Genome Sequence of Thermofilum pendens Reveals an Exceptional Loss of Biosynthetic Pathways without Genome Reduction

Abstract

We report the complete genome of Thermofilum pendens, a deeply branching, hyperthermophilic member of the order Thermoproteales in the archaeal kingdom Crenarchaeota. T. pendens is a sulfur-dependent, anaerobic heterotroph isolated from a solfatara in Iceland. It is an extracellular commensal, requiring an extract of Thermoproteus tenax for growth, and the genome sequence reveals that biosynthetic pathways for purines, most amino acids, and most cofactors are absent. In fact, T. pendens has fewer biosynthetic enzymes than obligate intracellular parasites, although it does not display other features that are common among obligate parasites and thus does not appear to be in the process of becoming a parasite. It appears that T. pendens has adapted to life in an environment rich in nutrients. T. pendens was known previously to utilize peptides as an energy source, but the genome revealed a substantial ability to grow on carbohydrates. T. pendens is the first crenarchaeote and only the second archaeon found to have a transporter of the phosphotransferase system. In addition to fermentation, T. pendens may obtain energy from sulfur reduction with hydrogen and formate as electron donors. It may also be capable of sulfur-independent growth on formate with formate hydrogen lyase. Additional novel features are the presence of a monomethylamine:corrinoid methyltransferase, the first time that this enzyme has been found outside the Methanosarcinales, and the presence of a presenilin-related protein. The predicted highly expressed proteins do not include proteins encoded by housekeeping genes and instead include ABC transporters for carbohydrates and peptides and clustered regularly interspaced short palindromic repeat-associated proteins.

Crenarchaeota is one of the two major divisions of the Archaea, and it is the least well represented taxon in terms of genome sequences. Only seven crenarchaeal genomes have been sequenced and published so far, and three of these are genomes of members of the genus Sulfolobus. For the order Thermoproteales, the complete sequence of only one organism, Pyrobaculum aerophilum, has been determined and published so far, although several more species of Pyrobaculum, Caldivirga maquilingensis, and Thermoproteus tenax have been or are currently being sequenced (29). Thermofilum pendens represents a deep branch in the order Thermoproteales, and this organism grows only in rich medium with a fraction of the polar lipids of T. tenax (64), a property that has not been seen before in archaea. Therefore, it was an attractive sequencing target. We report here the genome sequence of T. pendens and analysis of the type strain, T. pendens Hrk5.

T. pendens is an anaerobic, sulfur-dependent hyperthermophile isolated from a solfatara in Iceland. It forms long thin filaments and may have an unusual mode of reproduction in which spherical bulges form at one end of the cell. It requires complex media and a lipid extract from the related organism T. tenax for growth (64). The unknown lipid may be a cellular component or may make sulfur more available to the cells. Complex media, such as tryptone or yeast extract, are required for growth, and CO₂ and H₂S are produced, similar to characteristics of other anaerobic members of the Crenarchaeota and the euryarchaeal family Thermococcaceae. The genome shows that this organism appears to have lost most biosynthetic pathways, yet does not have a reduced genome size compared to other Crenarchaeota.

MATERIALS AND METHODS

Frozen T. pendens Hrk5 cells were obtained from Karl Stetter. Cells were resuspended in 0.25 M sucrose in Tris-EDTA buffer. Sodium dodecyl sulfate was added to a concentration of 1%, and cells were lysed by three cycles of freezing and thawing. Proteinase K was added to a concentration of 50 μg/ml, and the lysate was incubated at 60°C for 30 min. Undigested proteins were precipitated by addition of 0.5 M NaCl and were removed by centrifugation. Nucleic acids in the supernatant were precipitated by addition of an equal volume of cold isopropanol and collected by centrifugation. After digestion with RNase A, the DNA was purified by successive extractions with phenol and phenol-chloroform and recovered by ethanol precipitation. DNA was resuspended in Tris-EDTA buffer and sent to the Joint Genome Institute.

The genome of T. pendens was sequenced at the Joint Genome Institute using a combination of 3-, 6-, and 40-kb (fosmid) DNA libraries. All general aspects of library construction and sequencing performed at the Joint Genome Institute are described at http://www.jgi.doe.gov/. Draft assemblies were based on 21,478 total reads. All three libraries provided 11× coverage of the genome. The Phred/Phrap/Consed software package (www.phrap.com) was used for sequence assembly and quality assessment (7, 8, 10). After the shotgun stage, reads were assembled with parallel phrap (High Performance Software, LLC). Possible misassemblies were corrected with Dupfinisher (16) or transposon bombing of bridging clones (Epicenter Biotechnologies, Madison, WI). Gaps between contigs were closed by editing in Consed, custom primer walking, or PCR amplification (Roche Applied Science, Indianapolis, IN). A total of 465 additional reactions were necessary to close gaps and to increase the quality of the finished sequence. Genes were identified using a combination of Critica (2) and Glimmer (6), followed by a round of manual curation.

