INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

The archaeal 16S rRNA tree has four major branches of methanogens, three of which make methane almost exclusively from carbon dioxide. A family of the fourth branch, the Methanosarcinaceae, is exceptional in that representative species such as Methanosarcina barkeri are also capable of methanogenesis from acetate or methylotrophic substrates, such as methanol, methylated thiols, and methylamines (3, 46).

Methylamines are particularly important methane precursors in marine environments, where they arise from the breakdown of common osmolytes (22). Methanogenesis from trimethylamine (TMA) requires the intermediate formation of dimethylamine (DMA) and monomethylamine (MMA), which are subsequently converted to methane (16). The methylation of coenzyme M (CoM) with a methylamine initiates methanogenesis, and as with all substrates, methyl-CoM serves as the direct methane precursor (41, 47).

The different pathways of TMA-, DMA-, and MMA-specific CoM methyl transfer can be reconstituted in vitro with only three highly purified polypeptides. A single protein, MtbA, acts as the common CoM methylase for all three methylamines (11). However, different methyltransferase polypeptides are required to initiate metabolism by demethylation of TMA, DMA, or MMA and subsequent methylation of different corrinoid binding polypeptides. The methylated corrinoid is then demethylated by MtbA to methylate CoM. Each gene or gene product involved in CoM methylation with a methylotrophic substrate is designated according to the following convention. The first two letters, mt, indicate involvement of the gene or gene product in methyl transfer. The third letter indicates the substrate: a for methanol, s for methylthiols, m for MMA, b for DMA, and t for TMA. The final letter designates the polypeptide function, where B is the substrate-specific methyltransferase that methylates the corrinoid protein with substrate, C is the corrinoid binding polypeptide, and A is the CoM-methylating protein.

For MMA:CoM methyl transfer, the specific MMA methyltransferase is MtmB, which uses MMA to methylate its cognate corrinoid protein, MtmC (5). For DMA:CoM methyl transfer, the specific DMA methyltransferase is MtbB (44, 45), which methylates the DMA corrinoid protein, MtbC (D. J. Ferguson, N. Gorlatova, L. Paul, D. A. Grahame, and J. A. Krzycki, submitted for publication). The specific TMA methyltransferase, MttB, copurifies with the TMA corrinoid protein, MttC. MttB has not yet been shown to directly methylate MttC with TMA. However, by analogy with the mechanism of CoM methylation with MMA or methanol, it was proposed that MttB is a TMA methyltransferase that uses TMA to methylate MttC (10). Figure 1 illustrates the functions of the gene products methylating CoM with either DMA or TMA.

View larger version (18K): [in this window] [in a new window]   FIG. 1.   Schematic of the mtt-mtb1 transcriptional unit. Above the gene sequence are indicated the reactions demonstrated for the gene products, i.e., the DMA and TMA methyltransferases and their cognate corrinoid proteins. The function of mttP is proposed but has not been demonstrated. The locations of probes used in S1 protection studies are shown along with the reverse transcriptase PCR product confirming the presence of the UAG codon within the mttB transcript. The M. thermophila mttB gene was amplified by PCR and corresponds to the region indicated above it. Key restriction sites used during cloning of the complete set of genes are indicated.

Methanol:CoM methyl transfer also requires a specific methanol methyltransferase polypeptide, MtaB, which tightly binds and methylates its cognate corrinoid protein, MtaC (7, 36). Methyl-MtaC is then demethylated by a different CoM methylase, MtaA, which methylates CoM. Methylthiol:CoM methyl transfer has been shown to require only two polypeptides (39, 40). In this case, a third CoM methylase, MtsA, appears to methylate a corrinoid binding protein, MtsB, with methylated thiols like dimethylsulfide (MtsB is the only corrinoid protein which is not named according to the above nomenclature rules). MtsA then demethylates methyl-MtsB and methylates CoM.

The genes for MMA- (6), methanol- (35), and methylthiol-dependent (32) CoM methylation have been identified by reverse genetics. These studies have shown that the methylotrophic corrinoid proteins MtaC, MtsB, and MtmC share approximately 50% identity. These methylotrophic corrinoid proteins are also homologous to the cobalamin-binding domain of B₁₂ proteins, such as methionine synthase (28). The methylcobamide-CoM methyltransferases, MtsA, MtaA, and MtbA, are also 50% similar to one another (14, 26, 32). However, MtaB and MtmB, the methanol and MMA methyltransferases, have no significant homology with one another.

