hydrolysates (Mills et al. 2009 ).The mechanisms of furfural and HMF toxicity ar e mainl y by cell membr ane disruption, accum ulation of r eactiv e oxygen species, enzyme inhibition and dama ge, br eaks or m utations of deo xyribon ucleic acid (DNA), being furfural more toxic than HMF (Zaldivar et al. 1999, Mills et al. 2009, Allen et al. 2010, Willson et al. 2022 ).Furfur al also has a syner gistic toxic effect with other lignocellulosic-derived inhibitors such as acetic acid (Zaldivar andIngram 1999 , Mills et al. 2009 ).
In Esc heric hia coli , furfur al did not seem to cause membrane disruption despite its hydrophobicity (log P octanol/water = 0.41).Ho w e v er, deriv ativ es suc h as fur an-2-ylmethanol and fur an-2carboxylic acid did (Zaldivar et al. 1999, 2000, Mills et al. 2009 ).Membr ane dama ge was found in Sacc harom yces cerevisiae when cells were exposed to furfural (Allen et al. 2010 ).Regarding enzyme inhibition, se v er al authors hav e r eported that the competition for NAD(P)H cofactors by oxidoreductase enzymes used in furfural detoxification is detrimental to growth (Miller et al. 2009a, b , Arteaga et al. 2021 ).In addition, protein damage might be attributed to the reaction between their nucleophilic sites and the fur an aldehydes (Kur gan et al. 2019 ).Mor eov er, the inter action of furfural with duplex DNA causes single-strand breaks in sites with three or more thymine or adenine bases (Mills et al. 2009 ) while, in the plasmid pBR322, furfural produced DNA mutations (Mills et al. 2009 ).
Strategies to overcome the toxicity of furfural and HMF gener all y r el y on the detoxification of lignocellulosic biomass hydr ol ysates using chemical, physical, or biological methods (P almqvist andHahn-Hägerdal 2000a , Koopman et al. 2010 ).Among the biological methods, microbial degradation of furan aldehydes has been studied showing that redox reactions are the k e y points of the process (López et al. 2004, Wierckx et al. 2011 ).
The aerobic bacteria Cupriavidus basilensis HMF14 and Pseudomonas putida Fu1 can use furfural and HMF as carbon sources through the use of oxidoreductase enzymes (Koenig andAndreesen 1990 , Koopman et al. 2010 ).Under aerobic conditions, fur ans aldehydes ar e first oxidized to fur an-2-carboxylic acid and then metabolized to w ar ds 2-oxopentanedioic acid which feeds the tricarboxylic acid cycle (TCA) cycle to pr oduce ener gy and carbon building blocks (Trudgill 1969, Koenig and Andreesen 1990, Koopman et al. 2010, Nie v es et al. 2015 ).The use of furan aldehydes as carbon sources might be due to the presence of oxygende pendent o xidor eductases, whic h can limit the process in anaerobic micr oor ganisms (Koopman et al. 2010, Ran et al. 2014 ).Ho w e v er, it has been demonstrated that Desulfovibrio furfuralis , a strictl y anaer obic bacterium, can also use furfural as the sole carbon source yielding acetate.In this organism furfural is transformed into 5-oxo-4,5-dihydr ofur an-2-carboxylic acid, whic h can be hydr ol yzed or decarbo xylated to 4-o xobutanoic acid, transformed into acetyl-CoA and finally into acetate (Folkerts et al. 1989 ).
Model micr oor ganisms suc h as S. cerevisiae and E. coli , which do not have furan aldehyde o xidati ve degradation pathways and cannot use these compounds as carbon sources, are known to reduce furan aldehydes through the use of NAD(P)H-dependent oxidoreductases to furan alcohols, which are less toxic and not metabolized (Gutiérrez et al. 2002, Nieves et al. 2015 ).
Furthermore, it has been shown that several enteric bacteria such as E. coli , Enterobacter aerogenes , Citrobacter freundii , Klebsiella pneumoniae , Klebsiella oxytoca , Edwardsiella spp., Proteus vulgaris and Proteus mirabilis transform furfural and HMF into alcohols under aer obic and anaer obic conditions when carbon sources suc h as glucose, peptone and yeast extract are used (Boopathy et al. 1993, Gutiérrez et al. 2002 ).
