www.nature.com/scientificreports OPEN Evidence of mixotrophic carboncapture by n-butanol-producer Clostridium beijerinckii Received: 13 September 2016 Accepted: 13 September 2017 Published: xx xx xxxx W. J. Sandoval-Espinola1,4, M. S. Chinn2, M. R. Thon3 & J. M. Bruno-Bárcena1 Recent efforts to combat increasing greenhouse gas emissions include their capture into advanced biofuels, such as butanol. Traditionally, biobutanol research has been centered solely on its generation from sugars. Our results show partial re-assimilation of CO2 and H2 by n-butanol-producer C. beijerinckii. This was detected as synchronous CO2/H2 oscillations by direct (real-time) monitoring of their fermentation gasses. Additional functional analysis demonstrated increased total carbon recovery above heterotrophic values associated to mixotrophic assimilation of synthesis gas (H2, CO2 and CO). This was further confirmed using 13C-Tracer experiments feeding 13CO2 and measuring the resulting labeled products. Genome- and transcriptome-wide analysis revealed transcription of key C-1 capture and additional energy conservation genes, including partial Wood-Ljungdahl and complete reversed pyruvate ferredoxin oxidoreductase / pyruvate-formate-lyase-dependent (rPFOR/Pfl) pathways. Therefore, this report provides direct genetic and physiological evidences of mixotrophic inorganic carbon-capture by C. beijerinckii. Current societal efforts require solutions to address increasing greenhouse gas emissions1. Accordingly, carbon-capture and its biotransformation into useful value-added commodities, including biofuels, has become a growing area of research. Thanks to a better understanding of pathways such as the Wood-Ljungdahl (WL), and the recently described reversed-pyruvate ferredoxin oxidoreductase (rPFOR)/pyruvate-formate-lyase-dependent (Pfl) carbon assimilation2, more microorganisms can potentially be screened for presence of these pathways using physiological signals. These pathways allow the incorporation of one-carbon (C-1) molecules into Acetyl-CoA. This is a key molecule that can be subsequently bio-transformed into value-added molecules such as acetate or ethanol3–7. Currently, butanol is considered one of the ideal advanced renewable fuel due to a number of favorable properties and applications8–10. For example, it can be used unblended in unmodified car engines and is compat- siabtnaldegewCsiOtohf2)dciuenvrtreoelonbptumotaielnniontl5f,r,6ba,1ys2t,1nr3ua. ctAutusraraelr1o1er.sHugleotn,weseetivecekaril,nloygnmtloyoardecichfieieendvtlemyctiohcsreto-abcseossimm, hpialeasttibitoievneenobfausstysaennstosheledps,irsaongdaduscr(tecimoonna,tianmisnoiisnntgtrheHese2e,aaCrrcOlhy has focused on assessing the heterotrophic biotransformation of renewable feedstock by traditional solventogenic Clostridia. However, heterotrophic fermentations have the inherent limitation that 1/3 of carbon is lost in the finorgmtoowfaCrdOs2.oIvnetrelroeosktiendglmy,etthaebroelipcocratepdabdialittaiessh9o,1w4–s17s.iWgniitfhictahnitsvianrimabinildit,ywine apparent final product yields, pointperformed a deeper examination of the evolving fermentation gases as physiological signals, while assessing the assimilation of synthesis gas by the natural n-butanol producer C. beijerinckii. Results fRoermale-dtiamseeri(eisno-flifende-)baftecrhmfeermnteanttaiotionngs aofsCm. boeinjeirtionrckiniigwhreilveemaolnsitCoOrin2ga, nindreHal2-toimscei(lilna-tliionen)s, .th  e We perevolving leantdeologge-npohuassgeaasnsedst.hInetoenressettinogf lsyo, lwveenotbosgeernveesdisin(w-phheanseH, 2syanncdhCroOn2oruesaHch2eadn≈d 3C%O[2vo/svc]i)ll(aFtiiog.n 1s)c.oTihnecsideitnygpewsitohf 1Department of Plant and Microbial Biology, North Carolina State University, Raleigh, NC, USA. 2Department of Biological and Agricultural Engineering, North Carolina State University, Raleigh, NC, USA. 3Instituto Hispano-Luso de Investigaciones Agrarias (CIALE), Department of Microbiology and Genetics, University of Salamanca, Calle Del Duero 12, Villamayor, 37185, Spain. 4Present address: Department of Chemistry & Chemical Biology, Harvard University, 12 Oxford Street, Cambridge, MA, 02138, USA. Correspondence and requests for materials should be addressed to J.M.B.-B. (email: jbbarcen@ncsu.edu) SCIENtIfIC REPorTS | 7: 12759 | DOI:10.1038/s41598-017-12962-8 1 www.nature.com/scientificreports/ Figure 1.  