Analysis of the T. pendens genome was carried out with the Integrated Microbial Genomes (IMG) system (33). Protein families unique to T. pendens or missing from T. pendens but present in other Crenarchaeota were identified with the phylogenetic profiler in the IMG system. Analysis of signal transduction was carried out using the MiST database (60), which uses SMART (28) and Pfam (9) domain assignments. A cumulative GC skew plot was generated with a 35-kb sliding window using GraphDNA (56).

Predicted highly expressed (PHX) genes were determined with the EMBOSS (45) programs cusp and cai. The training set of PHX genes was compiled from the data of Karlin et al. (21). The genes with a codon adaptation index in the top 5% were considered to be PHX genes.

Nucleotide sequence accession numbers.

The sequences of T. pendens, consisting of the sequences of one chromosome and one plasmid, can be accessed using GenBank accession numbers CP000505 and CP000506. The Genomes On Line Database accession number for T. pendens is Gc00473.

RESULTS

General features.

The genome of T. pendens Hrk5 consists of a circular 1.78-Mbp chromosome and a 31,504-bp plasmid (Table 1). The G+C content is 58%, which is higher than the G+C contents of other Crenarchaeota. A total of 1,923 genes were identified, and 1,883 of these genes encode proteins. The percentage of the genome devoted to encoding genes is 91%, which is slightly higher than the values for other sequenced Crenarchaeota. About 59% of the protein-encoding genes begin with an AUG codon, 32% begin with a GUG codon, and 10% begin with UUG. About 66% of the protein-encoding genes have been assigned to clusters of orthologous groups (COGs) (55), and about 63% have Pfam (9) domains, similar to the values for other archaeal genomes. There is one copy of each rRNA. T. pendens has the highest percentage of fusion genes among members of the Crenarchaeota. Several proteins that are present in T. pendens have not been found in crenarchaeotes or in archaea previously (Table 2), and genes encoding several proteins found in all other crenarchaeotes are missing from the T. pendens genome (Table 3).

TABLE 1.

General statistics for T. pendens

Parameter	Value
Chromosome size (bp)	1,781,889
Chromosome G+C content (bp) (%)	1,027,538 (57.6)
Plasmid size (bp)	31,504
Plasmid G+C content (bp) (%)	17,813 (56.5)
Total genome size (bp)	1,813,393
Total genome G+C content (bp) (%)	1,045,351 (57.6)
Total no. of genes	1,923
No. (%) of RNA genes	40 (2.1)
No. (%) of protein-encoding genes	1,883 (97.9)
No. (%) of genes with function prediction	1,170 (60.8)
No. (%) of genes in ortholog clusters	1,541 (80.1)
No. (%) of genes in paralog clusters	805 (41.9)
No. (%) of genes assigned to COGs	1,264 (65.7)
No. (%) of genes assigned to Pfam domains	1,209 (62.9)
No. (%) of genes with signal peptides	134 (7.0)
No. (%) of genes with transmembrane helices	437 (22.7)
No. (%) of fusion genes	79 (4.11)

TABLE 2.

Unique genes in T. pendens with COG hits

CDS	COG no.	Function
COGs not found in any other sequenced Archaea
Tpen_1241	0698	Ribose-5-phosphate isomerase rpiB
Tpen_1297	3525	Glycosyl hydrolase family 20
Tpen_1097	3444	PTS IIB subunit
Tpen_1100	3715	PTS IIC subunit
Tpen_1100	3716	PTS IID subunit
Tpen_1090	4821	Phosphosugar binding protein, SIS domain
COGs not found in any other sequenced Crenarchaeota
Tpen_1155	1554	Glycoside hydrolase family 65
Tpen_1624	3836	2-Dehydro-3-deoxyglucarate aldolase
Tpen_0948	0207	Thymidylate synthase
Tpen_0017	3613	Nucleoside 2-deoxyribosyltransferase
Tpen_1467	5598	Trimethylamine:corrinoid methyltransferase
Tpen_1211		Monomethylamine:corrinoid methyltransferase
Tpen_1092	1080	Phosphoenolpyruvate-protein kinase (enzyme I of PTS)
Tpen_1091	1925	PTS HPr protein
Tpen_1098	2893	PTS IIA component
Tpen_1491	1268	Biotin transporter BioY
Tpen_0929	3601	Riboflavin transporter
Tpen_0191	2116	Formate transporter
Tpen_1479	2060	Potassium-transporting ATPase, A chain
Tpen_1480	2216	Potassium-transporting ATPase, B chain
Tpen_1481	2156	Potassium-transporting ATPase, c chain
Tpen_0197	0474	Cation transport ATPase (P-type ATPase)
Tpen_1427
Tpen_1010	1327	Predicted transcriptional regulator, Zn ribbon and ATP cone
Tpen_1048	1510	Predicted transcriptional regulator
Tpen_0270	4190	Predicted transcriptional regulator
Tpen_0889	2150	Predicted regulator of amino acid metabolism, ACT domain
Tpen_0253	2229	Predicted GTPase
Tpen_1457	1773	Rubredoxin
Tpen_1536	1811	Uncharacterized membrane protein, DUF554
Tpen_0198	2047	Uncharacterized protein, ATP-grasp superfamily
Tpen_0838	2164	Uncharacterized conserved protein, DUF369
Tpen_1835
Tpen_1118	2908	Uncharacterized conserved protein
Tpen_0381	3863	Uncharacterized relative of cell wall-associated hydrolases
Tpen_1090	4821	Uncharacterized protein with phosphosugar binding domain