A surprising result from the sequencing of the gene encoding the MMA methyltransferase, MtmB, was the presence of a single in-frame amber codon midway through the open reading frame that does not function as a stop codon during translation of the mRNA producing the abundant full-length 50-kDa protein (6). The functionally analogous, but nonhomologous, methanol methyltransferase gene does not contain such an in-frame amber codon. Here, the genes encoding the TMA and DMA methyltransferases and their cognate corrinoid proteins are characterized for the first time. Interestingly, single in-frame amber codons are found to be a common feature of the genes encoding the polypeptides that initiate methanogenesis from TMA, DMA, or MMA.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Nucleotide sequence accession numbers. The mtt-mtb1 operon has been deposited under GenBank accession number AF102623. The nucleotide numbering scheme used in this paper for this operon is relative to the mapped 5' end of the transcript, 304 bp upstream of the translation start of mtbC. The other sequences have been deposited in GenBank under accession numbers AF153453 (mtbB2), AF153454 (mtbB3 and mtbP), and AF153452 (M. thermophila mtbC and mttB, partial sequence).

Organisms. M. barkeri MS and Methanosarcina thermophila TM-1 were grown in PBBM (23) containing 80 mM TMA or 80 mM methanol, respectively. Escherichia coli DH5 containing pUC19 and derivatives was grown in Luria-Bertani broth supplemented with 80 µg of ampicillin/ml (34).

Isolation of nucleic acids. Genomic DNA and total RNA from Methanosarcina spp. were isolated as described earlier (32). Plasmid DNA was isolated by the QIAprep miniprep method (Qiagen Inc., Valencia, Calif.). Ultrafree-MC filter units from Millipore Corporation (Bedford, Mass.) were used to isolate DNA fragments from agarose gels.

Amplification and cloning of an mttC fragment. Degenerate oligonucleotides were designed from the N-terminal amino acids (EAITDFD) of TMA corrinoid protein MttC, GA(A/G)GC(A/G/C/T)AT(A/C)AC(A/G/C/T)GA(C/T)TT(C/T)GA, and a portion of the corrinoid binding signature (HDIGKNI) of MtsB, AT(A/G)TT(C/T)TT(A/G/C/T)CC(A/G/T)AT(A/G)TCGTG. These primers and Taq DNA polymerase were used to amplify a fragment of the gene mttC from the genomic DNA of M. barkeri MS. The PCR products were probed with an oligonucleotide derived from the N-terminal MttC sequence internal to the PCR primers labeled at the 5' ends with [-³²P]ATP (ICN Pharmaceuticals, Inc., Costa Mesa, Calif.), using T4 polynucleotide kinase (GIBCO BRL). PCR products approximately 300 bp long were cloned into competent E. coli DH5 (19) using pGEM-T vector (Promega, Madison, Wis.) by standard methods (34).

Southern and Northern hybridizations. Following the appropriate electrophoresis method (34), RNA or DNA was transferred electrophoretically to nylon membranes (Schleicher and Schuell, Keene, N.J.) and UV cross-linked. Hybridizations employed the sandwich method (20) in 0.36 M NaCl, 20 mM NaH₂PO₄, pH 7.4, 2 mM EDTA, and 0.5% sodium dodecyl sulfate. Autoradiograms were prepared using Biomax MS X-ray film (Eastman Kodak Company, Rochester, N.Y.). The sizes of the hybridizing bands were determined using a 1-kb DNA ladder or the RNA molecular weight standard from GIBCO BRL. Oligonucleotide probes were 5' end labeled, while DNA fragments were random primer labeled using the Prime-a-gene kit (Promega) and [-³²P]dATP (Amersham, Arlington Heights, Ill.).

