Hyperthermophilic Thermotoga Species Differ with Respect to Specific Carbohydrate Transporters and Glycoside Hydrolases

Abstract

Four hyperthermophilic members of the bacterial genus Thermotoga (T. maritima, T. neapolitana, T. petrophila, and Thermotoga sp. strain RQ2) share a core genome of 1,470 open reading frames (ORFs), or about 75% of their genomes. Nonetheless, each species exhibited certain distinguishing features during growth on simple and complex carbohydrates that correlated with genomic inventories of specific ABC sugar transporters and glycoside hydrolases. These differences were consistent with transcriptomic analysis based on a multispecies cDNA microarray. Growth on a mixture of six pentoses and hexoses showed no significant utilization of galactose or mannose by any of the four species. T. maritima and T. neapolitana exhibited similar monosaccharide utilization profiles, with a strong preference for glucose and xylose over fructose and arabinose. Thermotoga sp. strain RQ2 also used glucose and xylose, but was the only species to utilize fructose to any extent, consistent with a phosphotransferase system (PTS) specific for this sugar encoded in its genome. T. petrophila used glucose to a significantly lesser extent than the other species. In fact, the XylR regulon was triggered by growth on glucose for T. petrophila, which was attributed to the absence of a glucose transporter (XylE2F2K2), otherwise present in the other Thermotoga species. This suggested that T. petrophila acquires glucose through the XylE1F1K1 transporter, which primarily serves to transport xylose in the other three Thermotoga species. The results here show that subtle differences exist among the hyperthermophilic Thermotogales with respect to carbohydrate utilization, which supports their designation as separate species.

INTRODUCTION

Metabolic engineering efforts to develop microorganisms capable of producing biofuels through the fermentation of sugars have largely focused on manipulating central and intermediary metabolism to direct electrons to the desired bioenergy product (1, 54, 55). In most cases, this presumes that the carbon source is homogenous and readily utilized by the microorganism. However, if lignocellulosic feedstocks are to be processed to biofuels efficiently, microbial systems capable of simultaneously processing mixtures of simple and complex sugars are highly desirable since such systems could circumvent or minimize biomass pretreatment requirements. The strategies used by microorganisms capable of utilizing heterogeneous mixtures of complex carbohydrates rely heavily on the regulation of glycoside hydrolase inventories and the presence of ATP-binding cassette (ABC) transporters to move sugars across cell boundaries (51, 52). Understanding these strategies can ultimately lead to metabolically engineered microorganisms that coordinate the uptake and processing of carbohydrates with the synthesis of biofuel molecules.

Members of the genus Thermotoga have been isolated from locations throughout the world, including Italy, France, The Netherlands, the Azores, Japan, and Africa (4, 12, 19, 21, 25, 28, 39, 53). These bacteria possess a large number of glycoside hydrolases and ABC transporters relative to their genome size (7, 41), which allow them to ferment a wide variety of simple and complex carbohydrates, forming hydrogen (or hydrogen sulfide), acetate, and CO₂ as the primary metabolic products (10). Polysaccharides that can support growth of Thermotoga include xylan, laminarin, carboxymethyl cellulose, starch, glucomannan, galactomannan, and chitin (7, 11). However, Thermotoga species lack multifunctional enzymes with both endoglucanase and carbohydrate binding module (CBM) domains, at least partially explaining their inability to grow on crystalline cellulose (3). Nevertheless, Thermotoga strains have been extensively investigated as sources of hyperthermophilic mannanases, xylanases, and amylases (2, 40, 45). These bacteria are also noteworthy for high-yield hydrogen production (3 to 4 mol/mol glucose), which may be related to the synergistic utilization of NADH and reduced ferredoxin by a bifurcating hydrogenase (9, 46).

Regulatory networks related to carbohydrate utilization in Thermotoga have been predicted by transcriptomics and comparative genomics (11, 38) and, in some cases, validated by characterization of transcriptional regulators (D. Rodionov, personal communication). Transcription of genes in a particular regulon is generally repressed by the transcriptional regulator and induced by binding of the regulator to an effector (often a mono- or disaccharide). For example, genes in the CelR regulon are involved in catabolism and transport of β-linked glucans and glucomannan. Binding of the CelR regulator (TM1218) to the predicted effector (cellobiose) induces transcription of two endoglucanases (TM1524 and TM1525), a mannobiose transporter (TM1219 to TM1223), a transporter of oligosaccharides derived from xyloglucans (TM0300 to TM0304), and a cellobiose phosphorylase (TM1848) (33, 38). The BglR regulon is also induced by cellobiose and was found to be upregulated during growth on various β-linked glucans, including laminarin, barley glucan, and pustulan (11). This regulon's single operon includes the regulator (TM0032), a laminarinase (TM0024), a β-glucosidase (TM0025), and a cellobiose/laminaribiose transporter (TM0027 to TM0031) (33, 38).