Direct monitoring of hydrogen and carbon dioxide evolution in the gas-phase during fed-batch fBeiromsteant tBa+tiornesacbtyoCr u. bsienijgerdinefcikniei.dHm2 eadnidumCO202ceovnotlauintiionngf6r%om(wth/vre)esuincdroespeeansdleinmtietixnpgecriamrbeonntsapnedrfeonremrgeyd in a source with an initial and final volumes of 1000 mL and 1400 mL, respectively. Feeds (400 mL) contained 80 g sucrose fed at 0.08 mL/h, to reach a final concentration of 100 g/L (w/v) along with: Red line: only sugar was added; wCBlOiathc2konlsiitcnrieoll:gafetrineosngh.aFws ewhroamsleeanmcthaeitdeiviouenmds uwasneirndeg; cBmolunaetsrsloifnllloeew:d2acXto2tnr5tar0co relplcemorsm,.3Op7ou °ntCpeunanttsgd.aYpseH-lploh6wa.5sbeaoncxodemcsopsnhosostiwatinodtnelytwasipalsaorcfgotehndeti(nH1u22o.a4un8sd lLy/h) monitored and recorded using two analyzers: An EasyLine continuous analyzer, model EL3020 (ABB, Germany) and a Pfeiffer OmniStar quadrupole mass spectrometer. fluctuations are normally observed in feedback-loop controls as a response to metabolic pathway changes18,19. Indeed, diauxic growth accumulation (Fig. S1). Awlaths oeuvigdhewnteftreosmtedththerdeeecfreeeadsecoinmtphoessitpioecnisf,ic(ig.er.oawlltwhhrailteecaosnHta2inainndgCadOd2itrieosnuaml seudctrhoeseir: (i) fresh complete medium; (ii) 2X trace components20; (iii) sucrose only) these had no effect on the gas oscilla- tions. This suggested to us that neither organic carbon nor another medium component modulated this pheno- type. Furthermore, the cells failed to utilize all the provided sugar, and its utilization varied significantly among experiments (Fig. S2, D, D1). In contrast, the kinetic and yield parameters for products and biomass did not vary significantly (Fig. S2–A1, B1, and C1; and Table S1). We used nitrogen (F = 12.48 L/h) as carrier/stripping for gas measurements. reached 48 and In batch (i.e. closed-system) without 23%, respectively (not shown). stripping CO2 and H2 accumulated in the head-space and It was previously shown taining anoxic conditions) atlhloatwrsecfoirrcmuloatriensguegnadrocgoennsouums pHti2oanndanCdOa2ciddugrienngerbauttiaonno14l.fAerdmdeitniotantaiollny,(ifnocrrmeaasiend- product biosynthesis has been shown when electrochemical bioreactors were cultured with C. acetobutylicum in complex balance, hmeelpdiinugmtoalsounsgtawinitshuCbsOtr2a2t1e. Iunptetarekset2i2n.gNlye,vtehrethCe-l1esass,sCim-1ilaastsioimn iilnatbioanctheraisanisoat lbseoeanmpreecvhiaonuisslmy dfoesrcrreidboedx in Clostridium beijerinckii6,7. Genomic and indirect transcriptomic analysis indicates that C. beijerinckii has the genetic potential for C-1 assimilation.  To explain the gas oscillations and their potential assimilation, we explored the C. beijerinckii genome searching for genes related to C-1 assimilation, such as those associated to the WL or rPFOR/Pfl pathways2,5,7,23. We found open reading frames that putatively code for CO dehydrogenase (CODH) (Cbei_5054 and Cbei_3020), formate dehydrogenase and accessory genes (Cbei_3798 to Cbei_3801), formyl-THF ligase (Cbei_0101), methylene-THF dehydrogenase/cyclohydrolase (Cbei_1702) and methylene-THF reductase (Cbei_1828). The putative proteins encoded by these genes have high sequence identity to those of Clostridium ljungdahlii, the species most often utilized for ethanol generation from synthesis gas5,24 (Fig. 2A–B). However, as opposed to this species, C. beijerinckii does not contain annotated a gene coding specifically for an acetyl-CoA synthase, which is essential within the WL pathway. Furthermore, in C. ljungdahlii, most of the WL pathway genes are clustered, except a gene coding for a Ni-Fe-S containing CODH (CLJU_c17910), and a formate dehydrogenase (CLJU_c08930). The former is the main enzyme within the carbonyl branch of the WL pathway, and the final step of the methyl branch (or initial step, if CO is supplied). The latter initiates SCIENtIfIC REPorTS | 7: 12759 | DOI:10.1038/s41598-017-12962-8 2 www.nature.com/scientificreports/ Figure 2.  Partial Wood-Ljungdahl (WL) and reverse pyruvate: ferredoxin oxidoreductase/pyruvate-formate lyase (rPFOR/Pfl) pathways scheme in C. beijerinckii and chromosome localization of corresponding genes. (A) C-1 assimilation genes found in C. beijerinckii and C. ljungdahlii (WL pathway), mapped in their respective chromosomes. (B) Presumed partial WL and rPFOR/Pfl scheme pathways in C. beijerinckii. Red arrows indicate reactions predicted to be catalyzed by the CO dehydrogenase/Acetyl-CoA synthase complex, (acetyl-CoA is not coded in C. beijerinckii genome). Gray and blue arrows indicate reactions belonging to the methyl branch of the WL, and rPFOR/Pfl evolving) reactions. (C) pathways, respectively. Genes and their names Black and yellow arrows indicate associated to C-1 assimilation in CC.Ob2eaijnerdinHck2 iais. similation (and athneamceettyhl-yCl borAanscyhn,thaallsoew. CinognCveOr2seclayp, tCu.rbeeiinjetroinfocrkmii actoen. tTaoingsetthheer,htohmeyolloeagdoutos the generation genes scattered of acetyl-CoA via through its chro- mosome (Fig. 2A). Interestingly, the annotated CODH and the formate dehydrogenase from C. beijerinckii have 77.62 and 72.23% sequence identity, respectively, to the corresponding genes of C. ljungdahlii localized outside its WL cluster. In addition to these genes, C. beijerinckii contains Fe-only and NiFe-hydrogenases (Cbei_1773, Cbei_3796, Cbei_4110 and Cbei_3013) with similarities to those of C. ljungdahlii. In this species, along with HC-21g-efinxeartiaotniopna,tthhwesaey5e.nzymes have hydrogen uptake capabilities, providing extra reducing equivalents to its SCIENtIfIC REPorTS | 7: 12759 | DOI:10.1038/s41598-017-12962-8 3 www.nature.com/scientificreports/ Figure 3.  