Analysis of mttB transcript around the UAG codon. RNA (10 µg) was treated with 10 U of DNase, denatured at 90°C for 5 min, and then frozen on dry ice. A 5'-end-labeled oligonucleotide corresponding to the region +2050 to +2068 (3 pmol) was annealed to the denatured RNA at 42°C for 15 min, and then 2.5 U of reverse transcriptase was added and incubation was continued for 45 min. The product was purified by QIAquick (Qiagen) and was amplified by PCR using primers corresponding to +2050 to +2068 and +1518 to +1533. The purified PCR product was sequenced in both directions using the primers corresponding to the regions +1810 to +1826 and +2050 to +2068.

Sequencing methods. Both strands of the described DNA were sequenced. Nested deletion clones were generated in one direction using exonuclease III (15), and then internal primers were used to confirm the sequence in the opposite direction. For manual sequencing, the Sequenase version 2 deaza kit (United States Biochemicals Corp., Cleveland, Ohio) and [-³⁵S]dATP (Amersham) was used, followed by 6% acrylamide-8 M urea gel electrophoresis and autoradiography. Automated sequencing was carried out by dye terminator cycle sequencing using AmpliTaq polymerase and an ABI PRISM 310 genetic analyzer (Perkin-Elmer, Foster City, Calif.).

Mapping of the transcription start site. Primer extension was done with a primer complementary to RNA from positions +396 to +413 and SUPERSCRIPT II reverse transcriptase (GIBCO BRL). RNA (20 µg in distilled H₂O) and the 5'-end-labeled primer (0.4 pmol) were denatured at 85°C for 10 min and allowed to cool to 42°C over a period of 30 min. The extension reaction was carried out at 42°C for 50 min. A 240-bp EcoRV/EcoRI fragment (136 to +104) was used to determine the 5' end of the transcript by using S1 nuclease (34) incubated at 37°C with 12.5 ng of 5'-end-labeled probe and 50 µg of RNA. A pUC19 sequencing ladder made with reverse primer (Promega) served as a size standard.

Amplification of mttB. The mttB gene from M. thermophila was amplified using primers corresponding to the regions +760 to +775 and +2434 to +2464 with Vent DNA polymerase (New England Biolabs). The PCR product was sequenced directly.

Computer-aided sequence analysis. Homology searches were conducted using the BLAST program (1) maintained on the National Center for Biotechnology Information (NCBI) server. CLUSTAL W (42) maintained at Baylor College of Medicine, was used to align multiple sequences. Transmembrane regions were determined with the TMpred program maintained at ISREC (17).

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Cloning of the genes surrounding mttC. The 5' end of mttC, encoding the trimethylamine corrinoid protein, was amplified from genomic DNA by PCR and cloned into E. coli. The PCR primers were designed with the N-terminal sequence of purified MttC and a portion of the corrinoid binding motif of MtsB (32). This fragment was used as a probe to clone a 4.0 kb-KpnI DNA fragment with all of mttC, which encoded a 23-kDa polypeptide beginning with the N-terminal sequence of purified MttC (ANKEEIIAKAKEAITDPDDELAEEVANEALAAGI). Probing genomic digests with an oligonucleotide specific for the region upstream of mttC (+1333 to +1347) revealed a single hybridizing 3.4-kb SphI/SalI fragment. The cloned fragment was found to overlap the KpnI fragment by 40 bases. The sequences of both contiguous restriction fragments show that mttB, encoding the TMA methyltransferase, is directly upstream of mttC. The 5' end of mttB encodes AKNNAVAGFNALNGVEL, the N-terminal sequence of the TMA methyltransferase. A third gene, mtbC, encoding the DMA corrinoid protein, was found directly upstream of mttB, while a fourth gene, mttP, was found directly downstream from mttC (Fig. 1).