The XylR regulon is composed of xylan utilization genes and is organized into 7 or 8 operons, depending on the Thermotoga species. Binding of the XylR repressor (TM0110) to DNA is disrupted by the inducers xylose and glucose in vitro (D. Rodionov, personal communication). This large regulon includes two xylanases (TM0061 and TM0070), a xylobiose/xylotriose transporter (TM0071 to TM0075), a transporter of unknown xylan degradation products (TM0056 to TM0060), the xylose/glucose transporter XylE1F1K1 (TM0112, TM0114, and TM0115), a β-xylosidase (TM0076), a xylose isomerase (TM1667), and a xylulokinase (TM0116) (11, 33, 38, 50).

Another regulon of interest here (GluR) consists of two transporters organized in one operon that were, until recently, uncharacterized. This operon shows variability from one species to another, as Thermotoga petrophila is lacking one of the two transporters and Thermotoga sp. strain RQ2 is missing an α-amylase. Nearly the entire operon is missing from the Thermotoga maritima strain used to generate the whole-genome sequence published in 1999 (34), but a recent study has shown that the strain of T. maritima deposited in DSMZ has all of these genes, except for the gene coding for α-amylase (5). The transporters in this operon from T. maritima were characterized; TreEFG was shown to bind trehalose, sucrose, and glucose, while XylE2F2K2 was shown to bind xylose, glucose, and fucose (5). The regulator encoded in TM1847 is predicted to bind glucose (38).

Given the wide range of geographic locations inhabited by Thermotoga species, it is interesting to consider the underlying factors that make them so adaptive as well as what, if anything, differentiates one species from another with respect to growth physiology on carbohydrates. Genome sequences are available for several hyperthermophilic Thermotoga species, including T. maritima (34), T. petrophila (56), T. neapolitana, and Thermotoga sp. strain RQ2 (GenBank accession no. AE000512, CP000702, CP000916, and CP000969, respectively). An understanding of the physiological characteristics encoded in the “core” genome, as well as the features that are associated with the “non-core” open reading frames (ORFs), could provide new insights into the intricacies of microbial carbohydrate utilization strategies. To this end, growth of four hyperthermophilic Thermotoga species on glucose, a mixture of six monosaccharides, and a polysaccharide mixture was examined to probe the relationship between genotype and phenotype. Even among these closely related hyperthermophiles, sugar utilization patterns and corresponding transcriptomes revealed distinguishing features that could be important in selecting a platform for biofuel production.

MATERIALS AND METHODS

Design of the multispecies cDNA microarray.

A Thermotoga multispecies cDNA microarray was developed by using a combination of previously constructed cDNA probes from T. maritima (7, 23), along with new probes representing ORFs specific to three other species (T. neapolitana, Thermotoga sp. strain RQ2, and T. petrophila). At the time that the multispecies array was constructed, a complete genome sequence was available for T. petrophila, a draft genome sequence was available for T. neapolitana, and suppressive subtractive hydbridization had been used to compare T. maritima and Thermotoga sp. strain RQ2 (36). Due to the incomplete genome information that was available for Thermotoga sp. strain RQ2 at that time, in some cases, there are multiple probes for the same gene. Note that the Thermotoga sp. strain RQ2 genome is now available (GenBank accession no. CP000969).

The genomes of T. neapolitana and T. petrophila were compared to that of T. maritima using GenomeBlast (27). ORFs with amino acid sequences that were not 70% identical with 80% coverage compared to the other three species were considered to be unique to that bacterium. Vector NTI Advance 10 (Invitrogen, Carlsbad, CA) was used to design probes for these unique ORFs. Probes were produced by PCR using genomic DNA as the template and primers obtained from Integrated DNA Technologies (Coralville, IA). PCR products were spotted onto UltraGAPS slides (Corning), as previously described (7). A few of the smaller probes were ordered directly as oligonucleotides from Integrated DNA Technologies (Coralville, IA).

Growth of Thermotoga species on simple and complex carbohydrates.

Cell densities were determined by epifluorescence microscopy using acridine orange stain, as described previously (48). For monosaccharide utilization experiments, the four species were grown on modified sea salts medium to minimize the effect on the refractive index baseline during high-performance liquid chromatography (HPLC) analysis (with the concentration of sea salts lowered to 10 g/liter and resazurin omitted) (17). The monosaccharide mixture (containing glucose, xylose, arabinose, mannose, galactose, and fructose) was prepared in solution, sterilized by filtration, and added to autoclaved medium at a final concentration of 0.25 g/liter of each sugar. Samples taken were passed through a 0.22-μm-pore filter. Monosaccharide concentrations were determined by HPLC with a Shodex SP0810 column at 80°C (52). Water was used as the mobile phase at a flow rate of 0.2 ml/min.