TABLE 3.

Genes present in all Crenarchaeota except T. pendens with COG hits

COG	Function
0214	Pyridoxine biosynthesis enzyme (YaaD)
0311	Glutamine amidotransferase involved in pyridoxine synthesis (YaaE)
0413	Ketopantoate hydroxymethyltransferase
0452	Phosphopantothenoylcysteine synthetase/decarboxylase
0108	3,4-Dihydroxy-2-butanone-4-phosphate synthase
1985	Pyrimidine reductase, riboflavin biosynthesis
0054	Riboflavin synthase beta chain
0163	3-Polyprenyl-4-hydroxybenzoate decarboxylase
1635	Flavoprotein involved in thiazole biosynthesis
0112	Glycine/serine hydroxymethyltransferase
0189	Glutathione synthase/ribosomal protein S6P modification enzyme/l-2-aminoadipate N-acetyltransferase
0105	Nucleoside diphosphate kinase
2046	Sulfate adenylyltransferase
1650	Uncharacterized protein conserved in archaea
1701	Uncharacterized protein conserved in archaea

The plasmid is predicted to encode 52 proteins, only 2 of which have similarity to proteins in the GenBank nonredundant protein database. The Tpen_1849 protein is similar to a T. pendens chromosomal protein with an unknown function (Tpen_0735), and the Tpen_1875 protein is a predicted helicase. In addition, the Tpen_1891 protein is predicted to be a site-specific recombinase (COG4974). The function of the plasmid and whether it is beneficial to the host are currently unknown.

Cumulative GC skew analysis of the T. pendens genome was used to identify the potential origin(s) of replication (12). A global minimum was located at position 488884, which is near a 478-bp intergenic region between positions 487890 and 488368. The intergenic spacer contains several repetitive sequences similar to conserved crenarchaeal origin recognition boxes (46).

No repetitive elements were found when the ISfinder database (53) was searched with the T. pendens coding sequences. However, in a 100,000-bp section of the genome there are 12 stretches of clustered regularly interspaced short palindromic repeat (CRISPR) elements that are interspersed with protein-encoding genes. CRISPR elements and their associated genes constitute a defense against viruses (3). Among the 97 PHX genes (see Table S1 in the supplemental material) are 6 genes encoding CRISPR-associated proteins (Tpen_1263, Tpen_1287, Tpen_1288, Tpen_1316, Tpen_1342, and Tpen_1356). Interestingly, a group of seven consecutive genes (Tpen_1287 to Tpen_1293), including two genes encoding CRISPR-associated proteins, is PHX. While other thermophilic archaea have similar numbers of genes encoding CRISPR-associated proteins in their genomes, most do not have genes encoding CRISPR-associated proteins that are PHX genes, although Staphylothermus marinus and P. aerophilum do (four and two genes, respectively). However, T. pendens has the highest number of CRISPR-associated genes that are PHX genes and the highest percentage. Thus, protection against viral infection appears to be a major priority for T. pendens. Crenarchaeota from hot spring environments are known to host a wide variety of viruses with distinctive morphologies (for a review, see reference 42).

Central metabolism.

T. pendens contains complete glycolysis and gluconeogenesis pathways. Glyceraldehyde-3-phosphate:ferredoxin oxidoreductase, found in some archaeal hyperthermophiles as part of an alternative step in glycolysis (37), is not present in T. pendens. Phosphoenolpyruvate synthase, used in the last step in glycolysis in Thermococcus kodakaraensis (19), is present in T. pendens (Tpen_0588). Phosphoenolpyruvate synthase could be involved in glycolysis and/or in gluconeogenesis. Starch synthesis and utilization pathways are also present.