Time-course transcription profile of C-1 assimilation and energy conservation genes in C. beijerinckii. Time-course expression profiles of partial Wood-Ljungdahl and pyruvate: ferredoxin oxidoreductase/pyruvate-formate lyase (rPFOR/Pfl) predicted genes, along with genes related to energy conservation in C. beijerinckii. The FPKM (fragments per kilobase per million) were calculated from publicly available RNA-seq defined medium20, data27. 37 °C, 2L5in0 erspmrep, arensdencot nCsOta2natnlydsHpa2regveodluwtiitohnpiunrtehneigtraosg-pehna(s1e2o.4f8C L./bhe)i.jerinckii growing in C. beijerinckii also contains two putative Pfl-coding genes (Cbs_1009 and Cbs_1011), (both annotated as formate acetyltransferase, as is the case in Clostridium thermocellum [pflB, clo1313_1717)])2, and a putative pyru- vate formate-lyase activating enzyme gene (Cbs_1010). The proteins coded by Cbs_1009 and Cbss_1011, and Cbs_1010, have ~63.5 and 44.4% sequence identity to those of C. thermocellum, respectively. This bacterium, while relying on a partial WL pathway (i.e. the methyl branch without the formate dehydrogenase), contains a reverse pyruvate ferredoxin oxidoreductase generate pyruvate, which is then transformed (incltoo1f3o1rm3_a0t6e7a3ndanadceotythl-eCros)AthbaytPcfol2m. Ibnitnereessaticnegtylyl,-tChoeAC.abnedijeCrOin2cktioi pyruvate ferredoxin oxidoreductase (PFOR) (Cbs_4318) has 64.1% sequence identity to C. thermocellum rPFOR. The reverse reaction of PFOR has also been observed in other acetogenic and methanogenic bacteria, where this enzyme links the WL pathway and glycolysis25. Additional genes related to the rPFOR/Pfl pathway are a serine hydroxymethyltransferase and a methionine synthase, both of which are also encoded in C. beijerinckii chromo- some (Cbs_1868, and Cbs_3100, Cbs_2329 and Cbs_1401, respectively). With these C-1 assimilation genes in focus, we performed an analysis of publicly available transcriptomic data from batch cultures of C. beijerinckii26. An RNA-seq time-course experiment was previously reported by Wang et al.27 using cells growing in P2 medium sparged with pure nitrogen. After quality trimming and normalization for gene length and number of assembled reads, we found the putative genes required for C1-assimilation being expressed, either constitutively (Cbei_5054, Cbei_1828 and Cbei_4318) or differentially over time (Cbei_1702, Cbei_0101, Cbei_3801, Cbei_3794, Cbei_3798, Cbei_3799, Cbei_3800, Cbei_3020, Cbei_1010 and Cbei_1011) (Figs 3 and S3). After mapping this transcriptomic response to our gas oscillation data, we identified expression changes that coincided with this phenotype, indirectly pointing towards the re-assimilation of these gases (Figs 1 and 3). Among these genes, formate dehydrogenase and its accessory genes showed the lowest expression in ihtshoemweaevxvaielmru, aaitnle(cdFreiegax.sp i3ne)rg. iTmtohewensaetradgl escnomensiddbi-telioloognn-gp(ih.teoa.stleoh,wecombnieoctumhryarlsi-nsbgraanwndicthCh Oothf2e/thHtei2mWceo-Lnp,coaeinnndttrarwPthiFoeOnre,RaC/nPOdfl2nCaitn1r-doagHsesni2maacitlcmautoimosnuphlpaetaritoeh)n-; ways. Among the four annotated hydrogenase genes, two displayed the highest expression levels: the correlation between expression and hydrogen oscillation/evolution suggested that Cbei_3013 and Cbei_1773 are primar- aailnvyahuiylsadebrdialsifteoysr2a8H.lsT2ohusiphstoianwk-eesdialiencxodparHnes2asleiyovsnoisl(upCtribooenvii_,d4ree4ds2p5ae/ncCtibinvedeii_lry1e.0cA3td1od)v.ieTtrihoviniseawelnlyoz,yftmthheeegagelalnosewosssecfnoilrclaoetdvieoinnngsmfaootrrtehpieuntttarrataincveselclcruialpartbrooCmnOiicc2 level, while suggesting that these gases may regulate the C1-assimilatory phenotype in C. beijerinckii. Similar gas-dependent behavior has been previously observed in cultures of acetogens Clostridium thermoautotrophicum (reclassified as Moorella thermoautotrophica)29 and C. ljungdahlii, which contain a complete WL pathway30,31. Functional evaluation shows inorganic carbon capture by C. beijerinckii.  