For transcriptional analysis experiments, four Thermotoga species (T. maritima MSB8, T. neapolitana DSM 4359, T. petrophila RKU-1, and Thermotoga sp. strain RQ2) were each grown in pure culture at 80°C on artificial seawater (ASW) medium (6), supplemented with either glucose (0.25% [wt/vol]) (Sigma-Aldrich, St. Louis, MO) or a polysaccharide mixture (0.083% [wt/vol]) consisting of equal parts by mass of lichenan (>85% purity) (Megazyme, Wicklow, Ireland), konjac glucomannan (90% purity) (Jarrow Formulas, Los Angeles, CA), pectin (>85% purity) (Sigma-Aldrich), galactomannan (95% purity) (Sigma-Aldrich), carboxymethyl cellulose (∼99% purity) (Sigma-Aldrich), xylan from oat spelts (90% purity) (Sigma-Aldrich), and xylan from birchwood (90% purity) (Sigma-Aldrich). Polysaccharides were added to autoclaved medium without sterilization. Prior to extracting total RNA, cultures were passed at least six times on either glucose or the polysaccharide mixture, using a 1% (vol/vol) inoculum. Polysaccharide- or glucose-grown cultures were harvested during mid-exponential phase (3 × 10⁷ to 7 × 10⁷ cells/ml) and quickly cooled in a dry ice-ethanol bath, and cells were collected by centrifugation. Cultures grown on polysaccharides were passed through a coffee filter prior to centrifugation. RNA extraction, reverse transcription, labeling, hybridization, and slide scanning were performed with modifications of previously described protocols (7, 23). Loess normalization, analysis of variance (ANOVA) normalization, and ANOVA row-by-row modeling were performed using JMP Genomics (SAS Institute, Cary, NC). Fold changes at or above 2-fold with significant values at or above the Bonferroni correction (typically equivalent to a P value of approximately 10⁻⁵) were considered to be significant differential transcription.

Microarray data accession numbers.

Full results, additional experimental details, and more information about the multispecies array have been deposited in NCBI's Gene Expression Omnibus (13) and are accessible through GEO platform accession no. GPL13655 and GEO series accession no. GSE29557.

RESULTS AND DISCUSSION

Genomic comparison of Thermotoga species.

The 10 Thermotoga species currently identified can clearly be divided into one group with optimum growth temperatures of 65 to 70°C and another group with optimum growth temperatures of 75 to 80°C (Table 1). 16S rRNA sequences indicate that the isolates with higher optimum growth temperatures are more closely related (≥99% identical) to each other than to the less thermophilic Thermotoga strains (15). Complete genome sequences are available for seven Thermotoga species, and all of these, except for T. lettingae and T. thermarum, belong to the group of higher-temperature species (Table 1). Only the 16S rRNA sequence was available for T. naphthophila when this study was initiated, so it was not included; note that T. petrophila, which is closely related to T. naphthophila, was included (35). To probe the significance of differences among these species, the genomes of T. maritima, T. neapolitana, Thermotoga sp. strain RQ2, and T. petrophila were compared on an ORF-by-ORF basis. Based on 70% amino acid sequence identity over 80% of the ORF, a core genome of approximately 1,470 ORFs (Fig. 1) was defined for the four hyperthermophilic Thermotoga species. In general, genes involved in central metabolism are highly conserved. All four species have complete glycolytic, pentose phosphate, and Entner-Doudoroff pathways. T. maritima and T. neapolitana have an extra transketolase that is not present in the other two species, but the physiological role and phenotypic implications of this enzyme are unknown.

Table 1.

Genomic characteristics of Thermotoga species^a

Species or strain	Isolation site	Location	Topt (°C)	% 16S rRNA identity to T. maritima	No. of protein-coding ORFs
T. maritima	Geothermally heated sea floor	Vulcano, Italy	80	100	1,858
Thermotoga sp. strain RQ2	Geothermally heated sea floor	São Miguel, Azores	80	99.7	1,819
T. neapolitana	Submarine thermal vent	Naples, Italy	77	99.4	1,905
T. petrophila	Oil reservoir production fluid	Niigata, Japan	80	99.2	1,785
T. naphthophila	Oil reservoir production fluid	Niigata, Japan	80	99.0	1,768
T. thermarum	Continental solfataric spring	Djibouti, Africa	70	91.8	1,945
T. lettingae	Sulfate-reducing bioreactor	Wageningen, Netherlands	65	90.8	2,040
T. elfii	Oil-producing well	Africa	66	90.7
T. subterranea	Deep continental oil reservoir	East Paris Basin, France	70	90.6
T. hypogea	Oil-producing well	Cameroon, Africa	70	90.2

^aShading indicates species with an optimum growth temperature (T_opt) of 75 to 80°C.

Fig 1.

Venn diagram summarizing the shared ORFs of the selected hyperthermophilic Thermotoga species, based on 70% identity and 80% coverage at the amino acid level over the entire ORF. Numbers indicate the ORFs shared by subsets of species.