In a pr e vious study has been r eported the use of Acinetobacter strains to reduce furfural into the less toxic compound difurfuryl ether when acetate was used as a carbon source.Acinetobacter baylyi ADP1 possesses two NAD(P)H-dependent alcohol dehydrogenases (EC 1.1.1.284,FrmA and EC 1.1.1.90,Ar eB) whic h ar e responsible for this biotransformation (Arteaga et al. 2021 ).
NADPH is the pr efer ential electr on donor used in the biosynthesis of most cellular components (Ying 2008 , Spaans et al. 2015 ) while NADH is mainly used as the electron donor in catabolic re-actions and for aerobic energy production (Ying 2008, Spaans et al. 2015 ).Se v er al authors hav e described that both NADH and NADPH are needed for furfural detoxification (Gutiérrez et al. 2006, Arteaga et al. 2021 ;Miller et al. 2009b ;Mills et al. 2009, Nie v es et al. 2015 ).Ther efor e, v ariations in the central carbon metabolism and o xidati ve phosphorylation might occur in the presence of furfural due to the competition for NAD(P)H cofactors used for the detoxification of this compound and the production of biomass and cellular energy.
For this reason, we evaluated the expression level of central carbon metabolism and o xidati ve phosphorylation genes to study the influence of furfural biotransformation on the physiology of A. baylyi ADP1.

Bacterial str ains, gro wth media, and cultiv a tion conditions
The strain A. baylyi ADP1 was kindly donated by Professor Veronique de Berardinis (Genoscope CNS, France).
Cultivations of ADP1 were performed at 30 • C in 500 mL quadruple baffled Erlenmeyer flasks containing 50 mL of M9 medium using 4 g L −1 acetate as the carbon sour ce.When needed, tw o pulses of 0.5 g furfural L −1 were added with a period of 30 min between them.Inocula and M9 medium were prepared as reported elsewhere (Sigala et al. 2017 ).
The A. baylyi ADP1 frmA and A. baylyi ADP1 areB mutant str ains wer e cultiv ated as follows.Pr einocula was performed in 15 mL conical centrifuge tubes at 30 • C in l ysogen y br oth (LB) medium added with 2 g glucose L −1 .Inocula were performed in 150 mL quadruple baffled Erlenmeyer flasks at 30 • C either in LB broth added with 2 g glucose L −1 or in M9 medium containing 3 g acetate L −1 .Chloramphenicol was added at a final concentration of 20 μg mL −1 in m utants pr einocula and inocula.Cultiv ations were performed at 30 • C in 250 mL quadruple baffled Erlenmeyer flasks containing 25 mL of either LB broth with 1 g glucose L −1 or M9 medium with 2 g acetate L −1 and 0.6 g furfural L −1 .
In all the cultiv ations, two tec hnical r eplicates fr om two independent biological experiments were performed.Additionally, for the quantification of cofactors, an unpaired t -test was performed, with a significant difference P < .05.

Analytical methods
Bacterial gro wth w as monitor ed spectr ophotometricall y at 600 nm using an Eppendorf BioPhotometer TM .The OD 600nm was converted to dry cellular w eight accor ding to the correlation 1 OD 600nm = 0.55 g DCW L −1 for A. baylyi ADP1 as described (Sigala et al. 2019 ).
Cofactors quantification was performed with the nicotinamide adenine dinucleotide phosphate / reduced nicotinamide adenine dinucleotide phosphate (N ADP/N ADPH) (Cat.No. MAK038) and nicotinamide adenine dinucleotide / reduced nicotinamide adenine dinucleotide (N AD/N ADH) (Cat.No. MAK037) kits from Sigma-Aldrich (MO, USA) following the manufacturer instructions, sampling at exponential and stationary growth phase 10 min before and after two pulses of 0.5 g furfural L −1 were added with a period of 30 min between them.
