Discussion
In this work we describe a novel strain of Thermoanaerobacterium thermosaccharolyticum (strain GSU5) through genome analysis and metabolic characterization. The phylogenetic analysis revealed that strain GSU5 belongs to the T. thermosaccharolyticum species and is closest to strain TG57. A genomic analysis of all sequenced strains of Thermoanaerobacterium was also performed, including two strains that have not yet been assigned to any species, PSU2 and RBIITD. When strain PSU2 was first described, it was classified as a strain ofT. thermosaccharolyticum (O-Thong, Prasertsan, Karakashev, & Angelidaki, 2008), and when its genome was published, the authors reported that according to in silico studies, it did not belong to theT. xylanolyticum species (O-Thong et al., 2017). The results presented in this work indicate that neither PSU2 nor RBIITD belong toT. saccharolyticum, T . thermosaccharolyticum, T. xylanolyticum or T. aotearoense . It remains to be studied whether these strains belong to species with no sequenced representatives, or to previously undescribedThermoanaerobacterium species.
T. thermosaccharolyticum synthesizes several products of biotechnological importance, among which ethanol and butanol are the most relevant, along with hydrogen (Bhandiwad, Guseva, & Lynd, 2013). Strains GSU5 and DSM 571 produced both alcohols from all carbon sources tested in this study, including glucose, fructose, galactose, xylose, arabinose, cellobiose and sucrose. The organization of genes involved in butanol synthesis is the same in both strains, and different from the organization of the well-known bcs operon found in solventogenicClostridium (Berezina, Brandt, Yarotsky, Schwarz, & Zverlov, 2009) in which the thl gene is not part of the bcs operon, but constitutes a monocistronic operon in another part of the genome (Wietzke & Bahl, 2012).
It is known that in C. acetobutylicum the transcriptional regulator Rex plays a fundamental role in the regulation of solvent synthesis (Panitz et al., 2014) and binds to specific sequences identified upstream from thl, crt(the first gene of the bcs operon), and adhE (Wietzke & Bahl, 2012). A gene coding for Rex was identified downstream from thebsc operon in T. thermosaccharolyticum GSU5 and DSM571, and putative Rex binding sequences (ROP) were found both upstream fromthl (the first gene of the bsc operon) and of adhE , suggesting that Rex could regulate the expression of these genes, as described in Clostridium . Butanol synthesis has been reported previously not only in several strains of T. thermosaccharolyticum but also in Thermoanaerobacterium sp. RBIITD (Biswas et al., 2018). In this last organism, the structure of the bcs operon is the same as in T. thermosaccharolyticum,except that rex is upstream from the bcs operon, as inC. acetobutylicum .
All genomes analyzed carry adhE , coding for the bifunctional alcohol/aldehyde dehydrogenase, while ald , corresponding to an aldehyde dehydrogenase, was only found in GSU5 and in some butanol producing and non-butanol producing strains. T. thermosaccharolyticum DSM 571 and M5 do not have genes coding for this enzyme, but produce both butanol and ethanol (Bhandiwad et al., 2013; Jiang et al., 2018; Li et al., 2018). These results suggest thatald is not essential for alcohol synthesis inThermoanaerobacterium , in agreement with previous reports that indicated that AdhE was responsible for the synthesis of n-butanol from butyryl-coA and ethanol from acetyl-coA (Bhandiwad et al., 2014). All strains also carried genes coding for other dehydrogenases, including Bhd, so a possible role of these enzymes in the synthesis of alcohols in Thermoanaerobacterium cannot be ruled out.
When T. thermosaccharolyticum GSU5 or DSM 571 were grown in different carbon sources, several of the typical metabolites associated to butanol producers were detected, such as butyrate, acetate, lactate, ethanol and butanol, but no acetone could be found in any of the cultures. In addition, it has been reported that butanol producing strains RBIITD and TG57 do not produce acetone during fermentation (Biswas et al., 2018; Li et al., 2018). A search of the genome of GSU5 for the presence of genes related to acetone synthesis revealed that it lacks adc , coding for an acetate decarboxylase, and ctfAB , which codes for both units of a butyrate-acetoacetate CoA-transferase (Figure 6).
In recent years analysis of the genome of several strains of T. thermosaccharolyticum revealed important differences in the biosynthesis of butanol and butyrate when compared to the pathway known in the well characterized C. acetobutylicum (Jones & Woods, 1986). Previous reports indicated that adc was absent inT. thermosaccharolyticum M5 (Jiang et al., 2018), and that bothadc and ctf were absent in T. thermosaccharolyticumTG57 (Li et al., 2018). The analysis performed in this work revealed a general absence of these genes in all the genomes, indicating that none of the strains analyzed would be able to produce acetone during fermentation due to the absence of adc and ctf . Since the analysis involved all available Thermoanaerobacterium genomes representing half of the known species, these results suggest that this is a common trait in this genus.
Butyrate production has been reported for several T. thermosaccharolyticum strains (Biswas et al., 2018; Freier-Schröder, Wiegel, & Gottschalk, 1989; Li et al., 2018), and is the main metabolite produced from most carbon sources by both strains tested in this work. Among the known butyrate producing pathways, the most common is the acetyl-CoA pathway, that has two variants: i) the two step conversion catalyzed by the phosphotransbutyrylase (Ptb) and butyrate kinase (Buk) with a phosphorylated intermediate that allows the formation of ATP, commonly found in bacteria that have the ABE pathway, and ii) the one step conversion of Butyryl-CoA to Butyrate catalyzed by a butyryl-CoA:acetate-CoA transferase (But) (Vital, Howe, & Tiedje, 2014). Only the one step pathway seems to be performed by strain GSU5 and all other T. thermosaccharolyticum strains, as they carrybut (denominated ach in T. thermosaccharolyticumTG57), and not ptb or buk (Bhandiwad et al., 2013; Li et al., 2018) (Figure 6). The two step pathway has been proposed to exist in strain M5 (Jiang et al., 2018) but its genome does not carry genes coding for Ptb or Buk.