These four species have notable genomic differences with respect to carbohydrate utilization and transport (Table 2). Non-core genes include an arabinose utilization locus, present only in T. petrophila and Thermotoga sp. strain RQ2 (TRQ2_0657 to TRQ2_0667), which consists of a predicted arabinose transporter (Table 2) and four enzymes that are apparently targeted to the secretome. Mannan-derived oligosaccharides are likely transported into the cell by the products of TM1746 to TM1750, before being further hydrolyzed by two intracellular glucomannanases (TM1751 and TM1752) (8), which are only present in T. maritima and Thermotoga sp. strain RQ2. A recently characterized glucose/xylose transporter (5) is missing from T. petrophila (Table 2), while a nearby α-amylase (CTN_0781) is missing from T. maritima and Thermotoga sp. strain RQ2. There are two relatively large loci which are not well characterized and are unique to T. neapolitana. The first (CTN_0355 to CTN_0373) includes a glycoside hydrolase family 31 (GH31) enzyme (CTN_0355) and two putative monosaccharide transporters (Table 2), while the second (CTN_1540 to CTN_1555) includes another GH31 enzyme and an uncharacterized transporter (Table 2). The question arises as to whether these variations in glycoside hydrolases and ABC transporters relate to carbohydrate utilization patterns.

Table 2.

Carbohydrate transporter binding proteins identified in genomes of hyperthermophilic Thermotoga species

Predicted substrate(s)	Gene of species or strain:
	T. maritima	T. neapolitana	T. petrophila	Thermotoga sp. strain RQ2
α-Arabinosides	TM0277	CTN_0408	Tpet_0646-7	TRQ2_0671
Cellobiose, laminaribiose	TM0031	CTN_0664	Tpet_0892	TRQ2_0914
Chitobiose	TM0810	CTN_1767	Tpet_0118	TRQ2_0116
Maltose	TM1204	CTN_1367	Tpet_1552	TRQ2_1614
Maltose	TM1839	CTN_0765	Tpet_0966	TRQ2_0984
Mannobiose	TM1223	CTN_1348	Tpet_1545	TRQ2_1595
Rhamnose related	TM1067	CTN_1502-3	Tpet_1677, Tpet_1687	TRQ2_0510
Trehalose, maltose	treE	CTN_0780	Tpet_0954	TRQ2_0970
Xylan related	TM0056	CTN_0638	Tpet_0868	TRQ2_0890
Xylobiose, xylotriose	TM0071	CTN_0622	Tpet_0853	TRQ2_0876
Xylose, glucose	TM0114	CTN_0576	Tpet_0810	TRQ2_0833
Unknown carbohydrate	TM0309	CTN_0378-9	Tpet_0608	TRQ2_0622
Unknown carbohydrate	TM1235	CTN_1337	Tpet_1534	TRQ2_1583
Unknown monosaccharide	TM0595	CTN_0067	Tpet_0322	TRQ2_0340
Galactoside	TM1199	CTN_1375		TRQ2_1619
Glucose, xylose	xylE2	CTN_0777		TRQ2_0973
Mannose-glucuronic acid related	TM1855	CTN_0790	Tpet_0942
Myo-inositol	TM0418	CTN_0252	Tpet_0503
Pectin/galacturonate-related	TM0432		Tpet_1679	TRQ2_0512
Ribose	TM0958	CTN_1618	Tpet_1794
Arabinose			Tpet_0636	TRQ2_0661
Mannan related	TM1746			TRQ2_1079
Mannooligosaccharides	TM1226			TRQ2_1592
Xyloglucooligosaccharides	TM0300			TRQ2_0631
β-Glucoside		CTN_0660
Fructose (PTS)				TRQ2_0639
Galactoside		CTN_1372
Pectin/galacturonate related			Tpet_0485
Pectin/galacturonate related		CTN_0240
Unknown carbohydrate		CTN_1541
Unknown monosaccharides		CTN_0358
Unknown monosaccharides		CTN_0364

Utilization of a mixture of monosaccharides.

The existence of monosaccharide transporters that are not conserved among all four species (Table 2) suggests that these species may show different sugar preferences. To test this, all four species were grown on equal amounts of glucose, galactose, mannose, fructose, arabinose, and xylose. Consumption of the individual monosaccharides was monitored by HPLC. T. maritima and T. neapolitana have very similar monosaccharide utilization patterns (Fig. 2A and B), and there are no well-annotated monosaccharide transporters that differ between them (Table 2). Both species clearly coutilize glucose and xylose, while utilizing arabinose to a lesser extent. Thermotoga sp. strain RQ2 (Fig. 2D) has a similar pattern, except that fructose is consumed almost as quickly as glucose and xylose. This is most certainly due to the fructose-specific phosphotransferase system (PTS), which is present in Thermotoga sp. strain RQ2 but missing from the other three species (Table 2). All six genes of the PTS locus were significantly upregulated on fructose compared to glucose in Thermotoga sp. strain RQ2 (A. D. Frock and R. M. Kelly, unpublished results), supporting the annotation of this PTS as fructose specific. Similar results were seen for a homologous PTS in Caldicellulosiruptor saccharolyticus (52). Although T. petrophila was initially reported to be unable to grow on xylose (49), in this experiment, xylose was depleted faster than the other monosaccharides (Fig. 2C). Comparison of T. petrophila's utilization pattern to the other species reveals that T. petrophila shows less of a preference for glucose, which can be traced to the absence of a transporter that will be discussed in further detail below.

Fig 2.