Three DMA methyltransferase genes exist, one immediately downstream of the mtt genes. The first 20 residues of the purified predominant DMA methyltransferase from M. barkeri MS, MATEYALRMGDGKRVYLTKE (Ferguson et al., submitted) were found to be encoded by a gene 523 bp downstream of mttP near the 3' end of the KpnI fragment. The predicted protein sequence also matched 14 of 16 amino acid residues of the N terminus of the DMA methyltransferase from M. barkeri Fusaro (45). In order to complete cloning of the DMA methyltransferase gene linked to the TMA methyltransferase genes, genomic DNA restriction digests were hybridized with TCP10, an oligonucleotide specific for the region just downstream of the N terminus-encoding sequence (+5215 to +5237). Surprisingly, three HindIII fragments of 4.0, 2.5, and 2.1 kb were identified that strongly hybridized to TCP10. All three HindIII fragments were cloned, and the sequences averaged 93% identity for the first kilobase. One 5' end of the 4.0-kb HindIII fragment was completely identical to the last 80 bp of the KpnI fragment, while the other HindIII fragments had a single-base-pair mismatch with this sequence.

View larger version (15K): [in this window] [in a new window]   FIG. 2.   Three nearly identical copies of the DMA methyltransferase gene were cloned as HindIII fragments from M. barkeri MS and sequenced. In order to establish that mtbB1 is linked to the mtt genes, a restriction map of genomic DNA was made using Southern blots probed with oligonucleotides TCP10 and 18A (+4837 to +4852) (A). The map was compared with restriction sites sequenced in the 4.0-kb HindIII fragment (B), the 2.5-kb HindIII fragment (C), or the 2.0-kb HindIII fragment (D). Unsequenced DNA in the 2.5-kb HindIII fragment is indicated by the dashed line. Restriction sites are indicated as follows: Hd, HindIII; Hn, HincII; S, SalI; RI, EcoRI; D, DraI; RV, EcoRV; K, KpnI.

Cotranscription of mtbC, mttB, mttC, mttP, and mtbB1. Probes specific to all five genes were hybridized to blots with total RNA from TMA-grown cells (Fig. 3). A 6.8-kb transcript was detected with all gene probes, while probes to mtbC, mttB, and mttC detected an abundant 3.2-kb transcript. A probe immediately upstream of the transcription start site (determined below) did not hybridize to any RNA in these same blots. Probes corresponding to genes closer to the 5' end of the 6.8-kb transcript hybridized to smaller bands. The 5.0-kb band was detectable with probes corresponding to all the genes except mtbB1. There was also a 4.0-kb band whose 3' end was predicted to be within mttP. Probes to mtbC and the 5' untranslated region revealed smaller bands which were ~2.0 and ~1.0 kb in size. This pattern is consistent with transcript degradation beginning at the 3' end of the longest message.

View larger version (54K): [in this window] [in a new window]   FIG. 3.   Northern blot analysis of transcripts produced from the DMA and TMA methyltransferase gene cluster. The hybridization targets for the probes are indicated by the lines connected to the boxed probe numbers above the resulting autoradiograms. The oligonucleotide probes were complementary to the following regions: 1, 97 to 115; 2, +181 to +196; 3, +396 to +413; 4, +833 to +848; 6, +2434 to +2464; 7, +2569 to +2591; 8, +4095 to +4114; 9, +4786 to +4800; and 10, +5215 to +5237. Probe 5 (+1299 to 1644) was a 5'-end-labeled fragment of mttB. Size standards are indicated to the right in kilobases. The migrations of standards, which varied slightly for different blots, are indicated by horizontal lines next to each blot.

Mapping of the mtt-mtb1 transcript start site. A primer close to the 5' end of mtbC gave a single extension product of about 420 bp. From this approximate start site, a 240-bp EcoRV/EcoRI probe was identified for S1 analysis. Only two closely spaced protected bands of equal intensity were detected after S1 digestion, corresponding to a transcript start site 303 or 304 bp from the translation start of mtbC (Fig. 4). There was no protected fragment when E. coli tRNA was used in place of M. barkeri RNA. A 12-bp sequence (GAATAATCGTGA; +226 to +237 and +240 to +251) is directly repeated in the extended leader region of the transcript. Three sets of indirect repeats are also in the transcript leader that could form stem-loops encompassing regions +17 to +78, +84 to +114, and +180 to +205. There is a putative promoter sequence (TATATA) 21 bp upstream of the mapped 5' end of the transcript.