Interested in direct evidence of C-1 assimilation, we performed mixotrophic (sucrose 3% and fructose 1.5% [w/v]) chemostat fer- nmiternotgaetino.nWs (eDo b=s e0r.1v3ed5 hst−e1a)dwyh-sitlaetsetecaodnisluymsppatriogninogfCCOO22aannddHH22aatlohnigghwainthd low concentrations, balanced with proportional increases of apparent SCIENtIfIC REPorTS | 7: 12759 | DOI:10.1038/s41598-017-12962-8 4 www.nature.com/scientificreports/ product-yield values above theoretical levels (Fig. S4 and Table S2). If sucrose and fructose were the only car- bon and energy sources, apparent yields should have remained at or below the theoretical maximum (i.e. 0.66% C-mol, as a result of one decarboxylation from C-3 pyruvate to the C-2 acetyl group of acetyl-CoA). The higher-than-maximum apparent yields indicated additional carbon assimilation that was only possible by inor- ganic carbon capture. Considering current efforts to transform surplus synthesis gas into biofuels6,13, we also sparged this gas at increasing step-wise concentrations (Table S3). Specifically, we sparged synthesis gas mixtures from low (9%), to medium (32%), to high (60%) concentrations, balanced with nitrogen (100% synthesis gas contained 20% CO, 2b0y%thCe Oste2,a1d0y%-stHat2eavnadlu5e0s%(FNig2). .4AAt,lToawbcleo Sn4c)e.nAtrcactoiornd,inCg.lby,eitjheerirnecwkiaisonxoid sizigendiCficOa,nrtedleiaffseirnegnHce2iannadpCpaOr2enast shown C-mol yields and carbon-energy recovery balance compared to the control. Additionally, with increasing electron sink availability, the steady-state sugar utilization improved, resulting in higher product titers (Fig. 4B). Similar behav- ior has been observed by acetogens Clostridium thermoaceticum (Now Moorella thermoacetica)32, C. autoethano- genum, Rhodopseudomonas gelatinosa and also Carboxydothermus hydrogenoformans, according to the following raTHesh2saeicwmrteeiiofrloeanrt:meioC, aniOnxbi +amyg CaHrel.)2eb,Omteh i→ejiensr tCpinwhOcyikt2sih +iio.oA lHuocgr2c,itocwraradhlnibisncechghrliaiyps,vtwimoomhraeisinnculgaycgnuuealtssluetysrdseitssfho(ewris.eeerre.gedoaeovsxexeprsbeomxaspleaadrynetcsboseeihowginrhgohoewnef rtCghs-r-y1olniwmcthanipetmistnuisigrxgefoaagtscretocnoopernhssciwdceuanhlrtelirynna1tg2Ci,2oCO2n,3-2s31/., higher-than-theoretical apparent C-mol yields and carbon/energy mass balances were detected with the con- cgoams ciotanncteinntcrraetaiosends,c1o1nsaunmdp1t7io%nmofoHre2,cCarOb,oann,danCdO129, (aFnigd. 24A7%,Cm,Do)r.eSpcaecrbifoicnalalyn,datemlecetdriounms,arnedspheicgthivseylyn,thweesries recovered. Interestingly, butanol and butyric acid increased by 5.5- and 1.85-fold, respectively, while biomass did not change significantly, which is typical for C1-assimilation pathways such as WL, or rPFOR/Pfl. To confirm gas assimilation into products, we collected steady-state cells (D = 0.135 h−1) growing under high synthesis gas con- centration and cultured them in batch 48 h of incubation and NMR analysis, conditions peaks at ~ in the presence of 178 and 180 ppm rpeuvreeal1e3dCOth2einprtehseenhceeadosfp1a3Cce-laasbterlaecdera.cAetfateter and butyrate, respectively (Fig. 5). These peaks were likely from the enriched 13C quaternary carbon of these com- pounds. We confirmed this by adding acetate and butyrate standards (12.8 and 15 g/L, respectively), and observed an increase of peaks at 178 and 180 ppm (quaternary Cs of acetate and butyrate), at 22 (primary C of acetate), and 14 and 40 ppm (primary, secondary (also at 14 ppm), and tertiary Cs of butyrate, respectively). This is a typical spectrum of unlabeled compounds, where quaternary Cs are difficult to detect, as seen in the control in panel A (Fig. 5A). Overall, this provides direct evidence of C-1 assimilation by C. beijerinckii. Although butanol is the main target in ABE fermentation and the proportion of total carbon in the form of n-butanol increased by 92%, butyric acid is also a value-added product and can be re-assimilated into n-butanol through multi-stage fermentations20,34. The generation of C-4 compounds, such as butyric acid and butanol, require more NADH than C-2 compounds (such as ethanol)16, underscoring the cells emphasis in recycling electrons. The rate of gas assimilation, larger than the saturation values in each condition, also indicated biological activ- ity (Figs 4A and S5). We also tested higher synthesis gas concentrations (up to 100%, containing 20% CO, 20% CpoOis2o, 1n0in%g3H5 2(Faingd. S560)%. TNh2i)s,isnhdoiwcaitnegs carbon capture but generating lower yields, that the fermentation working-window for possibly due to carbon monoxide mixotrophic capture of synthesis gas by C. beijerinckii is between 30 We have also performed batch and 60% (balanced fermentations of C. with N2). beijerinckii under a continuous flow of high synthesis gas concentration as sole carbon and energy source. We observed transient cell proliferation and CO assimilation (not shown). However, cell growth and gas assimilation stopped in early exponential growth phase, as the cells initiated sporulation. As a result, no products were detected. Transcriptomic analysis of the partial WL and rPFOR/Pfl pathways in C. beijerinckii.  