The one step conversion of butyryl-CoA to butyrate has been extensively studied in C. kluyveri , that seems to compensate the lack of the ATP producing step by means of an electron bifurcating mechanism involving the crotonyl-CoA reductase, that couples the reduction of crotonyl-CoA to the reduction of ferredoxin using NADH as the electron donor for both reactions. The reduction of crotonyl-CoA to butyryl-CoA is catalyzed by the cytoplasmic butyryl-CoA dehydrogenase complex, coded by bcd and etfAB , while the reduction of the ferredoxin is catalyzed by NfnAB, a NADH-dependent reduced ferredoxin:NADP oxidoreductase (Wang, Huang, Moll, & Thauer, 2010). These genes are present in all the butanol and butyrate producing strains: T. thermosaccharolyticum and Thermoanaerobacterium sp. RBIITD (table 2). Furthermore, the ferredoxin dependent activity of the butyryl-CoA dehydrogenase was experimentally demonstrated in T. thermoanaerobacterium DSM 571 (Bhandiwad et al., 2014). Based on this information, it can be hypothesized that all butanol producingThermoanaerobacterium are able to obtain energy during the one step butyrate synthesis through the ferredoxin mediated electron bifurcation mechanism (Figure 6). In contrast, genes coding for Ptb and Buk were identified in T. xylanolyticum , T. saccharolyticum , T. aotearoense and Thermoanaerobacteriumsp. PSU-2. These strains lack genes needed for butanol synthesis (bcs operon), and are also devoid of genes coding for the cytoplasmic butyryl-CoA dehydrogenase complex (bcd andetfAB ), suggesting that they are unable to synthesize either butanol or butyrate. In these microorganisms pbt and bukare clustered together with a gene that codes for a leucine/valine/phenylalanine dehydrogenase. This genetic organization has been previously observed in Bacillus megaterium . In this organism Ptb expression was induced in the presence of valine and isoleucine, and the enzyme could use butyryl-CoA and 2-methyl-propionyl CoA as substrates (Vazquez, Pettinari, & Méndez, 2001). In B. subtilis these genes are part of the bkd operon, involved in the degradation of branched chain amino-acids (Debarbouille, Gardan, Arnaud, & Rapoport, 1999). It is possible that in T. xylanolyticum ,T. saccharolyticum , Thermoanaerobacterium sp. PSU-2 andT. aotearoense Ptb and Buk could be involved in branched chain amino-acid degradation as in Bacillus .
Strain GSU5 was able to simultaneously produce both organic acids and alcohols in different carbon sources in a similar manner as the control strain DSM 571. The main acid produced was butyrate, the most abundant metabolic product in most conditions. Both strains were able to synthesize ethanol and smaller amounts of butanol from all carbon sources tested. These results are similar to those obtained in previous studies performed using T. thermosaccharolyticum DSM 571 grown on cellobiose (Bhandiwad et al., 2013), and also using strain M5 grown on xylan (Jiang et al., 2018). In contrast, Li et al report that T. thermosaccharolyticum TG57 produce butyrate, acetate and butanol, but no ethanol, when grown using glucose, cellobiose, cellulose or xylan, in spite of the fact that it has all enzymes required for the synthesis of both alcohols (Li et al., 2018).
When GSU5 was grown in bioreactors, it achieved a higher biomass, and some differences were observed in the relative amounts of acids and alcohols produced. For example, while butyrate was the most abundant metabolite produced in 5 ml tube cultures using all substrates tested, including glucose and xylose, fermentor cultures with those substrates had a greater proportion of alcohols. Furthermore, ethanol was the most abundant metabolite in glucose cultures. This could suggest that in higher density cultures, carbon flow towards the synthesis of alcohols increases contributing to the reoxidation of excess NADH or NADPH, thus favoring the production of alcohols over acids. Solventogenic anaerobes like the broadly studied model bacterium C. acetobutylicum have a fermentation behavior characterized by two distinct phases: formation of acids during the first phase is followed by a solventogenic phase in which growth slows down, and solvents are produced (Amador-Noguez, Brasg, Feng, Roquet, & Rabinowitz, 2011). When the dynamics of acids and alcohols production was analyzed in T. thermosaccharolyticumGSU5, metabolite synthesis was observed to accompany growth, and no clear solventogenic phase could be distinguished. This had been reported for T. thermosaccharolyticum M5 (Jiang et al., 2018) and careful observation of metabolite curves displayed in studies carried out with different solventogenic Thermoanaerobacterium strains show that in all cases alcohols and acids are synthesized throughout growth. This can be clearly perceived in an early work performed using strain DSM 571 (Freier-Schröder et al., 1989) and in the metabolite curves shown in studies that analyze the production of hydrogen in different strains ofT. thermosaccharolyticum (Cao et al., 2009; Cao, Zhao, Wang, Wang, & Ren, 2014; Khamtib & Reungsang, 2012), using both sugars or lignocellulosic biomass for growth. These results suggest that solventogenesis in Thermoanaerobacterium is not subject to the same regulatory mechanisms described for the ABE metabolism in