View larger version (53K): [in this window] [in a new window]   FIG. 4.   Mapping of the mtt-mtb1 transcript start site by S1 analysis. Total cellular RNA was denatured and cooled in the presence of the radiolabeled restriction fragment shown in Fig. 1 and as described in Materials and Methods. The protected fragments of 103 and 104 nucleotides correspond to 303 and 304 bp from the translational start site of mtbC. The sequencing ladder was generated from pUC19. E. coli tRNA was used in place of M. barkeri RNA in the control lane (C).

Conservation of sequences in and preceding the mtt-mtb1, mtmCBP, and mtbA operons. As shown in Fig. 5, there is a 21-bp sequence that ends 99 bases 5' of the mapped start site of the mtt-mtb1 transcript which is highly similar to a sequence found 43 to 21 bases before the mapped transcript start site of the mtmCBP operon (6). The same sequence is also partially conserved 42 to 22 bases upstream of the mapped transcription start site of mtbA (6). In the cases of mtmCBP and mtbA, the conserved sequence ends with the putative promoter sequences identified previously by their spacing from their mapped transcript start sites using RNA from cells grown on MMA. However, the conserved sequence is unlikely to serve as the promoter giving rise to the 5' end of the mtt-mtb1 transcript detected in the experiments above. The transcript start site is 99 bases from the 3' end of the conserved sequence (Fig. 5), and most archaeal promoters are found within 27 (±4) bases of the transcript start site (33, 48).

View larger version (22K): [in this window] [in a new window]   FIG. 5.   Schematic of the upstream region and transcriptional start site of the mtt-mtb1 unit. In the lower part of the figure is the nucleotide sequence upstream and around the transcript start site. The first T of the two thymine bases identified by S1 mapping (boldface italics) is considered as the start of the transcript and is located 304 bases from the mtbC translation start site. As seen in the upper part of the figure, 99 bases upstream from the mtt-mtb1 transcript start site there is a highly conserved sequence in the regions 5' of the mtmCBP and mtbA transcript start sites. The location of this sequence relative to the mapped start sites of these transcripts is indicated to the left of each sequence. The putative promoters for the mtbA, mtmCBP, and mtt-mtb1 transcripts are underlined.

The TMA methyltransferase gene mttB from either M. barkeri or M. thermophila contains an in-frame amber codon. The determined N terminus of MttB is encoded by mttB, but a UAG codon, which was represented in both strands of the DNA, follows at codon position 334 (Fig. 1). Cessation of translation at the UAG codon would result in a 34-kDa product rather than the abundant 53-kDa polypeptide purified from TMA-grown cells (10). However, following the UAG codon, the open reading frame continues for another 483 bp before ending with a UAA codon. The predicted molecular mass of the gene product that would be produced if translation ceased at the UAA codon is 54 kDa. Numerous UAA and UGA codons are found in other reading frames both before and after the in-frame amber codon.

View larger version (87K): [in this window] [in a new window]   FIG. 6.   Comparison of the deduced amino acid sequence of mttB amplified from M. thermophila (Mth) and mttB cloned from M. barkeri (Mb). The UAG codon position (at position 334) is indicated by X in both sequences. Identical residues are boxed.

Presence of the amber codon in transcripts encoding mttB. In order to test if the UAG codon was actually present in mttB transcripts, RNA isolated from cells grown on TMA was incubated with a primer complementary to mttB and reverse transcriptase. The cDNA product was then amplified by PCR using primers complementary to sequences on both sides of the UAG codon (Fig. 1). The PCR product was then sequenced directly and had the same sequence as mttB, including the UAG codon. Reactions without reverse transcriptase gave no PCR product, indicating that the amplified product did not arise from contaminating DNA in the RNA preparation.

All three copies of the DMA methyltransferase gene contain an in-frame amber codon. An in-frame amber codon was also discovered in the reading frame of mtbB1. This codon was represented in both strands of the DNA. The molecular mass of the purified DMA methyltransferase isolated from TMA-grown M. barkeri strain Fusaro (45) or MS (Ferguson et al., submitted) is 50 kDa. The mtbB1 gene product would be only 38 kDa if the single in-frame UAG codon at position 356 acts as the translation stop. However, if translation proceeds through the in-frame UAG codon to the following UGA codon 333 bases later, then a 50-kDa mtbB1 gene product is predicted. This indicates that the DMA methyltransferase gene mtbB1 contains a single in-frame amber codon within the gene itself that does not act as a translation stop during expression of the 50-kDa methyltransferase.