To com- plement the time-course transcriptomic data previously described, we performed a RNA-seq experiment using chemostat cultures (D = 0.135 h−1) of C. beijerinckii SA-1, continuously sparged either with nitrogen (control), low or high synthesis gas. Figure 6 shows constitutive expression of each transcripts per million [TPM]), but differentially expressed under both sgyenntehuesnidsegraNs c2ocnodnidtiiotinosn.sU(nnodremr laolwizecdonto- centration of synthesis gas, there was a significant overexpression of a putative formate-THF ligase (Cbs_0101), which belongs to the WL pathway, a PFOR (Cbs_4318), a carbonic anhydrase (Cbs_4425), and a hydrogenase (Cbs_1773). Conversely, there was a repression of a putative CODH (Cbs_5054), a gene that belongs to the for- mate dehydrogenase complex (Cbs_3799), a flavodoxin (Cbs_3109), and several genes that putatively code for the Rnf-complex (Cbs_2449/54). Under high concentration of synthesis gas, there was a significant overexpression of the same genes under low synthesis gas, and a Pfl-activating enzyme gene (Cbs_1010), which belongs to the rPFOR/Pf pathway. However, under that condition, the formate dehydrogenase complex, a CODH (Cbs_3020), a hydrogenase (Cbs_3796), and a cytochrome c biogenesis coding protein (Cbs_2976) were completely shut down. Overall, this transcriptomic data suggests that C. beijerinckii constitutively expresses its putative genes associated to C-1 capture, by preferentially activating those belonging to the rPFOR/Pfl pathway. Specifically, when cultured under synthesis gas, the repression of the formate dehydrogenase, and the overexpression of PFOR and Pfl suggest that CO2 is assimilated via rPFOR/Pfl pathway (and the carbonic anhydrase), as observed with C. thermocellum2. Nitrite as an electron sink for energy conservation.  Under mixotrophic growth, the C-1 assimilation pathways operate mainly for electron recycling2,22,36. Consequently, an alternative way to demonstrate an active pathway is to inhibit CO assimilation by providing an alternative and preferred electron acceptor. Both nitrate and nitrite are known to have this effect on CO assimilation by acetogenic bacteria37,38. To test this hypothesis SCIENtIfIC REPorTS | 7: 12759 | DOI:10.1038/s41598-017-12962-8 5 www.nature.com/scientificreports/ Figure 4.  Kinetic, yield parameters and carbon and energy balances of steady-state cells (D = 0.135 h−1) ocwcoufenlortesuuttroapenbdutttalmyignsaixepsdoaptrahrgfoaetpsederhasfitcoua1rbl2etly.ra4aco8chn tLicn/pohgrn.etSdshetieetniancodeneyto.-vZfsatsealyrutnoeetsvvhaaoellfsuuiHesesg2inoaadsfnig(dcCaaCstOegO,seC2inngOeper2nuatatei nr=oadn toeH(udHt2pu)2.una(tdnA. edN)rCEesgpaOacat2hri)gvtaeeendsvdtanelsdiutyregnosatgshienemndsiiifscxraotgutmaersemutwhtoiearlisevzaagltauisoeenss being produced than input. Positive values indicate the steady-state amounts continuously assimilated. (B) Sucrose and fructose consumed in steady states under the different tested gas phase condition. (C) Apparent yield (C-mol product/C-mol carbon source utilized) and (D) carbon and energy balance, calculated as ∑( for Yp . γp eaγcs h )fe+rmeYnxγt.saγxtio=n 1, where Yp and Yx represent C-mol product58. Synthesis gas mixture was ratios on figure (C); γ represents electrons available balanced with nitrogen (100% synthesis gas contains 20% CO, 20% replicates and CthOe 2r,e1p0r%eseHn2teadndm5e0an%s Nar2e).vTahlueerseastuslttseapdreys-estnatteedfrhoemre were obtained from three biological at least three samples extracted at different retention time intervals. Significance at 0.05 refers to comparisons between whole columns. in C. beijerinckii, we performed chemostat pulse experiments under high synthesis gas concentration (i.e. 60% [v/v]). Interestingly, nitrate showed no effect on C. beijerinckii. However, less-reduced nitrite partially inhibited CO assimilation (1 mol per mol Additionally, biomass increased opfroNpOo2r)t,iownhaitleelyin, schreifatsininggthheydpraotghewnacyofnrosmumcpattaiobnol(i2sm.5 mtooalnoafbHo2lipsmer (mFiogl. o7f)N. BOo2t)h. SCIENtIfIC REPorTS | 7: 12759 | DOI:10.1038/s41598-017-12962-8 6 www.nature.com/scientificreports/ Figure 5.  