Pentoses are synthesized through the ribulose monophosphate pathway that is common in archaea (for a review, see reference 22). T. pendens genes encode two ribose-5-phosphate isomerases, one RpiA-type enzyme (Tpen_0327) and one RpiB-type enzyme (Tpen_1241). This is the first time that an RpiB enzyme has been found in archaea. The rpiB gene is adjacent to a uridine phosphorylase gene (Tpen_1240), suggesting that this protein has a function in nucleoside utilization. Under conditions in which excess ribonucleosides are present, RpiB may be involved in conversion of ribose phosphate to hexoses through the ribulose monophosphate pathway, a reversal of the pathway from its predicted function in archaea.

ATP can be generated from pyruvate through the consecutive action of pyruvate:ferredoxin oxidoreductase and ADP-forming acetyl-coenzyme A (acetyl-CoA) synthase, similar to the activity in Thermococcales (32). The T. pendens pyruvate:ferredoxin oxidoreductase (Tpen_0571 to Tpen_0574) is similar to the Thermotoga maritima enzyme that has been characterized (23). In Thermococcales and Crenarchaeota, ADP-forming acetyl-CoA synthase has two subunits, the alpha and beta subunits. T. pendens contains one alpha subunit (Tpen_0336), one beta subunit (Tpen_0109), and one protein with alpha and beta subunits fused together (Tpen_0602). Two AMP-forming acyl-CoA synthases are also present (Tpen_0893 and Tpen_1611). T. pendens has four other enzymes similar to pyruvate:ferredoxin oxidoreductase (Tpen_0540 to Tpen_0543, Tpen_0781 and Tpen_0782, Tpen_0856 and Tpen_0857, and Tpen_1455 and Tpen_1456), and these enzymes are likely to be involved in amino acid degradation pathways in which the amino acid is first converted to the 2-ketoacid, then to the acyl-CoA, and finally to an acid, with ATP generated by acyl-CoA synthases (32). Four aldehyde:ferredoxin oxidoreductases are also present (Tpen_0094, Tpen_0176, Tpen_1413, and Tpen_1817), and these enzymes could be involved in peptide fermentation (1). The 2-oxoacid oxidoreductases produce aldehydes, which are converted to acids. Reduced ferredoxin is produced, but there is no ATP production in this pathway.

T. pendens appears to assimilate glycerol. There is a glycerol kinase gene (Tpen_1128) adjacent to the gene encoding subunit A of glycerol-3-phosphate dehydrogenase (Tpen_1127). Next to these genes are three genes with similarity to genes encoding subunits B, C, and D of succinate dehydrogenases (Tpen_1124 to Tpen_1126). A gene encoding subunit A of succinate dehydrogenase has not been found in the genome. It appears that the three succinate dehydrogenase-related subunits along with the glycerol-3-phosphate dehydrogenase subunit A may form a novel glycerol-3-phosphate dehydrogenase that may transfer electrons to a quinone or another acceptor.

Unlike most of the sequenced Crenarchaeota, T. pendens has a ribulose-1,5-bisphosphate carboxylase (Tpen_1227). It also has the recently discovered enzymes involved in conversion of the ribose phosphate group of AMP to ribulose 1,5-bisphosphate (50): AMP phosphorylase (Tpen_0093) and ribose-1,5-bisphosphate isomerase (Tpen_0384). Under conditions in which acetate is incorporated into the gluconeogenesis pathway, the AMP-forming acetyl-CoA synthetase and phosphoenolpyruvate synthase could produce substantial amounts of AMP. The T. pendens AMP-forming acetyl-CoA synthetase (Tpen_0893) has a very high level of similarity to the P. aerophilum enzyme that has been characterized (4). A large amount of AMP may also be generated by ribose-phosphate pyrophosphokinase, which is required for pyrimidine synthesis, and phosphoribosyltransferases.

Biosynthesis.

T. pendens is known to require an extract of T. tenax for growth. While the specific compound from T. tenax required could not be identified, the genome analysis revealed a vast reduction in the ability of T. pendens to synthesize basic metabolites. T. pendens appears to be dependent on its environment for purines, most cofactors, and most amino acids. A list of 125 COGs involved in synthesis of nucleobases, amino acids, and cofactors was compiled (see Table S2 in the supplemental material). COGs encoding archaeal biosynthetic enzymes were included where they are known. The presence of these COGs in all complete bacterial and archaeal genomes was determined using the function profile feature in the IMG system. T. pendens possesses only 11 of these COGs. The COGs found in T. pendens include genes encoding five enzymes involved in pyrimidine synthesis, aspartate carbamoyltransferase (COG0540), dihydroorotase (COG0044), dihydroorotate dehydrogenase (COG0167), orotate phosphoribosyltransferase (COG0461), and orotidine-5′-phosphate decarboxylase (COG0284). Also present are genes for threonine synthase (COG0498), 2-polyprenylphenol hydroxylase (COG0543), protoporphyrinogen oxidase (COG1232), 4-hydroxybenzoate polyprenyltransferase (COG0382; two copies), 1,4-dihydroxy-2-naphthoate octaprenyltransferase (COG1575; two copies), and a methylase involved in ubiquinone/menaquinone biosynthesis (COG2226; six copies). The only organisms with fewer members of this COG set were obligate parasites or commensals. In fact, some obligate parasites, such as Rickettsia species, have greater biosynthetic capabilities than T. pendens.