View larger version (48K): [in this window] [in a new window]   FIG. 7.   Predicted gene products of available sequence from mtbB1, mtbB2, and mtbB3. The complete sequence of MtbB1 is shown and is numbered on the left-hand side. The first 48 residues shown for MtbB1 are not presently available for MtbB2 or MtbB3. The sequences of MtbB2 and MtbB3 are identical to that of MtbB1 for the next 100 residues; different residues that follow are indicated in the figure for MtbB2 or MtbB3 above and below, respectively, the MtbB1 sequence. The X at position 356 marks the UAG codon found in all three genes. The nucleotide sequence following the corresponding UAA or UGA codon of each gene is shown for comparison (lowercase).

C-terminal sequencing of purified DMA methyltransferase. The three copies of the DMA methyltransferase gene, mtbB1, mtbB2, and mtbB3, have the same predicted C-terminal sequence if translation of the mRNA does not end at the in-frame amber codons found in each gene and instead proceeds to the next canonical stop codon (Fig. 7). A DMA methyltransferase, MtbB1, has been purified from TMA-grown cells (Ferguson et al., submitted), and C-terminal sequencing of this DMA methyltransferase was undertaken. The sequence obtained was NLFXKQIA, where X was a residue that could not be assigned. These residues match the C termini of the gene products predicted from the sequences of all three DMA methyltransferase gene copies if translation ended at the UAA or UGA codons following the in-frame UAG codon sequenced in each gene.

The methylamine methyltransferases are not homologous. No significant deduced sequence similarity among the three methyltransferases specific for TMA, DMA, and MMA (MttB, MtbB, and MtmB, respectively) was found using the BLAST programs at the NCBI. The deduced amino acid sequences of both the TMA and DMA methyltransferase lack the complete motif of corrinoid binding found in the methanogenic methylotrophic corrinoid proteins.

The DMA and TMA methyltransferase cognate corrinoid proteins. MttB and MttC were previously found to copurify and were required to methylate CoM with TMA. However, the two were difficult to separate, and therefore it could not be demonstrated that both were required for TMA:CoM methyl transfer. The genes encoding both proteins are found adjacent and cotranscribed. This reinforces the previous biochemical data showing that both proteins are required for TMA:CoM methyl transfer (10).

Putative membrane proteins are encoded near the genes for TMA and DMA methyltransferases. Transmembrane proteins are predicted to be encoded by mttP and mtbP. MttP is encoded on the mtt-mtb1 transcript. It is predicted to be a highly hydrophobic protein spanning the membrane nine times. BLAST searches found a number of homologs to MttP; however, all were proteins of unknown function also predicted to be integral membrane proteins. Next to one of the DMA methyltransferase genes, mtbB3, is mtbP. MtbP is predicted to be a membrane protein most similar to permeases for cationic amines from different sources. For example, MtbP had 34% identity over a region of 84 amino acids (residues 31 to 113 in MtbP) with the YecA protein, a putative amino acid permease from Bacillus subtilis (expect value, 6e-06). CAN1, a known cationic amino acid permease from Candida albicans, had 40% similarity with residues 23 to 154 in MtbP (expect value, 0.02). The reading frame of mtbP ended with a UAG codon, which was followed by a UAA codon 12 bp later.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

Most methanogens generally lack the catabolic versatility so often found in prokaryotes and produce methane only from carbon dioxide. The acquisition of methylamine methyltransferase genes, including those described here, by Methanosarcina spp. was a significant step towards the diversification of the substrate range that is characteristic of this genus and its near relatives.

Biochemical analysis has shown that CoM methylation is initiated with TMA, DMA, or MMA by proteins with the N termini predicted for the products of the mttB, mtbB1, and mtmB genes, respectively (5, 6, 11; Ferguson et al., submitted). Each of these methyltransferases specifically methylates a cognate corrinoid protein with its substrate methylamine. Each methylamine methyltransferase is approximately 50 kDa, but perhaps contrary to expectation, these proteins performing analogous functions have little deduced sequence similarity. This provides a rationale as to why the different, yet homologous, cognate corrinoid proteins have evolved for TMA, DMA, and MMA metabolism. The homologous cognate corrinoid proteins serve as methyl acceptors for very different methyltransferases while still interacting with a single CoM methylase, MtbA, for methylamine-dependent methanogenesis. Homologs of this group of small corrinoid proteins have recently been found in bacteria that are involved in the catabolism of methoxylated aromatics (21) or chloromethane (43).