Furfural was spectrophotometrically determined at 520 nm based on a colorimetric reaction with aniline and chlorane.For this, 500 μL of cultur e wer e collected and completed to 2 mL with fr esh cultur e medium and sonicated for 4 cycles of 15 s on and 15 s off in a probe tip sonicator.A volume of 2 mL of ethanol was then added, mixed b y v ortexing and sonicated into an ultrasonic bath for 5 min.Samples were centrifuged at 10 000 rpm for 5 min.A volume of 1 mL of the obtained supernatant was added to a glass cuvette, follo w ed b y 20 μL of aniline and 10 μL of c hlor ane, mixing b y v ortex, and incubation at r oom temper atur e for 20 min after which the absorbance at 520 nm was determined.

Ribonucleic acid (RNA) isolation and purification
Cell samples for RNA isolation were collected from cultures in exponential and stationary phases 10 min before and after two pulses of 0.5 g furfural L −1 .Cell lysis and total RNA isolation and purification were carried out using the RNeasy Mini Kit according to the manufacturer's recommendations (Qiagen, Hilden, Germany).All RNA samples were subjected to DNase treatment using the TURBO DNA-free kit (Ambion, MA, USA).The RNA concentration was determined by UV spectrophotometry in a Nanodrop 2000 spectrophotometer (Thermo Fisher Scientific Inc, MA, USA).The absence of DNA contamination in the RN A samples w as confirmed by pol ymer ase c hain r eaction (PCR) amplification of specific control genes.RNA integrity was e v aluated by electr ophor etic separation on a microfluidic chip in a 2100 Bioanalyzer using RNA 6000 Nano Reagents Part I and RNA Nano Chips (Agilent, CA, USA).The RNA integrity number (RIN), which gives a numerical assessment of RN A integrity, w as automatically calculated by the included 2100 Expert Software (Agilent, CA, USA).All samples had RIN values > 8.

cDNA synthesis and transcriptional analysis
The Re v ertAid H Minus First Str and cDN A Synthesis Kit w as used to synthesize cDNA (Thermo Fisher Scientific Inc, MA, USA).For eac h r eaction, a ppr oximatel y 5 μg of RNA and a mixtur e of 1 × 10 −5 M DNA r e v erse primers ( Supplementary Table 1 ) specific for each analyzed gene were used.The secA gene was chosen as the r efer ence gene because its Cq pr acticall y r emained constant under all the tested conditions.Re v erse tr anscriptionquantitativ e pol ymer ase c hain r eaction (R T-qPCR) w as performed in a 7500 Real-Time PCR System (Applied Biosystems , C A, USA) using a SYBR Green/R O X qPCR Master Mix kit (Thermo Fisher Scientific Inc, MA, USA).The size of all amplimers was 101 bp.For each gene, all experiments were performed in duplicate from two differ ent cultiv ations, and v ery similar v alues wer e obtained.A nontemplate contr ol r eaction mixtur e was included for each gene.The 2 − Cq method was used to analyze the data (Schmittgen and Livak 2001 ).For downregulation, negative values were obtained by 1/2 − Cq for each condition.All R T-qPCR experiments w ere compliant with the minimum information for publication of quantitativ e r eal-time PCR experiments (MIQE) guidelines (Bustin et al. 2009 ).For each gene analyzed after furfural pulses, the transcriptional le v el of the same gene anal yzed without furfur al pulses was considered equal to one and was used as a control to normalize the data.

Inactiv a tion of frmA and areB genes in A. baylyi ADP1
Mutant str ains A. bayl yi ADP1 frmA and A. bayl yi ADP1 areB wer e gener ated based on the general methodology described (de Berardinis et al. 2008 ) ( Supplementary Fig. 1 ).To confirm the gene interruptions, c hr omosomal DN A w as extr acted fr om cultur es of A. baylyi ADP1 frmA and A. baylyi ADP1 areB and used as a template in specific PCR reactions using the primers described in Supplementary Table 2 .Single-gene knoc k out m utants wer e con-F igure 1. Gro wth profile of Acinetobacter baylyi ADP1 in minimal medium with acetate as carbon source and two pulses of furfural (arrows) either in exponential (red, circles) or in stationary growth phase (blue, squar es).A contr ol condition cultiv ation without furfur al is also shown (black, triangles).Error bars represent the experimental error from two independent biological experiments.
firmed by observing the correct size of the two expected PCR products in 1% a gar ose gels ( Supplementary Fig. 2 ).