While it is possible that T. pendens has different pathways for metabolite synthesis or has many enzymes that were replaced through nonorthologous gene displacement, this is unlikely to account for the lack of biosynthetic enzymes because other Crenarchaeota have recognized pathways for basic metabolites. For example, all Crenarchaeota except T. pendens have homologs of the pyridoxine biosynthesis genes pdx1 and pdx2 (yaaD and yaaE in Bacillus subtilis) and the bifunctional CoA biosynthetic enzyme phosphopantothenoylcysteine synthetase/decarboxylase. Table 3 shows the COGs missing from T. pendens that are found in all other sequenced Crenarchaeota. Nine of these COGs are involved in pyridoxine, CoA, riboflavin, ubiquinone, and thiamine biosynthesis. In addition, most Crenarchaeota have homologs of several heme biosynthetic enzymes, but T. pendens lacks these enzymes. They are not shown in Table 3 because they are also not present in the Staphylothermus marinus genome. Also, COG1731, encoding archaeal riboflavin synthase, is not shown in Table 3 because it is not present in either T. pendens or Cenarchaeum symbiosum; however, C. symbiosum has the bacterium-type riboflavin synthase (COG0307), but T. pendens lacks both the bacterial and archaeal enzymes.

In accordance with the predicted lack of biosynthetic capacity, T. pendens is the only crenarchaeote that has a BioY family biotin transporter and a riboflavin transporter (Table 2). In addition, T. pendens has an expanded number of ABC transporters related to the ABC transporters involved in cobalt uptake. While most Crenarchaeota have zero to two representatives of this family, T. pendens has seven. One of these transporters has an additional membrane protein related to B. subtilis YkoE, and such transporters are predicted to transport the thiamine precursor hydroxymethylpyrimidine (47).

T. pendens has genes for limited amino acid synthesis. It has a putative cysteine synthase (Tpen_1605) related to a previously characterized Aeropyrum pernix enzyme (41), but no serine acetyltransferase. Probably, like A. pernix, T. pendens uses O-phosphoserine rather than O-acetylserine as the intermediate in cysteine synthesis. Cysteine synthesis may have been preserved in T. pendens so that cysteine can help protect the cell against oxidative stress, a phenomenon that is thought to occur in some parasitic protists (for a review, see reference 38).

Glutamine can be synthesized from glutamate on its tRNA (Tpen_0360 and Tpen_0361) and also by a cytosolic glutamine synthase (Tpen_1089). Cytosolic glutamine synthesis has probably been preserved because of its role as a nitrogen donor. T. pendens has six proteins with glutamine amidotransferase domains, including CTP synthase (Tpen_1163) and glucosamine-6-phosphate synthetase (Tpen_0085 and Tpen_1094). Asparagine can be synthesized by a tRNA synthetase-related, archaeal asparagine synthetase (Tpen_1140) (48).

T. pendens has a cobalamin-independent methionine synthase (Tpen_1819) but no homoserine biosynthesis genes; thus, it can probably not make methionine de novo, but it can recycle homoserine resulting from S-adenosylmethionine-dependent methylation reactions. Interestingly, T. pendens has genes related to genes encoding monomethylamine and trimethylamine methyltransferases in Methanosarcinales (Tpen_1211 and Tpen_1467). The T. pendens monomethylamine methyltransferase is related to the Methanosarcina enzymes, and this is the first time that this protein family has been found outside the Methanosarcinales. Both putative methyltransferase genes are adjacent to corrinoid protein genes (Tpen_1212 and Tpen_1468), supporting their function as methyltransferases. Where the Methanosarcina proteins have pyrrolysine residues, both proteins of T. pendens have leucine. The methyl groups transferred from methylamines could be used to recycle methionine after methylation reactions.

T. pendens can synthesize pyrimidines but not purines de novo. Carbamoyl phosphate for pyrimidine synthesis is generated by carbamate kinase (Tpen_0172), not by carbamoyl phosphate synthase, similar to the activity in Pyrococcus furiosus (61). There are a variety of phosphorylases and phosphoribosyltransferases that could be used for salvage of bases. In addition, T. pendens has an open reading frame (Tpen_1649) with 66% similarity to A. pernix APE0012, which encodes a broad-range nucleoside kinase as well as a phosphofructokinase (17); thus, nucleosides may also be salvaged. No genes encoding transporters belonging to known families of nucleobase or nucleoside transporters could be identified in the genome.