Utilization of a mixture of six monosaccharides by four Thermotoga species. Mannose and galactose were not significantly utilized by any of the four species, so these two sugars were omitted from the figures for clarity. (A) T. maritima; (B) T. neapolitana; (C) T. petrophila; (D) Thermotoga sp. strain RQ2.

Design of the Thermotoga multispecies cDNA microarray.

Transcriptomic data for the four Thermotoga species were obtained to provide further insights into metabolic differences arising from carbohydrate processing. Previously, a whole-genome cDNA microarray for T. maritima was used to investigate a range of physiological and ecological issues pertaining to the growth of this bacterium (7, 10, 11, 22–24, 29, 30, 42, 43, 47). The high level of sequence identity between these four species suggested that common microarray probes could be used, expanding the transcriptomics analysis to several Thermotoga species. To include those non-core genes absent from the original T. maritima array, probes were added based on sequences from T. neapolitana (304 probes), Thermotoga sp. strain RQ2 (284 probes), and T. petrophila (60 probes).

Monosaccharide and polysaccharide transcriptomes for Thermotoga species.

To compare the genes expressed for carbohydrate utilization in these species, transcriptomes for each species during growth on a mixture of polysaccharides (including glucan, mannan, xylan, and pectin) were compared to that during growth on a single monosaccharide (glucose). As discussed above, these species possess a significant number of carbohydrate active enzymes and transporters that are not conserved among all four species. However, few of these genes were differentially transcribed on the polysaccharide mix compared to glucose. The most significant differential transcription of non-core genes involved a mannan utilization locus (TM1746 to TM1752) (8) and a mannooligosaccharide binding protein (Table 2) (33), which are present only in T. maritima and Thermotoga sp. strain RQ2. For these two species, these genes were upregulated on the polysaccharide mix compared to glucose (Table 3). It is not known how the absence of these genes affects mannan utilization by T. neapolitana and T. petrophila. All four species do have an extracellular endo-1,4-β-mannanase (TM1227) and a mannobiose transporter (TM1219 to TM1223) (33), which may be sufficient for mannan consumption.

Table 3.

ORFs responding in hyperthermophilic Thermotoga species during growth on polysaccharide mixture compared to growth on glucose

Probe	Product	Fold change in expression ina:
		T. maritima	T. neapolitana	T. petrophila	Thermotoga sp. strain RQ2
Mannan utilization
TM1746	Mannan ABC transporter, periplasmic oligopeptide-binding protein	4.6	Absent	Absent	3.7
TM1747	Mannan ABC transporter, permease protein	3.7	Absent	Absent	8.3
TM1748	Mannan ABC transporter, permease protein		Absent	Absent
TM1749	Mannan ABC transporter, ATP-binding protein	6.4	Absent	Absent	4.3
TM1750	Mannan ABC transporter, ATP-binding protein	5.5	Absent	Absent	6.7
TM1751	Endoglucanase	5.9	Absent	Absent	2.5
ManR operon
TM1224	ManR transcriptional regulator, ROK family	3.3	2.4		8.7
TM1226	Mannooligosaccharide ABC transporter, sugar-binding protein	5.4	Absent	Absent	18.3
TM1227	Endo-1,4-β-mannosidase	24.1	12.7	3.9	15.3
CelR regulon
TM0312	Predicted dehydrogenase	2.1	2.1
TM0313	Predicted aldo/keto reductase	2.0		2.6
TM1218	CelR transcriptional regulator, LacI family	3.8		3.9
TM1219	Mannobiose ABC transporter, ATP-binding protein		2.6		2.5
TM1223	Mannobiose ABC transporter, sugar-binding protein	10.0	25.4	4.9	18.8
TM1524	Endoglucanase	7.7		6.8	3.9
TM1525	Endoglucanase	9.5		7.3
TM1848	Cellobiose phosphorylase	10.2	4.7	19.7	9.4
TRAP transporter
TM0322	TRAP transporter, periplasmic substrate-binding protein, putative		8.1		14.7
TM0323	TRAP transporter, small transmembrane component				7.0
TM0324	TRAP transporter, large transmembrane component				3.9
TM0325	Predicted sugar dehydrogenase		2.8		12.0
TM0326	Transcriptional regulator, RpiR family	−3.6			2.6
TM0327	Phosphoglycerate dehydrogenase, putative				6.2
Processing of α-glucans
TM0752	α-Glucosidase, putative	5.3			2.2
TM0767	Maltodextrin glycosyltransferase		−2.7		−6.1
TM1834	α-Glucosidase				−3.6
TM1839	Maltose ABC transporter, periplasmic maltose-binding protein		−3.8	−2.3	−6.6
TM1841	Hypothetical protein				−2.5
TM1842	Hypothetical protein		−2.6	−3.2	−5.9
TM1843	Hypothetical protein				−3.2
TM1844	Hypothetical protein				−5.1
TM1845	Pullulanase			−5.1
CTN_1407	α-Glucan phosphorylase				−2.3
TRQ2_1647a	α-Glucan phosphorylase		−2.2		−2.5
BglR operon
TM0023	Methyl-accepting chemotaxis protein			−7.8	−2.4
TM0024	Laminarinase	−2.6		−11.8
TM0025	β-Glucosidase	−2.1	2.1	−6.7
TM0026	Hypothetical protein			−3.1
TM0027	Cellobiose/laminaribiose ABC transporter, ATP-binding protein	−2.0		−4.7	−2.6
TM0028	Cellobiose/laminaribiose ABC transporter, ATP-binding protein	−2.0		−8.0	−2.4
TM0029	Cellobiose/laminaribiose ABC transporter, permease protein			−5.0	−2.6
TM0031	Cellobiose/laminaribiose ABC transporter, sugar-binding protein	−2.7		−11.6	−2.7
TM0032	BglR transcriptional regulator, ROK family			−4.2

