In the present study the PDNT promoter was used in the comparison of the induction profiles of DntR, NtdR and NagR. This promoter was found in the two DntR-expressing strains Burkholderia sp. DNT and Burkholderia cepacia R34 [29, 30] and it is nearly identical to the PNTD promoter found in the NtdR-expressing strains Comamonas sp. strain JS765 and Acidovorax sp. strain JS42 , as well as the PNAG promoter found in Ralstonia sp. strain U2 (for alignment of the promoter regions, see Lessner et al ). From the consensus NahR binding region  to the -10 region (counted from the dntA transcription start) the promoters are identical. Therefore, the binding of DntR, NtdR and NagR to PDNT, and to PNTD or PNAG, is expected to be very similar. Although the four nucleotide changes observed outside the core promoter region might affect the binding affinity, it is assumed not to influence the inducer specificity.
E. coli as a model system for monitoring the inducer responses
In the study by Lessner et al.  it was reported that the ability of NtdR to respond to nitroaromatic compounds seen in Comamonas and Acidovorax could not be restored in E. coli. In the present study, very little response was seen for some nitroaromatic compounds in E. coli DH5α when using similar conditions as in the above-mentioned study (NtdR was expressed in trans, growth in LB medium). However, by optimizing a number of experimental conditions (LTTR expressed from its own promoter, choice of growth medium) the responses to 2,4-DNT, 2-nitrobenzoate, 4-nitrobenzoate, and benzoate were significantly increased for NtdR (see Figure 4). There was no significant response to 2-nitrotoluene or 4-nitrotoluene, as reported for NtdR in the original 2-nitrotoluene-degrading Acidovorax strain. This effect could be due to differences in the metabolism or uptake/transport of these compounds between the different bacterial species. The 2-nitrotoluene-degrading Acidovorax strain might be capable of accumulating the mono-nitrotoluenes intracellularly, since 2-nitrotoluene is a growth substrate in this strain.
When the fluorescence response was followed over time for the LTTRs expressed in the one-plasmid system in LB, salicylate gave a steady increase over time, while the small response to 2,4-DNT displayed a maximum after a few hours followed by a decrease. In contrast, the response to 4-nitrobenzoate only occurred several hours after addition (Additional file 1). The transient response to 2,4-DNT may be due to reduction of the nitro-groups of 2,4-DNT to the corresponding amine or hydroxylamine, as reported previously for TNT in E. coli AB1157 . This is also supported by our observation of an orange colour of the growth media formed in both LB and minimal media a few hours after addition of 2,4-DNT to the bacterial cultures, suggesting the formation of Meisenheimer complexes also observed in the AB1157 strain. The delayed response to 4-nitrobenzoate, on the other hand, may be explained in terms of response to a degradation product of this compound or the presence of an additional binding site for 4-NB that alters the equilibrium of the relative concentrations of the different states of DntR, thereby changing the kinetics of transcriptional repression/activation.
The additive effect upon addition of both salicylate and 4-nitrobenzoaate
As seen in Figure 6 the response upon simultaneous addition of both salicylate and 4-nitrobenzoate was larger than the sum of responses to each of these compounds. The existence of a second, low affinity site might provide an explanation to this observation. Also, the higher background fluorescence observed in LB compared to the minimal media may be due to an inducing compound(s) that is present exclusively in the LB medium and gives an additive effect together with 4-nitrobenzoate. The additive effect with 4-nitrobenzoate could also be due to a metabolic factor. However, two binding sites have previously been found for another LTTR, BenM, where a synergistic effect was observed upon addition of cis, cis-muconate and benzoate . In the case of BenM, the distinct binding sites have been identified in a truncated form of BenM , where the binding site for cis, cis-muconate is situated between the two subdomains in the IBD, analogous to the salicylate-binding site of DntR. The benzoate moiety found in crystals with BenM and cis, cis-muconate was buried in a hydrophobic region of subdomain 2, moving the two subdomains more closely around cis, cis-muconate. Thus, binding of benzoate could provide stabilization of the closed state (less stable with only cis, cis-muconate bound), and thereby increase transcriptional activation. A similar mechanism could be responsible for the effect observed for salicylate and 4-nitrobenzoate with DntR. Recently, new crystal structures of the truncated forms of DntR, co-crystallized with salicylate, were compared with the apo-structure of DntR . Two salicylate moieties bound/IBD monomer were found in a conformation that is suggested to be responsible for full transcriptional activation. In other words, there are two binding sites for aromatic inducers, but it remains to investigate whether or not also 4-nitrobenzoate binds to the other site.
