Fani et al. (2005) studied the organization of the histidine synthesis genes across the Proteobacteria. They reported that the unified operon hisGDC(NB)HAF(IE) found in Escherichia coli, where hisNB and hisIE are gene fusions, is broken up in many Proteobacteria. They considered that the common ancestor of the Proteobacteria may have had this unified organization and it was then broken up in several lineages, or that the unified operon may have evolved by joining several pieces. They favored the joining model, with the unified operon forming in the ancestor of the ß/γ-Proteobacteria, because they thought it unlikely that the operon would independently break up multiple times. However, they restricted their analysis to the Proteobacteria and considered other genomes as being outside the scope of their study.

As can be seen with the MicrobesOnline comparative genome browser (e.g., http://www.microbesonline.org/hisoperon), bacteria across many phyla have a compact histidine operon, including high-GC Gram positives (e.g., Mycobacterium tuberculosis has hisDCBH-impA-hisFI), low-GC Gram positives (Bacillus subtilis has hisZGDBHAFI), Thermotoga maritima (hisGDCBHAFI), Euryarchaeota (Pyrococcus furiosus has hisZGDBHAF(IE)C), and Crenarchaeota (Sulfolobus solfataricus has hisCGABFDEHI). This broad phylogenetic distribution suggests that the unified histidine operon is ancient. Consistent with the hypothesis of a single origin, hisDBHAFI are in the same relative order in all of these groups except the Crenarchaeota. Also consistent with this hypothesis, some δ-Proteobacteria have a unified histidine operon (e.g., Geobacter metallireducens has hisGDCBHAFI). The δ-Proteobacteria branched early in the evolution of the Proteobacteria (probably after the ε‘s and definitely before the α‘s diverged from the ß/γ‘s) and were not included in Fani et al.’s study (the data may not have been available).

Could the ß/γ-Proteobacteria have formed the unified histidine operon and then transported it to other lineages? To test this hypothesis we built a concatenated protein tree from the eight genes hisGDCBHAFI. These eight genes have easily identifiable orthologues (from bidirectional best hits or single-copy COGs) in most genomes. Given 122 genomes that contain all 8 genes, we built 8 alignments with MUSCLE, concatenated them, removed positions with gaps or adjacent to gaps, and built a consensus neighbor-joining tree with PHYLIP NEIGHBOR and CONSENSE. We used 100 bootstraps and gamma-distributed rates in eight categories with a coefficient of variation of 1.14 (estimated using tree-puzzle).

As shown in Fig. 1, the unified histidine operon is present across the tree and not only in sequences affiliated with the ß/γ-Proteobacteria. The low-GC Gram positives, the high-GC Gram positives, and diverse thermophiles form well-supported clades containing the unified operon. It also appears that the ß-Proteobacteria have rather distinct sequences from most (but not all) of the γ-Proteobacteria, so that the major γ-Proteobacterial group on the right in Fig. 1 (which includes E. coli and other γ‘s with the unified operon) may have arisen by horizontal transfer. The paraphyly of Proteobacterial hisG, hisD, and hisC was previously noted by Bond and Francklyn (2000).

Figure 1
figure 1

A concatenated protein distance tree of hisGDCBHAFI across 122 prokaryotes. The number on each leaf (species) indicates how many of these eight genes are present in a single operon: 7 and 8 indicate unified operons and 1 indicates that the genes are completely scattered. Additional members of a species with the same operon organization are marked with a period. The heavy circles mark well-supported major clades present in more than 90 of 100 boostraps. The vast majority of the branches within each major clade have similarly high boostraps. None of the larger groupings of major clades have bootstraps over 80.

Full size image

Thus, the unified his operon is indeed ancient. Because horizontal transfer is common in bacteria, we argue that it is essential to examine as many genomes as possible when drawing conclusions about the origins of a character such as operon structure. Furthermore, operon structure evolves rapidly relative to the age of the Bacteria, or even the age of the Proteobacteria: about half of E. coli operons are broken apart in non-γ-Proteobacteria. Reconstructions of operon evolution from parsimony may be misleading when used on large timescales.