In the third domain of life, the Archaea, the transcription machinery is composed of a multi-subunit RNA polymerase that shares homology to the eukaryotic RNA polymerase II, as well as two general transcription factors: the TATA-box binding protein (TBP) and transcription factor B (TFB) (Bell and Jackson, 2001; Jun et al., 2011; Gehring et al., 2016). The corresponding DNA elements of a basal archaeal promoter includes a purine-rich transcription factor B recognition element (BRE), which is recognized by the TFB, immediately followed by a TATA box centered at a distance of 26/27 bp upstream of the transcription start site (TSS) (Soppa, 1999; Bartlett, 2005; Gehring et al., 2016). To initiate transcription, TBP first binds to TATA box. This is followed by the binding of TFB to the DNA-TBP complex by recognition of the BRE sequence (Bell et al., 1999) and, finally, the recruitment of RNA polymerase to initiate transcription (Bell and Jackson, 2001). Mutations in either the TATA box or BRE can decrease transcription levels by reducing recruitment of TBP and TFB (Bartlett, 2005).

Although Archaea use a eukaryote-like basal transcription machinery, the genome structure and its transcription regulation are more like that found in Bacteria. In Archaea, a cluster of genes is co-transcribed into a poly-cistronic mRNA under the control of a single promoter, which can be regulated by repressors and/or activators (Peeters et al., 2013). Transcriptional activators typically bind to sites located upstream of the BRE and help in the recruitment of TBP or TFB. In contrast, repressors can bind to either the promoter region where they interfere with TFB or TBP binding by steric hindrance, or downstream of the promoter, sometimes even after the TSS, to prevent RNA polymerase recruitment or transcription elongation (Bell, 2005; Peeters et al., 2013; Karr, 2014). Transcriptional activators are often associated with promoters that have TATA box or BRE sequences that deviate from consensus sequences (Ochs et al., 2012; Peeters et al., 2013). They are believed to help overcome poor binding of TBP or TFB to weak TATA and BRE sequences to activate transcription (Ouhammouch et al., 2003; Peng et al., 2009; Ochs et al., 2012).

The methanogen Methanococcus maripaludis is a member of the phylum Euryarchaeota and a model organism for studies in Archaea. Here, the fla operon, encoding the components of the archaeal swimming organelle, the archaellum (Jarrell and Albers, 2012; Albers and Jarrell, 2015), begins with flaB1-B3 encoding the three major structural proteins (archaellins), followed by the fla-associated genes flaC-J (Chaban et al., 2007). Transcription of the fla operon is controlled by the transcriptional activator EarA (Ding et al., 2016b). Deletion of earA results in almost undetectable transcription of the fla operon and a corresponding disappearance of FlaB2 protein and archaella production (Ding et al., 2016b). Immediately upstream of the BRE in the fla promoter, four 6 bp consensus sequences were identified as EarA binding sites. When all four EarA binding sites were eliminated in the genome of wild-type Mc. maripaludis, similar phenotypes were observed as in the ΔearA mutant (Ding et al., 2016b). Recently, we have shown that EarA homologs from selected archaellated methanogens could successfully complement the function of EarA in the Mc. maripaludis ΔearA mutant, indicating that the EarA regulatory model is likely widespread in the methanogen fla promoters (Ding et al., 2017).

In addition to the direct control of transcription of the fla operon by EarA, transcription of the fla operon was also found to be regulated under several growth conditions. Global transcriptome analysis of Mc. maripaludis showed that the transcription of the fla operon is up-regulated when H₂ is limited and down-regulated under leucine starvation, for example (Hendrickson et al., 2008). In addition, we recently showed that transcription of the fla operon was severely impaired in cells grown at temperatures greater than 38°C (Ding et al., 2016a). The mechanism behind the regulation of the fla promoter under the above conditions, including any possible involvement of EarA or other putative transcriptional activators or repressors, is yet to be reported.