Growth of A. baylyi ADP1 in minimal medium with furfural
T he beha vior of A. baylyi ADP1 after pulses of furfural at exponential and stationary growth phase is shown in Fig. 1 .Furfur al negativ el y impacted gr owth because after its addition the biomass clearly decreased.This could be due to NADH competition (Arteaga et al. 2021 ) and/or metabolic enzyme damage (Mills et al. 2009 ).Ho w e v er, after a cell ada ptation, gr o wth w as recovered.This phenomenon occurred when pulses of furfural were added at exponential or stationary growth phases (Fig. 1 ).

Impact of furfural addition in the transcriptional levels of A. baylyi ADP1 genes during the exponential growth phase
The expr ession le v els of genes involv ed in acetate metabolism and o xidati ve phosphorylation in A. baylyi ADP1 were evaluated after furfur al exposur e (figs 2 A, 3 A).Some of the enzymes of these genes r equir e or pr oduce r edox cofactors (N AD + /N ADP + ).R T-qPCR assay sho w ed that the acetate transporter gene actP was underexpressed after furfural addition, suggesting a reduction in acetate transport (Fig. 2 A).This could be associated with one of the mechanisms of furfural toxicity which is cell membrane disruption (Mills et al. 2009 ).The ack and pta genes were also underexpressed indicating a possible limitation of the Pta-Ack pathway that transforms acetate into acetyl-CoA leading to a decrease in the assimilation of acetate since the Acs pathway was not affected (Fig. 2 A).
Under aerobic conditions, the TCA is responsible for the oxidation of acetyl-CoA resulting in the production of intermediates for amino acids , nucleotides , and cofactors synthesis (Han et al. 2008, Kwong et al. 2017 ).Regarding the TCA, acnA , icd , and fumC wer e ov er expr essed while aceA fr om the gl y oxylate shunt w as not.Under these conditions , TC A is likely to be preferred over the glyoxylate pathway to generate NADPH for anabolic reactions and NADH for o xidati ve phosphorylation.
Regarding ana pler otic r eactions, the under expr ession of pc kG, whose pr oduct tr ansforms oxaloacetate and guanosine triphosphate (GTP) into phosphoenolpyruvate and CO 2 , suggests a flow Figure 2. Acetate catabolism of Acinetobacter baylyi ADP1 in (A) exponential and (B) stationary growth phase with furfural.Key pathwa ys , metabolites and genes are highlighted.Green, red, and black arrows represent underexpression, overexpression, and no difference of the involved genes, r espectiv el y, when furfur al is pr esent in comparison with the non-furfur al condition.
pr efer ence to w ar ds the formation of phosphoenolp yruvate via malate and pyruvate with malic enzymes.An increase in the expression of sfcA instead of maeB demonstrates the priority of gener ating NADH ov er NADPH, r espectiv el y, when conv erting malate to pyruvate.In this case, the necessity of NADPH seems to be supplied by the ov er expr ession of icd (Fig. 2 A).Thus, the NADH generated by the TCA, the malate dehydrogenase (EC 1.1.1.38,SfcA) and the tr anshydr ogenase genes pntA-1 and pntA-2 , whic h ar e known to increase their expression when furfural is present (Arteaga et al. 2021 ), seems to be sufficient for o xidati ve phosphorylation and, at the same time, for the biotransformation of the two pulses of 0.5 g furfural L −1 added in this study.