MtmB, MtbB1, and MttB have the common trait of being encoded by genes with in-frame amber codons. Recognition of the UAG codon in each gene as a stop codon would result in truncated proteins of 38 (MtbB1), 32 (MttB), and 23 kDa (MtmB). However, the measured molecular mass of each isolated methyltransferase is 51 (MtbB1), 52 (MttB), and 50 kDa (MtmB). To obtain products of this size, the in-frame UAG codons must be read through and translation must continue to the following UAA or UGA codons. This clearly occurs in the DMA methyltransferase, MtbB1, isolated from TMA-grown cells. The C terminus of this protein matches that predicted by all three DMA methyltransferase gene copies, but only if translation passed through the UAG codon and continued to the next canonical stop codon. The presence of amber codons is unprecedented in the other genes that encode methyltransferases for methylthiol-dependent (32, 40) and methanol-dependent (35) methanogenesis. Indeed, in-frame UAG codons interrupting the reading frames of known genes were not noted in the recent sequencing of two different methanogen genomes (4, 37). The restriction of in-frame amber codons to the methylamine methyltransferase genes indicates that UAG in methanogens does not serve as a global sense codon in methanogens. Some protists, for example, use UAG to encode up to 16% of the glutamine in cellular protein (12), and this is clearly not occurring in methanogens.

A lower limit of the abundance of the methylamine methyltransferase is given by their recovery during isolation from methylamine-grown cells. From 0.5 to 1.6% of the total soluble protein of cells grown on methylamine was recovered as the 50-kDa TMA, DMA, or MMA methyltransferases (5, 10, 45; Ferguson et al., submitted). Since this is the yield of purified protein, and purification inevitably entails loss of protein, the actual amount of the full-length methyltransferases in methylamine-grown cells is likely to be considerably higher. These considerations indicate an active mechanism by which UAG-directed termination of translation is suppressed during expression of the 50-kDa methylamine methyltransferases. The truncated products may also be produced by recognition of the UAG codon as a stop codon. However, inspection of sodium dodecyl sulfate gels of total soluble protein from MMA-grown cells indicate no prominent band of 20 to 25 kDa, which would result if the UAG codon in mtmB were recognized as a stop codon, while a predominant 50-kDa band comigrates with purified MtmB (5).

Several mechanisms by which termination at the in-frame UAG codons is suppressed during translation of the 50-kDa methyltransferases can be considered. For example, it is unlikely that the amber codons represent mutations that have been suppressed by second-site mutations. The amber codon is at corresponding positions in the methylamine methyltransferase genes sequenced from different strains (6) and species (this work). It is also unlikely that the UAG-containing methylamine methyltransferase genes are not functional and that other methyltransferase gene copies without amber codons exist in the genome. Extensive probing of genomic DNA blots revealed only a single TMA methyltransferase gene copy, which contains the UAG codon. Three copies of the DMA methyltransferase gene were found, but all three contain UAG codons at the same position. The aligned sequences of the DMA methyltransferase genes are nearly identical both before and after the amber codon until they diverge immediately after a UAA or a UGA stop codon. This pattern of sequence conservation is further evidence that translation of the DMA methyltransferase genes extends through the UAG codon.

Several examples have been found in both prokaryotes and eukaryotes in which stop codons in transcripts are simply bypassed, for example, by frameshifts or by translational hopping (9). However, the other reading frames of the mtmB, mttB, mtbB1, mtbB2, and mtbB3 genes contain numerous UAA and UGA stops, which makes a reading frame shift an unlikely mechanism to circumvent a premature stop.