NMR spectra of C-13 labeled products. Cultures of C. beijerinckii SA-1 growing mixotrophically and dbeow2x0efuiipt%1rjheio0nrCs0siget%nOad4cn,k8[td2voi hia/0hgvr%idn]rigos1Cphw3bCrOsnueyslt2iany,enb1nrthie0tcchle%ee(sedo9iHspr µCrg2ageOaba)sssne2acendntonrcda5ncec0ecaoe%ecofrnef1itNtn1i0rc3a20Ct)tah%icwoleain[debhvre(e(/e6l7vaec0d.]d6o%sC8lpCl OeµsaOcygc2tne)2ei..;tndhAN(CtealMhslt)eisssCRathgemue1aa3lapdstCduly4esrs0pssep%tacemacoctNeeent;rdt(2a(aD,iBaiwon )=ffherCcd oe0u.rm.ale1bts3ue1cii5r0ojnee0nhtrem%t−irrn1noec)sdlakyacliinnaiustdgthflatrregunoosrrmwdioesanws:gr(sdnaiAuns0ip)nct.ph0Cobl8eneoa1mtpnta µcrtieehgrnnossoetle,fndCce. C-13 methanol. electron acceptors have also been shown to increase biomass in acetogens C. thermoautotrophicum and Moorella nthoetrgmeonaecreattiecaa3g7,a38i.nTihneAHT2P-d36e.pTehnudse, nntitCriOte2r, eodruCcOtioanssisimprileafteirornesdaarseathleesrsmeoxdpyennsaimveicwaallyy unfavorable as they to recycle electrons. do As such, the nitrite reductase reaction requires only electrons, in the form of hydrogen and reduced ferredoxin (i.e. not ATP). C. beijerinckii contains a putative ferredoxin-nitrite reductase (Cbei_0832), likely responsible for the observed phenotype, which also unveils this species as a facultative nitrite dissimilator. Transcription of alternative energy-conservation genes.  Considering that there is no net ATP generation through C-1 assimilation pathways, autotrophic bacteria rely either on substrate-level phosphorylation or on chemiosmosis for energy conservation7,22,36. Examples of the latter include cytochromes, Na+ pumps, or the Rnf-complex, whereby acetogens generate an ion gradient for energy generation through ATP-synthases. It has been suggested that B-type cytochromes are responsible for H+-dependent ATP generation, and can be coupled to a membrane-bound methylene-THF reductase. Its location suggests the role of this enzyme in energy conservation39. C. ljungdahlii contains a Rnf-complex but not cytochromes5,40,41. Interestingly, the C. beijerinckii genome encodes cytochromes (also involved in nitrite reduction42) b-type (Cbei_2439), c550 (Cbei_2762), c551 (Cbei_4151), c biogenesis protein (Cbei_2976), cytochrome-bound flavoproteins (Cbei_3109), and also genes coding for the Rnf-complex (Cbei_2449-2454). Additionally, the methylene-THF reductase of C. beijerinckii is predicted43 to contain transmembrane domains. The transcriptomic analysis of the publicly available RNA-seq data27 showed high expression of all these energy-conserving genes, especially the Rnf-complex (Fig. 3). In line with this observation, our transcriptomic analysis of chemostat cultures of C. beijerinckii shows constitutive expression of these genes under nitrogen exposure, and a modest repression under low and high concentrations of synthesis gas (Fig. 6). Since sporulation prevents C. beijerinckii to growth autotrophically, these chemiosmotic mechanisms are potentially useful only during mixotrophic growth. Similarly, C. ljungdahlii requires the Rnf-complex when cultured mixotrophically41. Discussion The variability on apparent product yields reported in the literature and the empirical records of microbial solvent production, demonstrated the need for a deeper study of the evolving gas-phase as signals for overlooked pathways. We have shown that C. beijerinckii captures inorganic carbon and hydrogen under mixotrophic conditions, increasing apparent product yields above theoretical heterotrophic values. Among the putative WL pathway SCIENtIfIC REPorTS | 7: 12759 | DOI:10.1038/s41598-017-12962-8 7 www.nature.com/scientificreports/ Old_locus_ tag A) Expression values (Normalized Log[2]TMP) B) Low Syngas VS Nitrogen C) High Syngas VS Nitrogen Log₂ fold Log₂ fold change P-value Significance Old_locus_tag change P-value Significance Cbs_5054 subunit 9.550747 8.347843 9.01653 Cbs_5054 -1.2 2.90E-05 **** Cbs_5054 -0.5 2.50E-02 * Cbs_3020 subunit Cbs_3798 sulfurtransferase FdhD 3.247928 2.232661 3.070389 1.201634 0 0 Cbs_3020 Cbs_3798 -1.1 1.00E+00 -1.5 1.00E+00 Cbs_3020 Cbs_3798 -10 9.40E-01 -10 1.00E+00 Cbs_3799 Cbs_3800 Cbs_3801 Cbs_0101 biosynthesis protein MobB molybdopterin molybdenumtransferase MoeA formate dehydrogenase subunit alpha formate--tetrahydrofolate ligase 3.925999 3.620586 4.548437 7.309249 0 0.584963 0.765535 8.649256 0 0 0 8.62315 Cbs_3799 Cbs_3800 Cbs_3801 Cbs_0101 -10 8.60E-01 -3 9.00E-01 -4 9.30E-01 1.3 3.20E-04 *** Cbs_3799 Cbs_3800 Cbs_3801 Cbs_0101 -10 7.60E-01 -10 7.60E-01 -10 3.90E-01 1.3 1.60E-04 *** C-1 Fi n Cbs_1702 Cbs_1011 Cbs_1010 Cbs_1009 dehydrogenase/5,10-methylenetetrahydrofolate cyclohydrolase formate acetyltransferase formate acetyltransferase 2.560715 4.620586 5.0917 7.020147 6.852998 7.195741 6.033423 7.175924 7.240314 Cbs_1702 Cbs_1011 Cbs_1009 2 8.10E-01 -0.2 9.60E-01 1.1 9.60E-02 Cbs_1702 Cbs_1011 Cbs_1009 2.6 2.60E-01 0.2 9.00E-01 1.2 5.20E-02 Cbs_1828 Cbs_4318 Cbs_1031 Cbs_4425 Cbs_1773 Cbs_3796 Cbs_4110 Cbs_3013 Cbs_2439 Cbs_2762 Cbs_4151 Cbs_2976 Cbs_3109 Cbs_2449 Cbs_2450 Cbs_2451 Cbs_2452 Cbs_2453 Cbs_2454 methyltransferase/methylenetetrahydrofolate reductase PFOR gamma carbonic anhydrase family protein carbonate dehydratase Hydrogenase hydrogenase Fe-only ferredoxin Ni/Fe hydrogenase cytochrome b5 cytochrome c550 cytochrome c551 cytochrome c biogenesis protein CcdA flavodoxin electron transporter RnfC NADH:ubiquinone oxidoreductase