The ndh gene sho w ed an increase in its expression level (Fig. 3 A).This gene encodes the enzyme NADH: quinone reductase (EC 1.6.5.9, Ndh) which is part of the first complex of the electron tr ansport c hain in o xidati ve phosphorylation.T he o v er expr ession of ndh ma y ha ve the purpose of capturing NADH more efficiently given the competition with the enzymes FrmA and AreB which r equir e NADH for furfur al r eduction.As expected, the genes of these last two enzymes also showed an increase in their expression, mainly areB .This is consistent with previous studies (Arteaga et al. 2021 ).Finally, the o xidati ve phosphorylation genes atpB , cydA , c ydB , and nuoB w er e under expr essed, so it could demonstr ate that the process is not well-performed in the presence of furfural.

Impact of furfural addition in the transcriptional levels of A. baylyi ADP1 genes during the sta tionary gro wth phase
All the cultivations reached the stationary phase upon exhaustion of acetate (data not sho wn).Ho w e v er, the pr esence of furfural caused a decrease in the expression of actP , but not of the pta -ackA genes (Fig. 2 B).Interestingly, a decreased expression of icd suggests the pr efer ence for the glyoxylate shunt altogether with the ov er expr ession of glcB and acnB (Fig. 2 B).Furthermore, the expression of aceA did not change compared to what occurred in the exponential phase.Under this scenario, the decarboxylations of the lo w er TCA r eactions ar e pr e v ented at some le v el in favor of the glyoxylate shunt to save some carbon but with the concomitant decrease in NADPH and NADH.In the early stationary growth phase, biomass is generated at the same rate that it is lysed because the surviving cells use cell debris as a substrate (Navarro Llorens et al. 2010 ).
In the case of acnB , a study has shown that furfural can increase the expression level of this gene in E. coli to impr ov e metabolic activity (Miller et al. 2009a ).Based on this information, we can deduce that an increment in the expression level of acnA and acnB in the exponential and stationary phases, r espectiv el y, can be attributed in part to the pr esence of furfur al. Ho w e v er, it m ust be demonstrated if the effect of furfural over the expression levels would be direct or indirect through an intermediary which causes a gener alized r esponse to str ess as it has been demonstrated in Zymomonas mobilis where transcriptional regulators and universal stress genes sho w ed higher expression in presence of furfural (He et al. 2012 ).
As in the exponential phase , sfcA was o v er expr essed and confirms its role as a generator of NADH via the malate-pyruvatephosphoenolpyruv ate r oute to support furfur al biotr ansforma-tion.The increase in the expression level of ppsA gene demonstr ates the pr eferr ed formation of phosphoenolpyruv ate fr om pyruvate instead of using the phosphoenolpyruvate carboxykinase (PckG) pathway from oxaloacetate, a phenomenon that was also observed during the exponential growth phase (Fig. 2 A).
The genes frmA and areB wer e also ov er expr essed and their pr oducts ar e still r esponsible for furfur al biotr ansformation at stationary growth phase.It is interesting to note that the overexpr ession le v el of areB was decreased while that of frmA was increased in the stationary compared to the exponential growth phase.
The o xidati ve phosphorylation genes atpB, nuoA, cydA , and cydB wer e under expr essed (Fig. 3 B) in the presence of furfural, and the r espir ation r ate decr eased in this gr owth phase (Riedel et al. 2013 ).This could be also an indirect effect of the diminished levels of N ADH caused b y the use of this cofactor by the AreB and FrmA enzymes, whose genes were still highly overexpressed (Fig. 3 B).Additionally, ndh was not overexpressed as in the case of the exponential growth phase, probably because there is no longer a strong NADH competition.
Inter estingl y, compar ed to the exponential growth phase fewer genes were affected in their expression levels during the stationary growth phase, which is known to have lo w er metabolic activity and growth rate (Jaishankar and Sriv astav a 2017 ).Consequently, in this growth phase, more NAD(P)H could be available for furfur al biotr ansformation in order to decr ease to xicity.In ad dition, whether in the exponential or stationary growth phase, other genes of the studied pathwa ys , whic h r epr esent 54% of the total, did not show changes in their expression levels when two pulses of 0.5 g furfural L −1 were added ( Supplementary Table 3 ).
This implies that the effect of furfural on the expression of central and energy metabolism genes is specific and is mainly related to o xidati ve metabolism.