^aShaded values, significantly upregulated on the polysaccharide mix; unshaded values, significantly upregulated on glucose; blank spaces, no statistically significant change; Absent, absent from the indicated genome. The designation “TRQ2_1647a” indicates that there is more than one microarray probe for TRQ2_1647.

Assuming that conserved genes have similar functions in all four species, these genes would be expected to exhibit similar transcriptional responses. This is true in some cases, such as the genes regulated by CelR (Table 3), which are involved in utilizing β-glucans and mannans (11). Genes in this regulon include the previously mentioned mannobiose transporter (TM1219 to TM1223), intracellular and extracellular endoglucanases (TM1524 and TM1525), and a cellobiose phosphorylase (TM1848) with a broad substrate range in vitro (44). TM1223 and TM1848 were upregulated significantly in all four species on the polysaccharide mix compared to glucose. The only genes under the control of CelR that were not upregulated on the polysaccharide mix (TM0299 to TM0307) are putatively coregulated by GloR (38), suggesting that under these conditions their expression may be repressed by this poorly studied regulator.

In other cases, transcriptional responses were less conserved among the four species. All six genes in an operon (TM0322 to TM0327) that includes a tripartite ATP-independent periplasmic (TRAP) transporter were significantly upregulated on the polysaccharide mix in Thermotoga sp. strain RQ2 (Table 3). Two of these genes were upregulated in T. neapolitana, while none were in T. petrophila or T. maritima. This locus was not observed to be responsive in T. maritima grown on 10 different carbon sources (11). TRAP transporters have primarily been shown to bind organic acids (14), so this transporter's role in polysaccharide utilization is unclear. Other genes with different responses in at least one species include those regulated by BglR and those involved in processing α-glucans (Table 3). BglR-regulated genes were generally more highly transcribed during growth on glucose, except for in T. neapolitana, where only one gene was differentially regulated (TM0025), and it was more highly transcribed on the polysaccharide mix. Genes listed in Table 3 involved in processing α-glucans were generally upregulated on glucose, except in the case of T. neapolitana, where one gene (TM0752) was upregulated on the polysaccharide mix.

Glucose transport.

Transcriptomes for growth on glucose and the polysaccharide mix facilitated the identification of a glucose-binding protein not present in the initial T. maritima genome sequence. Although Thermotoga species have been known to grow on glucose since the initial isolation of T. maritima (21), the corresponding genes responsible for glucose uptake were not identified. Glucose transport in T. neapolitana was previously shown to be dependent on ATP and sodium ions (16), and glucose binding activity was detected in cell extracts of T. maritima (31), but subsequent characterization of 15 binding proteins failed to identify one with a high specificity for glucose (33). XylE1 (the product of TM0114) is present in all four species (Table 2) and was shown to bind glucose (50), but this protein also binds xylose with high affinity (K_d [dissociation constant], ∼0.01 μM) (33). Furthermore, TM0114 was more highly transcribed in T. maritima during chemostat growth on xylose than glucose (C.-J. Chou and R. M. Kelly, unpublished results). Therefore, TM0114's involvement in in vivo glucose transport is unclear.

Here, four genes encoding sugar binding proteins were upregulated on glucose compared to the polysaccharide mix (Tables 3 and 4): TM1839, TM0031, TRQ2_0973, and TRQ2_0970 (note that TRQ2_0973 and TRQ2_0970 are homologous to CTN_0777 and CTN_0780, respectively). TM0031 was shown to bind cellobiose and laminaribiose (33), and TM1839 was shown to bind maltose, maltotriose, and trehalose (32). Both of these genes are present in all four species (Table 2). Homologs of TRQ2_0970 and TRQ2_0973 are not present in the published T. maritima genome (34), yet there was significant hybridization of T. maritima cDNA to the probes for these genes. Components of both transporters were significantly upregulated in T. maritima grown on glucose (Table 4). The microarray probes significantly upregulated in T. maritima grown on glucose correspond to a total of 2,658 bp ranging over 5,032 of the 8,862 bp that are present in Thermotoga sp. strain RQ2 but absent from the published genome for T. maritima, strongly suggesting that this entire region is actually present in the T. maritima strain used in this study. In fact, it was recently reported that the T. maritima isolate deposited in DSMZ (used in this study) contains 8,870 bp of genomic DNA that is missing from the isolate sequenced by TIGR in 1999 (5, 34). The product of the TRQ2_0970 homolog (treE) was found to bind trehalose, sucrose, and glucose, while the product of the TRQ2_0973 homolog (xylE2) was found to bind xylose, glucose, and fucose (5). Past studies that focused on T. maritima (33) failed to identify xylE2 because the sequence was missing from the sequenced strain. Noll et al. (37) predicted this genomic variation by examining gene synteny in related species and a β-glucosidase sequence in this region that was published prior to the complete genome (26). In the absence of these hints, genomic variations may not be discovered until the genome is resequenced or other analyses (e.g., RNA-seq) are used. The genomic variation was revealed by the multispecies microarray developed for this study, demonstrating the utility of microarrays in detecting mutations.

