Bioinformatic evidence for a widely distributed, ribosomally produced electron carrier precursor, its maturation proteins, and its nicotinoprotein redox partners

Abstract

Background: Enzymes in the radical SAM (rSAM) domain family serve in a wide variety of biological processes, including RNA modification, enzyme activation, bacteriocin core peptide maturation, and cofactor biosynthesis. Evolutionary pressures and relationships to other cellular constituents impose recognizable grammars on each class of rSAM-containing system, shaping patterns in results obtained through various comparative genomics analyses.

Results: An uncharacterized gene cluster found in many Actinobacteria and sporadically in Firmicutes, Chloroflexi, Deltaproteobacteria, and one Archaeal plasmid contains a PqqE-like rSAM protein family that includes Rv0693 from Mycobacterium tuberculosis. Members occur clustered with a strikingly well-conserved small polypeptide we designate "mycofactocin," similar in size to bacteriocins and PqqA, precursor of pyrroloquinoline quinone (PQQ). Partial Phylogenetic Profiling (PPP) based on the distribution of these markers identifies the mycofactocin cluster, but also a second tier of high-scoring proteins. This tier, strikingly, is filled with up to thirty-one members per genome from three variant subfamilies that occur, one each, in three unrelated classes of nicotinoproteins. The pattern suggests these variant enzymes require not only NAD(P), but also the novel gene cluster. Further study was conducted using SIMBAL, a PPP-like tool, to search these nicotinoproteins for subsequences best correlated across multiple genomes to the presence of mycofactocin. For both the short chain dehydrogenase/reductase (SDR) and iron-containing dehydrogenase families, aligning SIMBAL's top-scoring sequences to homologous solved crystal structures shows signals centered over NAD(P)-binding sites rather than over substrate-binding or active site residues. Previous studies on some of these proteins have revealed a non-exchangeable NAD cofactor, such that enzymatic activity in vitro requires an artificial electron acceptor such as N,N-dimethyl-4-nitrosoaniline (NDMA) for the enzyme to cycle.

Conclusions: Taken together, these findings suggest that the mycofactocin precursor is modified by the Rv0693 family rSAM protein and other enzymes in its cluster. It becomes an electron carrier molecule that serves in vivo as NDMA and other artificial electron acceptors do in vitro. Subclasses from three different nicotinoprotein families show "only-if" relationships to mycofactocin because they require its presence. This framework suggests a segregated redox pool in which mycofactocin mediates communication among enzymes with non-exchangeable cofactors.

Figure 1

Multiple sequence alignment of gene predictions for mycofactocin precursors. All detectable members of the family defined by TIGR03969 were collected, sorted by length, and aligned by MUSCLE [33], and then made non-redundant to 80% sequence identity, preferentially keeping sequences previously treated as genes and available through NCBI. Significant sequence similarity is restricted to the last 23 amino acids; the seven invariant residues occur among the last eight positions.

Figure 2

Mycofactocin gene cluster regions. Examples of the mycofactocin cluster are shown from Geobacter uraniireducens (Deltaproteobacteria), Pelotomaculum thermopropionicum (Firmicutes), Mycobacterium avium (Actinobacteria), Thermomicrobium roseum (Chlorobi), and Haloterrigena turkmenica (Archaea). Additional SDR family oxidoreductases for these species, beyond those shown in the mycofactocin cluster, include four Fe-dependent from G. uraniireducens and nine from P. thermopropionicum, twenty-five SDR and five Zn-dependent from M. avium, and four Zn-dependent from T. roseum.

Figure 3

SDR family dehydrogenase SIMBAL heat map and related structure. Panel A: SIMBAL map for 267-residue protein PTH_0592 from Pelotomaculum thermopropionicum SI. Data sets numbered 974 sequences in the TRUE partition and 6998 in the FALSE partition after being made non-redundant to less than 80% sequence identity. The X coordinate represents the location of the center of each subsequence along the full-length protein sequence, the Y coordinate represents the size of subsequence tested. The plot is triangular, narrowing as it rises, because longer subsequences are more constrained and have less room to slide from N-terminus to C-terminus. Panel B: Structure 1NFQ of Rv2002, an NADH-dependent 3alpha, 20beta-hydroxysteroid dehydrogenase in the SDR family. Bound NAD+ is shown in yellow and androsterone in blue. Active site resides Ser-140, Tyr-153, and Lys-157 are shown in cyan with their side chains as space-filling spheres. Highlighted in brown and red are regions from Rv2002 that map by pairwise alignment to the locally top SIMBAL hit sequences from PTH_0592 at subsequence sizes of 8 (brown) and size 12 (red), both taken from within the absolute highest scoring subsequence.

Figure 4

The SIMBAL heat map for gi|284167095 from Haloterrigena turkmenica, a Zn-dependent protein from family TIGR03989. SIMBAL training sets contained 590 sequences in the TRUE partition and 3620 in the FALSE partition after making the sets non-redundant to no more than 80% sequence identity. The apex score of 30.1 nears that of highest local score, consistent with the ability of TIGR03989 to identify large numbers of members found exclusively in mycofactocin-producing species. Interesting features include extended cool (blue and green) regions such as the N-terminal region of about 80 residues, which contains regions well conserved among Zn-dependent alcohol dehydrogenases yet apparently poorly predictive for whether or not a matching protein's genome of origin contains the mycofactocin cluster.

Figure 5

Iron-dependent dehydrogenase SIMBAL heat map and related structure. Panel A shows the SIMBAL heat map for Dtox_4270, a 419-residue group III iron-dependent dehydrogenase from Desulfotomaculum acetoxidans DSM 771. The training set contains 62 sequences in the TRUE partition and 1555 in the FALSE partition. This heat map shows a plume rising from a short subsequence TSNPKDYE, homologous to the sequence VPNPTITV in the sequence of lactaldehyde:1,2-propanediol oxidoreductase Of Escherichia Coli, which has a solved crystal structure 2BL4. Panel B shows the secondary structure cartoon of 2BL4 in green. The NAD cofactor is shown as space-filling spheres in yellow, and the iron atom as a blue sphere. The sequence identified by homology to the SIMBAL hot spot from panel A is shown in red, making close contact with the NAD but not with the iron atom. The location of the hot spot is consistent with a role in controlling the ability of NAD to exchange electrons to other redox carriers.