The complete set of genes analyzed by RT-qPCR in exponential and stationary growth phases is shown in the supplementary material .

Quantification of cofactors in the exponential and stationary growth phases
The N AD + /N ADH and N ADP + /N ADPH r atio offers a perspectiv e on the metabolic activity of the cell.Experiments with Bacillus subtilis in the presence of colistin sho w ed that an increase in the N AD + /N ADH r atio stim ulated the conv ersion of N ADH to N AD + (Yu et al. 2019 ).On the contrary, in this work, the presence of furfural sho w ed a decrease in the ratio of both cofactors (Fig. 4 A), which suggests the conversion of NAD + to NADH and NADP + to NADPH for furfural reduction in the exponential growth phase.A similar behavior has been shown in C. glutamicum and C. tropicalis (Tsuge et al. 2014, Wang et al. 2016 ).Ho w e v er, some micr oorganisms such as C. glutamicum and S. cerevisiae are also able to oxidize furfural to 2-furoic acid and consequently, a decrease in reduced cofactors might be attributed due to the changes in the cofactors proportion (Ask et al. 2013, Tsuge et al. 2014 ).Moreover, it is not clear what the expression level of central carbon metabolism genes was in these micr oor ganisms with furfural and carbon sources to compare with A. baylyi ADP1.Overexpression of sfcA and icd in A. baylyi ADP1 could k ee p the a ppr opriate le v el of reduced cofactors and avoid their exhaustion.This probably makes a difference between the behavior of Acinetobacter and other model micr oor ganisms that ar e dr ained in NAD(P)H when furfural is present.Figure 4 B shows the relative level of NADH and NADPH in the cell before and after the addition of furfural.The NADH level slightl y incr eased after the addition of furfur al whic h coincided at least with the ov er expr ession of sfcA whose product has NADH as a cofactor like the rest of the reactions that generate this cofactor in the TCA (Fig. 2 A).A similar trend is seen with NADPH which is the produced cofactor of isocitrate dehydrogenase (EC 1.1.1.42,Icd)whose gene was also ov er expr essed (Fig. 3 A).The increase in the le v el of NADH confirmed its demand by FrmA and Ar eB whic h ar e r esponsible of furfur al r eduction.Ther e was no significativ e difference in the levels of NAD + and NADP + (Fig. 4 B).
The N AD + /N ADH and N ADP + /N ADPH ratios did not sho w significativ e differ ences in the stationary gr owth phase compar ed to the exponential one (Fig. 5 A).Ho w e v er, a do wnw ar d trend w as observed in the N AD + /N ADH ratio, which w ould benefit the generation of the NADH r equir ed for the biotransformation of furfural.By contrast, the N ADP + /N ADPH ratio sho w ed a slightly upw ar d tr end, minimall y favoring the generation of NADP + .
Figure 5 B shows the level of NADH and NADPH present in the cell before and after the addition of furfural in the stationary growth phase.In this case, no changes in their le v els wer e observed.After the addition of furfural, the level of NAD + had a clear decrease (Fig. 5 B).As in the exponential phase of growth, a constant r egener ation of the r educed form would support the detoxification of furfur al. Pr e vious studies sho w ed that, in the presence of furfural, the transhydrogenase genes pntA-1 and pntA-2 increased their expression during the stationary growth phase helping furfural reduction (Arteaga et al. 2021 ).

Inactiv a tion of frmA and areB in A. baylyi ADP1
The growth of A. baylyi ADP1 frmA and A. baylyi ADP1 areB was e v aluated on M9 medium with furfur al and acetate as carbon source (Fig. 6 A).Compared to the wild-type strain, the ADP1 m utant str ains did not gr ow after 6 h.In contrast, during this time the wild-type str ain r eac hed the stationary growth phase and furfural was completely biotransformed.By the end of the experiment, A. baylyi ADP1 frmA and A. baylyi ADP1 areB only biotransformed 17% and 11%, respectively, of the initial furfural (Fig. 6 A).