Effect of mutations that differ between the LTTRs
No significant differences were seen between the inducer-specificity profiles for DntR and NagR, which is not unexpected given that the two residues that differ between these two transcription factors are found in the DBD. In the case of NagR and NtdR, in which the DBDs are identical, while five residues differ in the IBD/linker region, NtdR displayed a significant increase in response to 2-nitrobenzoate, 4-nitrobenzoate and benzoate also in the less sensitive two-plasmid system. Thus, even if the two-plasmid system is less sensitive, the broadening of specificity can be observed when following the response to these compounds where a clear response is seen for NtdR, even though no conclusions can be drawn regarding the changes in response to individual compounds.
When considering mutations that are responsible for the inducer-specificity broadening, our data support the previously obtained results with NtdR mutants, suggesting the key importance of residues 169 and 227  (although different combinations of mutations were constructed in our study compared to that by Ju et al.). We choose to focus on combinations of mutations with the H169L mutation, because this residue lines the inducer-binding pocket previously identified in DntR , where a F111L/H169V double mutant was shown to respond better to 2,4-DNT than the wild-type DntR. In the study of Ju et al. , no NagR mutants with H169L in combination with other mutations were analyzed; instead the study was focused on combinations with residues 227 and 232. Thus, our study provides complementary information of the combined effect of H169L with the other mutations.
In the study of Ju et al. , the single mutations H169L, P227S and I232V in NagR were shown to give the greatest change in inducer specificity, although the H169L mutant had no activity in the assay used. The "opposite" L169H mutation in NtdR, however, resulted in an improved response to nitro-aromatic compounds and an improved repression compared to wild-type NtdR. The H169L mutant of DntR showed very little response to the potential inducers. However, the double mutants H169L/P227S and H169L/K189R displayed a clear broadening of the inducer response that is not seen for the single mutants P227S or K189R alone (Additional file 1). These double mutants were not studied for NagR, instead the double mutant P227S/I232V was shown to recognize a large number of nitroaromatics to which the wild-type NagR was insensitive . The triple mutant I232V/K189R/P227S responded to some additional nitroaromatic compounds compared to the double mutant, indicating that all these residues are involved in modulating the response . Our results with the H169L/P227S and H169L/K189R double mutants suggest that several combinations of the mutations occurring in NtdR (compared to NagR/DntR) can result in a broadening of the inducer specificity. This gives additional possibilities for mutational trajectories towards a response to nitro-aromatic compounds in addition to those suggested by Ju et al. . To perform an accurate reconstruction of the evolutionary history of nitroaromatic detection, a complete analysis of all 32 mutants between NagR and NtdR would be necessary, to exclude as many of the 120 possible trajectories as possible. Such a study has been made for β-lactamase, where 102 out of 120 theoretically possible trajectories were found to be inaccessible to Darwinian selection, giving 18 possible paths of protein evolution 
Experimental factors that influence the LTTR-mediated transcriptional regulation of the reporter gene gfp
In the present study, we observed that by varying the experimental conditions, such as the composition of the growth media, the background expression levels of the gfp reporter changed (and thereby the sensitivity changed). However, the factor that influenced the sensitivity the most was whether the LTTR was expressed from its own promoter in the one-plasmid system or from the T5/lac promoter in the two-plasmid system. The introduction of a His6-tag in the two-plasmid system could partly explain the lower response, but could not explain the total reduction of sensitivity.
Previous studies of LTTRs have mainly focused on the regulation of the genes downstream of the promoter, and in many reporter systems, the original gene arrangement has been broken and expressed from a separate plasmid [18, 25, 28]. Also, the composition of the growth media varies in different studies, where rich growth media have been used in some studies of inducer responses for LTTRs [25, 28]. Until now, there have been no comparative studies of how these experimental factors influence the LTTR-mediated response to various inducers.