In this study, we isolated a spontaneous mutant of the ΔearA mutant in which transcription of the fla operon, production of archaellins and archaellation were all restored to near wild-type levels, despite the absence of EarA. Analysis of the fla promoter region of this mutant revealed a deletion of three adenines in the BRE. Recreation of the three adenine deletion in the original ΔearA mutant by molecular biology techniques resulted in very similar archaella-related phenotypes as observed in the spontaneous mutant. Examination of the fla promoter wild-type BRE and the three adenine deletion BRE revealed that the mutant BRE were highly similar to BRE sequences associated with promoters with strong basal transcription levels in both Mc. maripaludis and a related hyperthermophilic methanogen Methanocaldococcus jannaschii.

Previous studies have shown that the euryarchaeal archaellum regulator EarA was critical for transcription of the fla operon in Mc. maripaludis via its binding to at least one of four consensus sequences located immediately upstream of the BRE and TATA box of the fla promoter. In a ΔearA mutant, transcription of the fla operon is barely detectable and cells are non-archaellated (Ding et al., 2016b). In this study, we have isolated a spontaneous mutant of a ΔearA strain in which the transcription of the fla operon and archaellation were restored. Analysis of the DNA sequence of the fla promoter region in this mutant, designated ΔearA-sp, revealed a deletion of three adenines in the BRE region. Recreation of the three adenine deletion in the stock strain of the ΔearA mutant also led to restoration of fla operon transcription and archaellation, indicating that this small deletion in the BRE overcame the requirement for EarA for activation of transcription of the fla operon. However, the expression of FlaB2 detected by western blotting was lower in the recreated strain than in the spontaneous mutant ΔearA-sp, suggesting that the three adenine deletion may not be the sole change in the ΔearA-sp strain affecting transcription of the fla operon. However, it seems clear from our studies on the directed mutant Δ3A strain, that the deletion of three adenines in the BRE of the fla operon promoter is sufficient on its own to result in all the phenotypes related to archaellation observed in the spontaneous mutant.

Since there is virtually no transcription detected from the native fla promoter if earA is deleted, it suggests that the fla promoter is intrinsically very weak or inactive. Two key elements that determine promoter strength in Archaea are the sequences of the TATA box and BRE (Bartlett, 2005). The TATA box is the site of binding of the TATA-binding protein TBP while the BRE sequence is the site of binding for TFB (Peeters et al., 2013). While relatively few transcriptional activators have been studied in Archaea, the mechanism of activation in these limited studies has been shown to involve recruitment of TBP or TFB to the TATA box or BRE (Karr, 2014). Consensus TATA box sequences vary for different subgroups of Archaea and mutations in the TATA box can reduce transcription efficiency (Soppa, 1999; Bartlett, 2005; van de Werken et al., 2006). For protein promoters in Mcc. jannaschii, the TATA box was determined to be TWTATATA (where W = A or T) (Zhang et al., 2009), very similar to the TTTATATA proposed previously for the promoters of stable RNA genes in Methanococcus vannielii (Thomm and Wich, 1988) and featuring the methanogen characteristic of strict alterations of T and A in contrast to TATA boxes in other major archaeal groups (Soppa, 1999). One of the best-studied archaeal transcriptional activators, Ptr2 of Mcc. jannaschii, binds to multiple sequences upstream of BRE in the rubredoxin 2 gene and has its stimulatory effect due to direct recruitment of TBP (Ouhammouch et al., 2003, 2005). Adding binding sites for Ptr2 upstream of heterologous promoters with sub-optimal TATA box sequences resulted in significant transcriptional activation (Ouhammouch et al., 2005). Analysis of the TATA box of the fla operon in Mc. maripaludis revealed a strong identity to the consensus sequence, including the alternating T and A stretch TATATAT, suggesting binding of TBP should not be impaired.