The mutant strains sho w ed a clear deficiency in their ability to r educe furfur al .Additionall y, when ADP1 and m utants wer e cultivated in M9 acetate without furfural, the specific gr owth r ate ( μ) of mutants was affected significantly, decreasing 36% and 26% in frmA and areB , r espectiv el y, in comparison to ADP1 (data not shown).These results suggest FrmA and AreB activities are important for growth and adaptation.
The main consequence of disrupting the frmA or areB genes was to extend the adaptation phase of bacterial growth in a mineral medium.Both genes are necessary to have a short adaptation phase and to r eac h maxim um biomass in short times to be able to biotransform furfural.
Figure 6 B also shows the growth behavior of mutant strains in comparison with the wild type but growing on LB media with glucose and furfural.An affectation was observed in the adaptation time of the mutants.In this rich medium, the mutants could gr ow and r eac h the stationary phase, although not as fast as the wild-type str ain. A. bayl yi ADP1 and A. bayl yi ADP1 frmA specific gr owth r ates ( μ) wer e 0.70 h −1 while that of A. bayl yi ADP1 areB was 0.68 h −1 (a 3% decrease).
Unlike with cultures in the M9 medium, after 6 h in the LB medium, the thr ee str ains biotr ansformed furfur al completel y despite the fact mutant strains have a longer lag phase (Fig. 6 B).Enric hed media, suc h as LB medium, contain a series of non-defined elements and a wide variety of carbon and nitrogen sources (Sezonov et al. 2007 , Kim andKim 2017 ), as well as a large amount of pr e-elabor ated building bloc ks for the formation of macromolecules (Tao et al. 1999 ), which favors the growth in the mutant strains without compromising energy and reducing po w er, and compensating the lack of frmA or areB .In all the above cases, a control sample without bacterial inoculum and under the same experimental conditions confirmed that furfur al r emains intact without the presence of Acinetobacter strains after incubation (data not shown).
The presence of furfural causes k e y changes in the central carbon metabolism of A. baylyi ADP1.During the exponential gr owth phase, furfur al favor ed the full TC A pathwa y from Icd over the glyoxylate shunt, with the consequent preferential formation of NADPH and NADH.On the other hand, during the stationary growth phase, the gly oxylate pathw ay seems to r ecov er its activity.In both gr owth phases, furfur al led to the production of N ADH b y the malic enzyme SfcA and the malate-p yruvatephosphoenolp yruvate pathw ay w as pr eferr ed.A decr ease in the expr ession le v els of most o xidati ve phosphorylation genes when furfural is present probably decreased the production of energy.Also, cofactors analysis suggests a preference for NADH over NADPH production in the cell as a result of the furfural detoxification by the activity of FrmA and Ar eB.Finall y, the disruption of frmA or areB in A. baylyi ADP1 proved the direct activity of FrmA and AreB in the biotransformation of furfural in this strain.

F
igure 4. (A) N AD + /N ADH and N ADP + /N ADPH r atios and (B) r elativ e quantification of reduced and oxidized cofactors in the exponential growth phase of Acinetobacter baylyi ADP1.Dark and light bars represent the condition before and after the addition of furfural, respectively.Significant difference ( * ).Error bars represent the standard error of the mean (SEM) from two independent biological experiments.P < .05,degrees of freedom = 2.

F
igure 5. (A) N AD + /N ADH and N ADP + /N ADPH r atios and (B) r elativ e quantification of reduced and oxidized cofactors in the stationary growth phase of A. baylyi ADP1.Dark and light bars represent the condition before and after the addition of furfural, respectively.Significant difference ( * ).Error bars represent the standard error of the mean (SEM) from two independent biological experiments.P < .05,degrees of freedom = 2. F igure 6. Gro wth profile (continuous lines) and remaining furfural (discontinuous lines) of Acinetobacter baylyi ADP1 (blue, circles), A. baylyi ADP1 frmA (r ed, squar es) and A. baylyi ADP1 areB (green, triangles) in (A) minimal medium with acetate and (B) LB-medium with glucose.0.6 g furfural L −1 = 100%.Error bars represent the experimental error from two independent biological experiments.