T. pendens appears to be able to synthesize phospholipids de novo. It may have a modified mevalonate pathway like that predicted for Methanocaldococcus jannaschii (13) as it has a homolog of the MJ0044 protein, which was shown to be an isopentyl phosphate kinase (Tpen_0607). It has the enzymes for synthesis of sn-glycerol 1-phosphate (Tpen_1231) and geranylgeranyl diphosphate (Tpen_0606) and for attaching the geranylgeranyl groups to glycerol 1-phosphate (Tpen0633, Tpen_0636, and Tpen_1449). Like many archaea, it has only one identifiable CDP-alcohol phosphatidyltransferase (Tpen_0218), and this enzyme is most closely related to archaetidylinositol synthases. myo-Inositol-1-phosphate synthase is present (Tpen_1660). It is not known whether T. pendens makes additional phospholipids.

Carbohydrate metabolism and transport.

T. pendens requires a complex growth medium, such as yeast extract, tryptone, or gelatin, and sucrose stimulates growth (64). It was concluded that T. pendens grows mainly by peptide fermentation. While T. pendens does have enzymes for amino acid degradation, the genome analysis revealed that sugars and sugar polymers may also be important growth substrates for this organism.

One source of evidence that carbohydrates are important growth substrates is the set of transporters encoded in the genome. T. pendens genes encode eight ABC transporters belonging to family 1 (Transport Classification Database [http://www.tcdb.org]), which are involved in sugar uptake (Tpen_1055 to Tpen_1057, Tpen_1149 to Tpen_1152, Tpen_1174 to Tpen_1177, Tpen_1255 to Tpen_1257, Tpen_1451 to Tpen_1453, Tpen_1547 to Tpen_1550, Tpen_1588 to Tpen_1590, and Tpen_1617 to Tpen_1619). Within the archaea, only Haloarcula marismortui possesses as many family 1 ABC transporters. T. pendens also has one ABC transporter belonging to family 2, which is likely to be involved in sugar uptake (Tpen_1208 to Tpen_1210). The only other family 2 ABC transporter in the archaea is a transporter in Sulfolobus acidocaldarius. One of the four family 5 ABC transporters in T. pendens (Tpen_1676 to Tpen_1680) is similar to a P. furiosus cellobiose transporter (24) and a T. maritima transporter for mannobiose (TM1223) (39). Two members of the glycoside-pentoside-hexuronide:cation symporter family are also present (Tpen_1599 and Tpen_1831).

T. pendens is the only sequenced crenarchaeote that has the phosphotransferase system (PTS) for carbohydrate uptake (Table 2). The only other sequenced archaeon that has a PTS transporter is H. marismortui. Haloquadratum walsbyi has enzyme I and HPr proteins of the PTS, but it does not have identifiable PTS transporters. An enzyme I phylogenetic tree shows that the T. pendens and halophile proteins are not closely related, suggesting that they were independently acquired through separate lateral transfer events (not shown). The enzyme I gene in T. pendens is adjacent to the genes encoding a predicted N-acetylglucosamine-6-phosphate deacetylase, suggesting that N-acetylglucosamine may be the substrate for this transporter.

T. pendens has a set of 15 glycosyl hydrolases, about the same number as Sulfolobus species and greater than the numbers in other Crenarchaeota. There are several genes that are involved in starch utilization. One cluster of genes encoding two glycosyl hydrolases and an ABC transporter (Tpen_1451 to Tpen_1454 and Tpen_1458) is similar to a cluster in Thermococcus sp. strain B1001 involved in extracellular formation of cyclomaltodextrins, transport of cyclomaltodextrins into the cell, and intracellular degradation of the cyclomaltodextrins (18). In addition, there is an alpha-glucosidase (Tpen_1511) similar to the characterized NAD⁺-dependent T. maritima enzyme (44).

Cellulose may also be utilized by T. pendens. This species has a secreted family 12 glycosyl hydrolase (Tpen_1681; for glycosyl hydrolase classification, see CAZy [http://www.cazy.org]) with weak similarity to cellulases, as well as an ABC transporter with a high level of similarity to a characterized cellobiose transporter in P. furiosus (Tpen_1676 to Tpen_1680) (24). Cellobiose and larger oligosaccharides may be broken down by an intracellular beta-glucosidase (Tpen_1494).

Sucrose stimulates the growth of T. pendens but does not serve as a sole energy source (64). The enzymes involved in sucrose metabolism could not be identified from the genome sequence. No beta-fructofuranosidase (invertase) or sucrose phosphorylase could be identified, and there is no homolog of PF0132, which encodes the invertase purified from P. furiosus (27).