The 6–7 nucleotide long BRE sequences are the major site of binding for TFB, with positions -3 and -6 of BRE (relative to the TATA box) showing the strongest specificity determinants (Qureshi and Jackson, 1998; Littlefield et al., 1999). There are no BRE consensus sequences reported for halophiles and methanogens (van de Werken et al., 2006). However, in Mcc. jannaschii, a hyperthermophilic relative of Mc. maripaludis, two studies have identified promoter sequences on a whole genome basis (Li et al., 2008; Zhang et al., 2009). The first study used the binding of TBP and TFB in EMSA studies to identify promoters (Li et al., 2008). These studies had a strong bias for strong promoters, especially for promoters of tRNA genes with only small percentage of promoters for protein genes being retrieved. These studies led to the identification of an extended BRE element sequence of 9–10 nucleotides (MRCCGAAAAG where M = A, C and R = A, G). The second study focused on identification of promoters for protein-encoding genes (Zhang et al., 2009). It was found for Mcc. jannaschii protein gene promoters that there was a greater variability in the BRE than in the TATA box (Zhang et al., 2009). The identified promoters for protein-encoding genes were shown to bind the general transcription factors less tightly than tRNA gene promoters. Notably, base frequencies at several BRE positions considered important for TFB binding were significantly different from the in vitro selected promoters (mostly for tRNA genes) in the earlier study. Examination of the BRE sequences in both protein-encoding genes and tRNA genes revealed that most had internal stretches of 3–5 adenines, far less than the seven adenines in the wild-type fla operon promoter. Interestingly, the fla operon BRE element has a G at position -1, the most commonly found base at that position in the strong tRNA gene BRE (Figure 6A), while in protein-encoding genes the most common base at -1 is C (Zhang et al., 2009).

The wild-type version of the BRE of the fla operon promoter, with its stretch of seven adenines, does not show strong sequence identity to what may be considered strong BRE sequences as reported for Mcc. jannaschii. On the other hand, it is apparent that the three adenine deletion version found in the ΔearA-sp mutant much more closely aligns with BRE sequences found in strong tRNA gene promoters of Mcc. jannaschii (Figure 6A). In addition, the mutated fla promoter in the ΔearA-sp strain shares high sequence identity with two studied promoter sequences in Mc. maripaludis, namely the nitrogen-regulated glnA and nifH promoters (Figure 6A) (Cohen-Kupiec et al., 1997, 1999). Both glnA and nifH promoters are regulated via the repressor NrpR, which binds to the nif operators located downstream of the TATA boxes just after the TSS in the two promoters leading to repression of transcription under ammonia growth conditions (Cohen-Kupiec et al., 1997, 1999; Lie et al., 2005). Both nifH and glnA expression is very low when cells are grown on ammonia and NrpR binds but high expression is observed under conditions of diazotrophic growth where NrpR does not bind or in a strain where nrpR has been deleted (Cohen-Kupiec et al., 1997, 1999; Lie and Leigh, 2003; Lie et al., 2005). This indicated that the basal transcription level of the two promoters was strong, suggesting that TFB and TBP in Mc. maripaludis could recognize BRE and the TATA box of these two promoters and initiate transcription (Cohen-Kupiec et al., 1997, 1999). The high sequence identity of the three adenine deletion BRE of the ΔearA-sp strain with that of the glnA and nifH promoters, as well as the BRE of the highly expressed tRNA genes of Mcc. jannaschii likely explains why the pre-initiation complex could be formed with the mutated fla promoter without the aid of EarA. The qRT-PCR results (Figure 3B) suggest that the wild-type fla operon promoter even with EarA is not as strong as the three adenine deletion version in the absence of EarA.