Table 4.

ORFs regulated by GluR in four hyperthermophilic Thermotoga species during growth on a polysaccharide mix compared to growth on glucose

Probe	Product	Fold change in expression ina:
		T. maritima	T. neapolitana	T. petrophila	Thermotoga sp. strain RQ2
TM1847	GluR transcriptional regulator, ROK family				−2.8
TRQ2_0975-0976	GluR transcriptional regulator/glucose ABC transporter, ATP-binding protein			Absent	−2.7
CTN_0775	Glucose ABC transporter, ATP-binding protein	−2.4		Absent	−4.6
TRQ2_0975a	Glucose ABC transporter, ATP-binding protein	−3.2		Absent	−2.5
TRQ2_0975b	Glucose ABC transporter, ATP-binding protein	−3.1		Absent	−2.2
TRQ2_0975c	Glucose ABC transporter, ATP-binding protein	−3.0		Absent	−2.5
TRQ2_0974-0975	Glucose ABC transporter, ATP-binding protein/permease protein	−3.0		Absent	−3.1
TRQ2_0974	Glucose ABC transporter, permease protein			Absent	−2.4
CTN_0777	Glucose ABC transporter, sugar-binding protein		−2.2	Absent	−7.0
TRQ2_0973b	Glucose ABC transporter, sugar-binding protein	−2.9	−2.0	Absent	−14.4
TRQ2_0973c	Glucose ABC transporter, sugar-binding protein	−3.5		Absent	−6.3
TRQ2_0972-0973	Glucose transporter sugar-binding protein/trehalose transporter permease protein	−2.2		Absent	−6.4
CTN_0779b	Trehalose ABC transporter, permease protein	−2.1		−2.4
TRQ2_0971	Trehalose ABC transporter, permease protein	−3.3		−3.0
CTN_0780	Trehalose ABC transporter, sugar-binding protein			−2.8	−8.4
TRQ2_0970a	Trehalose ABC transporter, sugar-binding protein			−4.0	−18.2
TRQ2_0970b	Trehalose ABC transporter, sugar-binding protein			−4.2	−13.1

^aThe values shown represent significantly upregulated on glucose. Blank spaces, no statistically significant change; Absent, absent from the indicated genome. When probe names include two ORFs (e.g., “TRQ2_0975-0976”), the microarray probe covers parts of both ORFs (e.g., TRQ2_0975 and TRQ2_0976), as well as the intergenic region.

Regulation of xylan utilization genes.

The most striking difference observed between these species involved genes under the control of the regulator XylR (Table 5) that are present in all four species. As described above, this regulon includes all genes necessary for hydrolyzing xylan, transporting degradation products into the cell, and assimilating the resulting sugars into the pentose phosphate pathway. One would expect that these genes would be upregulated on the polysaccharide mix, which included xylan. For three of these species, this was the case (Table 5), but in T. petrophila, these genes were surprisingly downregulated on the polysaccharide mix compared to glucose.

Table 5.

ORFs regulated by XylR in four Thermotoga species during growth on a polysaccharide mix compared to growth on glucose

Probe	Product	Fold change in expression ina:
		T. maritima	T. neapolitana	T. petrophila	Thermotoga sp. strain RQ2
TM0056	Xylan ABC transporter, periplasmic sugar-binding protein	6.1	4.8	−5.8	9.7
TM0057	Xylan ABC transporter, ATP-binding protein	5.7	3.6		6.2
TM0058	Xylan ABC transporter, ATP-binding protein	5.7		−3.5
TM0060	Xylan ABC transporter, permease protein	4.7
TM0061	Endo-1,4-β-xylanase A	8.2	11.7	−15.3	16.8
CTN_0632	Endo-1,4-β-xylanase A	2.2	5.9
TM0062	Hypothetical protein	19.4	Absent	−6.6	Absent
TM0070	Endo-1,4-β-xylanase B	10.6	2.7		3.1
TM0071	Xylooligosaccharide ABC transporter, periplasmic sugar-binding protein	13.8	2.6	−7.3	14.4
TM0072	Xylooligosaccharide ABC transporter, permease protein	2.1		−4.8
TM0073	Xylooligosaccharide ABC transporter, permease protein	2.1		−3.9	2.9
TM0074	Xylooligosaccharide ABC transporter, ATP-binding protein			−2.9
TM0075	Xylooligosaccharide ABC transporter, ATP-binding protein			−2.6
TM0076	Xylosidase			−2.6
TM0077	Acetyl xylan esterase		2.6	−2.2	2.6
TM0110	Transcriptional regulator, XylR related	2.5
TM0111	Alcohol dehydrogenase, iron-containing			−3.6
TM0113	xylU-related protein			−4.0
TM0114	Xylose/glucose ABC transporter, periplasmic sugar-binding protein			−4.6
TM0115	Xylose/glucose ABC transporter, ATP-binding protein			−2.7
TM0309	Xylan ABC transporter, periplasmic sugar-binding protein	7.5	3.0	−2.6	10.3
TM0310	β-d-Galactosidase	9.2		−2.1	2.8
TM1667	Xylose isomerase	7.2	2.8		6.3