Genes encoding three glycosidases (Tpen_1511, Tpen_1269, and Tpen_1458) and three ABC transporter-associated sugar-binding proteins (Tpen_1055, Tpen_1208, and Tpen_1257) are among the PHX genes of T. pendens, providing further evidence of the importance of carbohydrate metabolism. Genes encoding subunits of two peptide ABC transporters are also PHX genes (Tpen_1635 and Tpen_1636, Tpen_1638, Tpen_1245, and Tpen_1247 to Tpen_1249). This reflects the fact that T. pendens needs to obtain many amino acids from external sources and utilizes peptides for energy.

Electron transport.

T. pendens requires sulfur for growth and produces H₂S, and genes encoding some of the potential catalysts for this metabolism can be identified in the genome sequence. T. pendens does not have a hydrogenase related to sulfhydrogenase and hydrogenase II of P. furiosus, which reduce sulfur as well as protons (31). It also does not possess a sulfide dehydrogenase (30). There is a homolog (Tpen_0143; 48% identity and 66% similarity) of the recently identified CoA-dependent NADPH:sulfur oxidoreductase in P. furiosus (52). However, T. pendens does not have the mbx protein complex that is predicted to transfer electrons from ferredoxin to NADPH. Also, T. pendens does not have a homolog of bacterial ferredoxin-NADP⁺ reductases (COG1018), so the pathway for recycling ferredoxin is unknown. T. pendens has a large set of adjacent genes (Tpen_1070 to Tpen_1088) with similarity to genes encoding NADH dehydrogenases and membrane-bound hydrogenases. This cluster may encode one or more multisubunit enzymes that oxidize ferredoxin and transfer the electrons to NADP, a quinone, or another electron carrier.

T. pendens has an operon (Tpen_1121 to Tpen_1123) similar to the psrABC genes encoding the polysulfide reductase in Wolinella succinogenes (25). The protein similarity is weak, but the three proteins encoded by the T. pendens operon belong to the same protein families as the polysulfide reductase subunits. The A subunit is predicted by ProSite to have a twin-arginine signal peptide, so the enzyme probably reduces its substrate extracellularly.

A substantial amount of formate may be produced by fermentative organisms in the environment in which T. pendens lives, and T. pendens appears to have two pathways for utilizing formate. Like P. aerophilum and Hyperthermus butylicus, T. pendens has a three-subunit, membrane-bound, molybdopterin-dependent formate dehydrogenase. The alpha subunit has a predicted twin-arginine signal peptide, so the topology of the enzyme is likely to be similar to the solved structure of Escherichia coli formate dehydrogenase N with formate oxidation occurring outside the cell (for a review, see reference 20). This enzyme likely channels electrons from formate to a quinone or other carrier and then to sulfur as the final electron acceptor.

T. pendens is the only crenarchaeote that has a formate transporter (Tpen_0191). The transporter gene is adjacent to a putative operon (Tpen_0190 to Tpen_0178) with a high level of similarity to the genes encoding E. coli hydrogenase 4. In E. coli, hydrogenase 4 forms part of the formate hydrogen lyase complex which oxidizes formate and produces hydrogen under conditions in which no electron acceptors other than protons are present. The T. pendens operon contains the gene encoding a formate dehydrogenase alpha subunit, providing strong evidence that this operon encodes formate hydrogen lyase. The formate dehydrogenase protein does not have a signal peptide, suggesting that formate oxidation occurs in the cytoplasm, as observed for the E. coli complex (for a review, see reference 51). Formate hydrogen lyase contributes to the generation of a proton gradient in two ways: by using protons from inside the cell to make H₂, which then diffuses out of the cell, and under some conditions by pumping protons out of the cell (14). This enzyme complex is expressed in E. coli only when no electron acceptors are present, suggesting that T. pendens may use this pathway when sulfur is scarce.

T. pendens may also use hydrogen as an electron donor as it contains genes (Tpen_0591 to Tpen_0594) similar to the genes encoding four subunits of a membrane-bound uptake hydrogenase in Acidianus ambivalens (26). The A. ambivalens hydrogenase is predicted to use a quinone to transfer electrons from hydrogen to sulfur. This type of pathway is common among archaeal autotrophs, and in T. pendens it may supplement the energy derived from peptides and sugars.

Signal transduction.

Archaea have significantly fewer signal transduction systems than bacteria. On average, 2.63% of archaeal proteomes and 5.4% of bacterial proteomes consist of signal transduction proteins (60). Moreover, it has been shown previously that in archaeal signal transduction a substantially reduced repertoire of sensory (input) and regulatory (output) domains is utilized (59). The median level of archaeal one-component systems in each genome is roughly 50 times greater than the median level of two-component systems, and the majority of these systems regulate gene expression at the transcriptional level (60). Two-component systems have been found only in Euryarchaeota and appear to have been laterally transferred from bacteria. In general, crenarchaeal species have fewer signal transduction systems (only 0.7% of the proteome) than euryarchaeotes.