Studies in several archaea have indicated that promoters containing non-conserved BRE sequences can be weak or even inactive (Peng et al., 2009, 2011; Marschaus and Pfeifer, 2012; Ochs et al., 2012). Replacement of the BRE of inducible promoters with a BRE from strong promoters, for example, can greatly increase the transcription from the resulting chimeric promoter. In Sulfolobus solfataricus, transcription from the arabinose promoter is induced in the presence of arabinose, via an unidentified factor that binds to a consensus ara-box sequence located immediately upstream of the BRE and TATA box (Lubelska et al., 2006; Peng et al., 2011). When the BRE from the arabinose promoter was replaced with the strong BRE from the Sulfolobus shibatae viral (SSV) T6 promoter (Qureshi and Jackson, 1998), the resulting chimeric promoter was now constitutive and not regulated by the ara-box element (Peng et al., 2009). The apparent mechanism of transcription activation of the ara-box binding factor is thought to be by recruitment of TFB to a weak BRE (Peng et al., 2009, 2011). In Pyrococcus furiosus, transcription from the pf1089 promoter is activated by PF1088 (Transcription factor B recruitment factor 1, TFB-RF1). This activation is dependent on the weak BRE of the pf1089 promoter and is not observed if the pf1089 promoter BRE is replaced with the BRE from the strong gdh promoter (Spitalny and Thomm, 2003; Ochs et al., 2012). Electrophoretic mobility shift assays further revealed that the transcription activation of the wild-type pf1089 promoter was by the recruitment of TFB via TFB-RF1, thereby overcoming the weak BRE.

We have recently shown that EarA homologs are commonly found in the Euryarchaota and that EarA proteins from numerous methanogens can rescue the defects in archaellation in a Mc. maripaludis ΔearA strain (Ding et al., 2017). As shown in Figure 6B, examination of the fla promoter regions in selected archaellated Methanococcales containing an earA homolog revealed BRE sequences identical, or very similar, to that in Mc. maripaludis, i.e., with a string of seven adenines as in Methanococcus voltae, Methanococcus vannielii, and Methanothermococcus thermolithotrophicus or seven adenines in a stretch of eight nucleotides in the BRE of the fla promoter of Mcc. jannaschii. It would appear that in all these cases, the fla promoter requires the presence of EarA proteins to overcome weak BRE sequences, presumably to aid in the recruitment of TFB, as found for transcriptional activators TFB-RF1 and the ara-box binding factor.

The appearance of the ΔearA-sp mutant was surprising to us. The isolation of the original ΔearA mutant arose after it was discovered that, after repeated transfers in the lab, mutants carrying deletions of various fla or agl genes required for assembly of archaella stopped transcription of the fla operon and the fla operon reporter protein FlaB2 could not be detected in western blots (Vandyke et al., 2009; Ding et al., 2016b). It was determined that in at least some of these mutants, the reason for the cessation of fla operon transcription was a reading-frame shift mutation in earA. We reasoned that in these strains that carried mutations in fla or agl genes necessary for archaella assembly, it was an advantage to no longer synthesize several proteins, some of which, like archaellins, were required in large amounts when they could not be assembled in archaella. This led to a selective advantage for those cells that had stopped transcription of the fla operon, as in the earA mutants. Thus, it is not obvious to us why a mutation in the ΔearA that would restore transcription of the fla operon would arise and outgrow the original ΔearA mutant. The answer may lie in the presence of an additional mutation(s) in the ΔearA-sp strain that could be revealed by comparison of the complete sequence of the ΔearA and ΔearA-sp strains.

In this study, a spontaneous mutant with restored FlaB2 expression was isolated from a ΔearA mutant, indicating that in the spontaneous mutant the need for the transcriptional activator EarA for the transcription of the fla promoter was bypassed. Analysis of the DNA sequence in the fla promoter region from the spontaneous mutant revealed a three adenine deletion in the BRE region in the fla promoter. Sequence alignment showed that the mutated BRE in the fla promoter shares high similarity with BREs from strong promoters in methanogens, indicating that with this mutated BRE, the transcription initiation of the fla promoter could be conducted with components of the basal pre-initiation complex. We believe this is the first report of spontaneous mutation in the promoter region that overcomes the need for a transcriptional activator and it emphasizes the key role played by the BRE in promoter strength in Archaea.

Conceived and designed experiments: YD, AB, CK, and KJ. Performed the experiments: YD, AB, CK, and KJ. Analyzed the data: YD, AB, CK, and KJ. Wrote the paper: YD, AB, CK, and KJ.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.