Suppression of UAG-directed termination appears to occur at the level of translation itself. As we have shown here, transcripts of mttB with the in-frame UAG are present in TMA-grown cells. In the case of mtmB, the tryptic fragment of MtmB encoded by the UAG-containing region was recently sequenced by Edman degradation (C. James, L. Paul, G. Srinivasan, S. Burke, T. Hill, and J. Krzycki, unpublished data). The predicted amino acid sequence encoded immediately before and after the in-frame UAG codon was confirmed. A residue was found at the UAG position whose identity is now undergoing confirmation by mass spectroscopy. These results indicate that UAG within this methylamine methyltransferase gene is not bypassed, but translated. In-frame amber codons that are translated have not been found in other genes of any member of the Archaea, methanogenic or otherwise. One possible rationale is that an unusual, or a usual, amino acid is encoded by the UAG codon, which operates in a specialized regulatory and/or catalytic role involving the methylamine methyltransferases. An obvious precedent for an unusual amino acid exists in the UGA-directed insertion of selenocysteine (2). A rare sense codon for leucine is used as a regulatory mechanism for sporulation genes in Streptomyces species (25, 27). Many examples have also been found of nonsense codon suppression in which a natural tRNA serves to decode both a normal sense codon and the stop codon (13, 18, 24). In any case, the single in-frame UAG codon in each of these methylamine methyltransferase genes must be related in some way to the analogous function of these genes, that is, encoding the proteins responsible for initiating metabolism of methylamines by this methanogen.

The methylotrophic genes of Methanosarcina spp. are some of the best candidates for study of methanogen regulatory mechanisms. Comparison of the methylamine operons reveals some intriguing sequence conservation with implications for regulation. Nearly 100 bases upstream of the mapped putative promoter site of the mtt-mtb1 transcript is a 21-bp sequence nearly identical to a sequence upstream of the mtm transcriptional unit. This upstream sequence contains the putative promoter sequence of the mtmCBP transcript (6). The sequence is also found less perfectly conserved in front of the mapped promoter of the mtbA transcript (6). Both these transcripts were mapped using RNA from cells grown solely on MMA. The use of the conserved sequence in front of the mtt-mtb1 operon as a promoter in TMA-grown cells was not detectable in our experiments, and a different putative promoter was mapped. However, RNA from cells during early log growth on TMA was used. During later stages of growth on TMA, MMA and DMA accumulate in the medium (30). Under these and other growth conditions, coordinated regulation of different methyltransferase operons may be necessary. Similar sequences near or containing promoters are often used in such coordinated regulation schemes.

Multiple copies of genes are relatively rare in prokaryotes, and it is rarer still to find nearly identical multiple copies, such as the mtbB genes. These results recall the recently discovered extremely close duplicates of catabolic genes, such as those encoding methane monooxygenase (38) or ammonia monooxygenase (31). It has not yet been demonstrated that all copies of mtbB are expressed. No direct evidence of expression of mtbB2 or mtbB3 was obtained, since probing of RNA from TMA-grown cells with an mtbB1 fragment unequivocally revealed only the 6.8-kb transcript of the mtt-mtb1 operon. It is possible that the other two gene copies give rise to low-abundance transcripts that are obscured by the degradation of the larger 6.8-kb transcript and/or are expressed under growth phases or culture conditions different from those tested here. However, the similarity of the DMA methyltransferase genes indicates that all three genes are probably expressed under some conditions. Half of the residue substitutions found in mtbB1, mtbB2, and mtbB3 are conservative, and most of the nucleotide substitutions are at the third position of the codon. This is consistent with selective pressure to maintain the same reading frame in these genes, presumably in order to maintain functional methyltransferases following expression.

The mtm operon contains mtmP, whose predicted gene product is very similar to members of the APC family of transporters specific for cationic amines (6). MtmP has been suggested to be an MMA permease. Interestingly, we found that two more open reading frames predicted to encode transmembrane proteins are adjacent to genes encoding methylamine methyltransferases. MttP is encoded between mttC and mtbB1. A gene encoding a second putative transporter with similarity to cationic amino acid transporters, mtbP, is found following mtbB3. These putative transmembrane proteins may represent transporters with the highest affinity for the methylamine-specific methyltransferases which are adjacent to them on the genome and represent some of the first candidates for transport proteins involved directly with methanogenic substrates.