The T. pendens genome contains genes encoding 45 one-component systems (regulatory proteins that contain one or more sensory domains [59]) that comprise 2.4% of the proteome. This percentage is more than three times greater than the average percentage for Crenarchaeota. Thirty-one (69%) of these systems are encoded by operons containing predominantly enzymatic genes and are predicted to regulate their transcription in response to environmental and intracellular signals. Interestingly, T. pendens possesses three families of transcription regulators that have not been found previously in Crenarchaeota (Table 2). T. pendens encodes more PadR domains than any other crenarchaeal species, possibly indicating that there is a high level of phenolic acid metabolism. Unlike all other crenarchaeotes, T. pendens does not have a member of the fur family, which is responsible for regulating metal ion uptake, although it does possess an iron-dependent repressor gene (Tpen_0973), which is located beside an iron transporter gene.

Presenilin.

A gene encoding a protein belonging to the presenilin family is present in the T. pendens genome (Tpen_0870). In eukaryotes presenilin is an integral membrane protease and part of the gamma-secretase complex (63). Mutations in presenilin cause it to cut amyloid precursor peptide in a different place and generate amyloid precursor peptide forms that are more likely to aggregate and form plaques. A family of proteins weakly related to presenilins, known as presenilin homologs, has been identified in eukaryotes and archaea (11, 43). The T. pendens protein and a related protein in P. aerophilum are not closely related to these presenilin homologs; they represent a new subfamily of presenilins. These crenarchaeal proteins are about 150 amino acids shorter than the mammalian presenilins and lack hydrophilic regions at the N terminus and in an internal loop (Fig. 1). They contain seven to nine predicted transmembrane helices and the conserved YD, LGXGD, and PALP motifs. The gamma-secretase complex includes three other proteins, but none of these proteins are present in T. pendens or P. aerophilum. Characterization of this new subfamily of presenilins may shed light on the structure and function of the eukaryotic proteins.

FIG. 1.

Alignment of presenilins from a human and Thermoproteales. The alignment was constructed with Clustal W (57).

DISCUSSION

The genome sequence of T. pendens shows that this organism lost enough biosynthetic pathways that it is not a free-living organism and is a commensal that is dependent on another archaeon. Besides the lack of biosynthesis, several nutrient transporters that are not found in any other crenarchaeote are present in T. pendens. A lack of biosynthetic ability and an increase in nutrient transport ability are features commonly found in obligate parasites (49). However, T. pendens lacks other features of obligate intracellular parasites, such as a reduced genome size (40, 62), a loss of signal transduction and DNA repair proteins (36), an increased percentage of A·T base pairs (35, 62), and a decreased number of fusion proteins (K. Mavromatis and N. C. Kyrpides, unpublished results). It has been hypothesized that one reason that genome size reduction occurs in intracellular bacteria is that there is no possibility for lateral gene transfer from other bacteria (54); however, T. pendens does have the opportunity to be exposed to DNA of other bacteria and archaea, and this may help explain the finding that its genome is a normal size. Also, since T. pendens is an extracellular symbiont rather than an intracellular symbiont, it may require a larger genome to deal with environmental perturbations. Some extracellular symbionts (e.g., E. coli, Haemophilus influenzae, and Pasteurella multocida) encode biosynthetic pathways for most nutrients (34), while others lack some pathways. For example, Helicobacter pylori cannot synthesize purines and some amino acids (58); however, T. pendens is unique among extracellular symbionts in the extent of its pathway loss.

Most of the crenarchaeal genomes sequenced so far are genomes of heterotrophs, but they do not show the same extent of loss of biosynthetic pathways as T. pendens. For example, S. acidocaldarius encodes biosynthetic pathways for all amino acids and nucleotides (5), while C. symbiosum has biosynthetic pathways for all amino acids except proline (15). H. butylicus has lost most amino acid biosynthetic pathways, but cofactor synthesis pathways appear to be retained. T. pendens is unique in its loss of pathways for cofactor biosynthesis, as shown in Table 3.

T. pendens does not appear to be parasitic, as it is not known to cause harm to another organism. However, it is limited to growth in nutrient-rich environments, to the point of depending on a specific organism (T. tenax) for an essential nutrient. This type of dependence may be one reason why many microbes cannot be cultivated.

PHX genes in archaea are generally found to be housekeeping genes (21); however, this is not the case in T. pendens. Surprisingly, the T. pendens PHX genes include many CRISPR-associated genes and genes encoding ABC transporters for carbohydrates and peptides. These findings suggest that T. pendens is constantly under attack from viruses in its environment. The large number of CRISPR elements also supports this conclusion. The presence of peptide ABC transporter genes that are PHX genes suggests that T. pendens places a higher priority on nutrient acquisition than on maximization of cell growth and division, which is in agreement with its lack of biosynthetic pathways for most amino acids and cofactors.