^aShaded values, significantly upregulated on the polysaccharide mix; unshaded values, significantly upregulated on glucose; blank spaces, no statistically significant change; Absent, absent from the indicated genome.

One possible explanation for this unique regulation pattern is that T. petrophila is simply not using the xylan component of the polysaccharide mix. Initially, T. petrophila was reported to be unable to use xylose (49), but the cell density reached when xylose or xylan was supplemented at 2.5 g/liter was approximately 3-fold higher than that on ASW base medium (data not shown), suggesting that T. petrophila is fully capable of growing on these substrates. Also, if T. petrophila was simply not using xylan, one would expect transcript levels for these genes to be similar for both conditions, not higher on glucose. A second possible explanation for the downregulation of the XylR regulon on the polysaccharide mix in T. petrophila is that these genes are repressed by one of the other carbohydrates in the polysaccharide mix. To investigate this possibility, T. petrophila was grown with xylan as the sole carbon source, and transcript levels were compared to glucose. Again, most XylR genes were significantly downregulated on xylan compared to glucose, while no XylR genes were significantly upregulated (data not shown), suggesting that the other components of the polysaccharide mix did not repress the transcription of XylR genes.

Another possible explanation for T. petrophila's regulation of XylR genes is that these genes were actually induced by glucose. As previously mentioned, in T. maritima, Thermotoga sp. strain RQ2, and T. neapolitana, XylE2 is the only sugar binding protein whose transcript was upregulated on glucose compared to the polysaccharide mix that has also been shown to bind glucose (5). However, the transporter is missing from T. petrophila (Table 2), which causes this species to show less of a preference for glucose (Fig. 2C), as discussed above. However, T. petrophila does utilize glucose (Fig. 2C), so this species must bring glucose into the cell by some other means. The only other protein shown to bind glucose in these bacteria, XylE1 (encoded by TM0114), also binds xylose and is regulated by XylR (33, 50). The XylR transcriptional regulator (TM0110) can bind both glucose and xylose, which derepresses transcription of XylR regulon genes (D. Rodionov, personal communication). Therefore, when T. petrophila grows on glucose, it appears that glucose is brought into the cell via XylE1, and glucose binds to the XylR regulator, resulting in increased transcription of all of the XylR regulon genes. T. petrophila might be considered inefficient at utilizing glucose, because it must upregulate so many unneeded XylR genes. This is an interesting example of small differences between closely related species resulting in significantly different strategies for carbohydrate utilization and transcriptional regulation.

Summary.

T. maritima, T. neapolitana, T. petrophila, and Thermotoga sp. strain RQ2 share approximately 1,470 genes (about 75% of their genomes), yet there are differences in sugar utilization and regulatory phenotypes that relate to the genotypes of these species. The multispecies microarray revealed multiple genes that are present in the T. maritima strain from DSMZ, but missing from the published genome sequence, including a glucose transporter that had eluded previous research efforts. When given a mixture of monosaccharides, only Thermotoga sp. strain RQ2 consumed fructose, consistent with the fact that the fructose-specific phosphotransferase system is not present in the other three species. Also, the absence of xylE2 in T. petrophila impairs its ability to use glucose and necessitates the upregulation of xylE1 along with the rest of the xylan utilization regulon during growth on glucose.

The results presented here provide further evidence that the ability of Thermotoga species to ferment individual carbohydrates in heterogeneous growth substrates is dictated by the available set of ABC transporters. The inability of some model biofuel-producing microorganisms to coferment glucose and xylose has prompted metabolic engineering efforts to knock out certain native transporters and insert heterologous systems to remedy this deficiency (18, 20). Thermotoga species are known to naturally coferment glucose and xylose and merit close examination to understand the metabolic basis for this physiological trait. It is interesting that while Thermotoga species appear to transport glucose and xylose using separate ABC transporters, no transporters have been identified that are specific for glucose or xylose. The two sugar binding proteins known to bind either sugar (XylE1 and XylE2) in fact bind both glucose and xylose. This versatility may be important in utilizing renewable heterogeneous feedstocks, which otherwise could limit biofuel production. This underscores the importance of understanding the connection between carbohydrate active enzymes, carbohydrate transporters, and their coordinated regulation.