RnfABCDGE type electron transport complex subunit G electron transporter RsxE electron transport complex subunit RsxA RnfABCDGE type electron transport complex subunit B 6.598425 6.411087 5.100137 4.145677 6.490249 5.244126 5.193772 4.89724 6.270529 3.57289 7.417853 2.944858 5.364572 8.321928 6.562242 7.512543 9.467606 6.988685 0 8.476139 2.963474 4.145677 2.655352 4.432959 4.683696 7.712183 1.378512 3.925999 5.053111 3.392317 7.105385 Cbs_1828 9.879583 Cbs_4318 7.086402 C. Cbs_1031 3.378512 anhydrase Cbs_4425 8.656782 Cbs_1773 Hydrogenases 0 Cbs_3796 3.137504 Cbs_4110 2.765535 Cbs_3013 1.632268 Cbs_2439 Cytochromes 4.392317 Cbs_2762 7.712183 Cbs_4151 0 Cbs_2976 2.632268 Cbs_3109 4.638074 Cbs_2449 4.517276 Cbs_2450 Rnf-complex 6.53294 3.678072 4.201634 4.877744 2.560715 1.263034 4.392317 4.240314 2.378512 Cbs_2451 Cbs_2452 Cbs_2453 6.57289 3.678072 2.807355 Cbs_2454 0.9 8.40E-02 3.1 5.10E-19 **** 1.9 1.50E-02 * -10 7.50E-01 2 4.70E-06 **** -2.3 5.50E-01 -1 1.00E+00 -2.3 9.10E-01 -1.9 2.00E-01 1.1 1.00E+00 0.3 8.00E-01 -1.4 1.00E+00 -1.5 5.00E-01 * -3.3 4.40E-06 **** -3.2 3.00E-02 * -2.8 2.90E-02 * -2.3 9.00E-01 -0.1 1.00E+00 -2.8 3.10E-02 * Cbs_1828 Cbs_4318 Cbs_1031 Cbs_4425 Cbs_1773 Cbs_3796 Cbs_4110 Cbs_3013 Cbs_2439 Cbs_2762 Cbs_4151 Cbs_2976 Cbs_3109 Cbs_2449 Cbs_2450 Cbs_2451 Cbs_2452 Cbs_2453 Cbs_2454 0.5 5.80E-01 3.5 0.00E+00 **** 2 4.80E-03 ** -0.8 1.00E+00 2.2 1.90E-08 **** -10 1.10E-01 -2 5.40E-01 -2.2 7.50E-01 -4.9 3.40E-02 * 0.8 1.00E+00 0.3 6.20E-01 -10 1.00E+00 -2.7 3.50E-01 -3.7 4.70E-08 **** -2 5.40E-02 -2.4 5.00E-02 * -3.5 7.10E-01 -2 1.00E+00 -3.9 1.40E-02 * Figure 6.  Transcription analysis of C. beijerinckii C-1 assimilation and energy conservation pathways. RNA was ehxigtrha(c6te0d%f)rocomncceunltturraetisognrsoowfisnygnitnhecshiesmgaoss.tCate(lDls  w=e 0r.e1c3u5l hti−v1a)tespd,airngeddef(i1n2e.d48m Le/dhi)ae2i0t,h2e5r0w rpitmh ,N327, °loCwan(9d%p)H, o6r.5. (100% synthesis gas contains 20% CO, 20% CO2, 10% H2 and 50% N2). genes, C. beijerinckii does not contain annotated an acetyl-CoA synthase, but its CODHs have Fe-S and Ni-Fe-S metal centers, which are typical of CODH subunits of bifunctional CODH/acetyl-CoA synthases44–46. However, it is likely that this enzyme in C. beijerinckii does not lead to acetyl-CoA synthesis, and thus autotrophic growth. Indeed, as has been recently shown, a mutant strain of C. ljungdahlii with a SNP (single nucleotide polymor- phism) in its CODH gene located in its WL cluster (i.e. the one with lower sequence identity to that of C. bei- jerinckii, and associated to a acetyl-CoA synthase) loses its autotrophic phenotype47, even when its CODH with similarity to C. beijerinckii was intact. A similar phenotype has been observed with a CODH-mutant strain of C. autoethanogenum48. Nevertheless, C. beijerinckii contains the genetic potential for an active rPFOR/Pfl-based C-1 capture, including an additional formate dehydrogenase, not present in C. thermocellum2. Based on our physiologic data, we propose a logic model to explain the carbon-electron flow during mix- fCoSoitOmrroc2up a+lhtra biHcno2egn,orioucf awstplhytteh,usCoruefOp:Cp21l.)iibneCdetiOhjseue2rgigtnaoarcsskc-iappirrhcboauocslenteueaidrtseettpaeta r5itao%htnes. utilizes all the sugar-derived carbon and electrons. Otherwise, and if no external electron sink is provided, fer- mentation stops, limiting sugar #3 takes place. If #3 takes place, tuhteileizxattriaonac. eItnylt-hCeopAregseennecreaotefda,naeloxntegrwnaitlheltehcetrsotinll-srinuknn(sinugchAaBsEC-pOa/tChwOa2)y,, #2 or leads to 17 and 27% more carbon and carbon-energy recovered, respectively. Moreover, this physiological capability improves product titers by increasing sugar utilization. However, it is important to note that under our experi- mental conditions, we recovered an apparent ~86 and 100% total carbon and carbon/electron, respectively. Mixotrophic C-1 assimilation was previously shown by cultures of acetogen C. ljungdahlii, whereby exoge- inmoupsorCtaOn2tgimaspinliccraetiaosness carbon recovery49. The discovery of the same phenotype by cultures of C. beijerinckii has in our understanding of the biology of this industrial butanol-producer, and adds a new alternative for greenhouse gas-capture. Indeed, C. beijerinckii stands out among traditional solventogenic species because: (i) it contains genetic elements for cytochromes and the Rnf-system; (ii) it contains genes that code for catalytic enzymes that belong to the WL (except acetyl-CoA synthase) and rPFOR/Pfl pathways; and (iii) the sfeyendcbharcoknocounstHro2l/sCiOn 2bioossceinllsaotirosn50,i5s1.aFnuetuxaremwpolerkofinavnoalvtuinragl integrated oscillator, that can potentially be used for knockout/complementation and biochemical studies will expand our understanding of these processes. Our approach for in-line endogenous gas monitoring shows that it can readably be utilized to uncover new pathways, or potentially even survey a culture (or consortia) for volatile metabolic signatures, in real-time. SCIENtIfIC REPorTS | 7: 12759 | DOI:10.1038/s41598-017-12962-8 8 www.nature.com/scientificreports/ Figure 7.  Transient responses to nitrite pulses by C. beijerinckii growing in chemostat (D = 0.135 h−1). Experiments were performed on defined medium20 containing 3% sucrose and 1.5% fructose (w/v) and sparged with 60% (v/v) aconndc5e0n%traNti2o)nast by monitoring, s3y7n °tCh.eTsihsegaasddbiatliaonncseodfwsoitdhiunmitrnoigternite(1a0re0%indsyicnattheedsiws igtahsvceorntitcaainl dsa2s0h%edClOin,e2s0t%o rCeaOc2h, 10% final H2 as in froelallotwim; 3e.,1w mitMh a(nAE),a6s.y2L minMe c(oBn)tiannudo1u2s.4ga ms Mana(Cly)z.eTr,hme oCdOelaEnLd3H0220d(aAtaBsBh,oGwenrmwearneyo).bNtaiitnrietde concentrations higher than 24 mM proved toxic and led to washout. Steady-state values were re-established CprOiocrotonstuesmtipntgioenacdhisnpiltarciteedc, oanndce(nFt)rabtiioomn.aCssoirnrcerlaetaisoen,swoefrNe cOa2lcaudldateeddwfriothm(Dth)eHsl2ocpoenssauftmeredfi,tt(iEn)gatmheoduanttaotfo linear regressions. Materials and Methods Organisms.  Clostridium beijerinckii SA-1 (ATCC 35702)26 was obtained from the American Type Culture Collection (ATCC). Its identity was verified by PCR amplification and sequencing of the 16S rRNA gene using the prokaryotic 16S rDNA universal primers 515F (5-GCGGATCCTCTAGACTGCAGTGCCA-3) and 1492R (5-GGTTACCTTGTTACGACTT-3). Bacterial medium and inoculum preparation.  C. beijerinckii stocks were activated as previously described17 and were grown in a previously designed medium20. The base components were autoclaved and the sugar (6% w/v sucrose) and trace components were added aseptically to the medium reservoir by filtration (0.22 µm). The inocula were prepared as consistently performed by our lab52. Exact fermentation conditions are detailed in the Main Text section. ®Bacterial culture conditions.  Growth experiments were performed in fed-batch or chemostat modes of operation in a 2-Liter Biostat B plus fermenter equipped with controllers for pH, temperature, agitation, and gas mass-flow (Sartorius BBI Systems, Germany). The temperature was set at 37 °C, agitation speed at 250 rpm, and pH 6.5 culture by the automated addition of 0.5 N KOH or for fed-batch or 700 mL for chemostat. The 25% (v/v) fed-batch fHer3mPOen4,tiantitoonaswwoerrkeisntgarftinedalwviotlhum6%e of 1,400 mL of (w/v) sucrose, and 400 mL containing 80 g of the same sugar were added at constant feed rate (0.08 mL/h) to reach a final concentration of 100 g/L (w/v). The initial volume was 1 L and final 1.4 L. Exact feed components and times of feed start are detailed in the Main Text section. For the chemostat experiments the conditions were identical as described for fed-batch except the carbon and energy source were 3% (w/v) sucrose and 1.5% fructose. Once the cells reached exponential phase under sparged pure nitrogen (OD600 nm ~ 1), the feed and harvest flow were initiated and adjusted to a dilution rate D = 0.135 h−1. Exact sparged gas compositions are detailed in the Main Text section, steady-state conditions were verified for each condition and at least three retention times were allowed before sampling was initiated. Three samples at each steady-state condition were obtained from at least one retention time intervals. The discrete ratios of continuous gas streams were always sparged at 12.48 L/h. Different gas-phase conditions, from pure nitrogen gas to increased synthesis gas concentrations, were achieved by modifying the mix ratios between synthesis gas and nitrogen using two mass flow controllers; the exact con- centrations tested are detailed in the Results section (and Table Supplemental 2). Inlet and exhaust gases in the Hpgao2sl/-eCpmOhaaEssesass(yOpLe2ic,ntNreo2c,moCneOttei,nr.CuBoOiuo2s,mHgaa2ss,saapnnrdaollAyifzree)rrawst,ieomrneomidneoltnhEietLof3er0er2md0ae(nnAdtaBtrBieoc,noGrtedarnemdkaiwnnayrs)e,maaln-otdnimiatoePruefedsiifnafengrdiOnr-emlcinoneriSdOtead2r/CuqsOuian2dgarnaund- in-line biomass sensor (Fundalux, Sartorius, BBI Systems, and Germany) and also by (SmartSpec Plus, BioRad, SCIENtIfIC REPorTS | 7: 12759 | DOI:10.1038/s41598-017-12962-8 9 www.nature.com/scientificreports/ Figure 8.  Logic model of carbon-electron flow in C. beijerinckii grown mixotrophically. The data suggest that it(pnhvu/ettvhaa)teb;ivospeerrnecficsaneeranbolcolfyena,oin3cf)eaCtlnheOhcetyCardon-rnd1asbaCeso,Ostoit2mlr,et2nihl)eaetcCrikoeO,naA,oreBixfiEtCdh-arOfteei2reo m>npoe 5tnos%stgia.betSlineoimenpruaauttlttheialsCnizfeOoeosr2u ac+slall ryHtb,hs,oeuinfpstucphaglepiaetrCdu-Odrseeu2:rgi1inav)retCsdhpOecrag2oratcboseo-ecpndahratabnosodegnlieyasl tce