**4. Discussion**

#### *4.1. Comparisons of BenM and CatM Provide a Framework to Engineer E*ff*ector-Binding Changes in CatM*

Residues R160 and Y293 in BenM-EBD, which are critical for benzoate-induced transcriptional activation of PbenA, were introduced into CatM. These replacements were designed to create a hydrophobic binding pocket resembling one in BenM that binds benzoate, termed the secondary effector-binding site (Figure S2) [5,6]. Since benzoate decreased the muconate-activated gene expression for these CatM variants (Figure 3), it is likely that benzoate competes with muconate for binding in the primary effector-binding site of this LTTR rather than binding to a secondary site. In CatM(H160R,F293Y) and CatM(H160R), R160 increased the muconate-activated transcription of PbenA. In BenM, responses to effectors in the primary and secondary binding sites appear to connect through charge-based interactions among three residues that separate muconate and benzoate in the protein structure (R146, E162 and R160) [1].Without benzoate in the secondary binding site of CatM variants, it is unclear how the replacement of H160 with R160 increases the response to muconate. Changes in the local environment may mimic what occurs in BenM upon binding benzoate. Regardless of the mechanism, the increased muconate-responsive transcription due to R160 is consistent with our previous conclusion that this residue in BenM can impact the nearby primary binding site.

The importance of the DBD in PbenA regulation was revealed by a spontaneous Ben<sup>+</sup> mutant. In this mutant, one amino acid change in the HTH region altered transcriptional activation in response to muconate. CatM(I18F), in strain ACN1111, increased muconate-dependent PbenA expression more than five-fold relative to CatM, in ACN1307 (Figure 4). A more BenM-like DBD may improve promoter recognition at PbenA and thus enable the CatM-EBD to display its typical e ffector specificity. Some data support the interpretation that DBDs and EBDs follow expected patterns, whereby promoter recognition is governed by the DBD and e ffector specificity, is controlled by the EBD. For example, replacing the entire DBD of CatM with that of BenM significantly improves regulation of PbenA (Figures 3 and 4, ACN1239 compared to ACN1307).

Most results sugges<sup>t</sup> a more complicated relationship among the domains. For example, if the low level of CatM-mediated transcription from PbenA were due solely to poor interaction with CatM-DBD, then CatM-DBDBenM would be expected to display a typical BenM pattern of e ffector response but yield low levels of transcription. Instead, the transcriptional levels mediated by this chimeric protein at PbenA were higher than for BenM (ACN1303 compared to ACN1232, Figure 3). At PcatB this hybrid variant (CatM-DBDBenM) might be expected to increase transcriptional activation compared to BenM because CatM is the cognate regulator of *catB*. Instead, the overall expression levels were lowered, and the BenM-mediated pattern of response was altered (ACN1367 compared to ACN1375, Figure 5). Another example that is counter to simple predictions resulted from replacing CatM-DBD with BenM-DBD. This alteration to CatM had little e ffect at PcatB (ACN1366 versus ACN1370, Figure 5). Experimental data are lacking to show if the variant proteins are produced at comparable levels in vivo or whether these proteins are equally stable. Nevertheless, the regulatory patterns, in most cases, are complex. An additional layer of variability results from di fferences in the flexible helix (the linker helix) that connects the DBD and EBD. In the 30 amino acid residues corresponding to this helix, BenM and CatM are 50% identical and 83% similar in sequence.

#### *4.2. Complex Patterns of Regulation*

Even without e ffectors, DBD alterations impacted transcription. Since BenM and CatM repress transcription in the absence of e ffectors, a several-fold increase in PbenA basal expression, which results from the loss of BenM, is interpreted as de-repression [2,3]. Consistent with this model, basal expression in most strains encoding CatM or a CatM variant was approximately 6-fold higher than for the strain encoding BenM (e.g., ACN1307, ACN717, ACN673, or ACN694 compared to ACN1232, Figure 3A). In one notable exception, two replacements in the CatM-DBD, I18F and K38N, caused a 21-fold expression increase in basal expression relative to BenM. This high increase suggests that this variant activates transcription without e ffectors (ACN1251 compared to ACN1232, Figure 3A). This variant may cause a change in protein structure that relieves binding to Site 3 and helps recruit RNAP. When these two DBD changes were accompanied by changes in the EBD or the DBD, high-level basal expression was no longer observed (ACN1347 and ACN1239, respectively, Figure 3A). Yet when all these EBD and DBD changes were combined in a single variant, BenM-DBDCatM(H160R,F293Y), transcriptional activation in the absence of e ffectors resulted again (ACN1302, Figure 3A). These transcriptional patterns did not correlate meaningfully with changes in the a ffinity of these proteins for the operator–promoter DNA (Table 2). Better interpretation awaits additional crystal structures and experiments with RNAP. Until then, the crystal structures of BenM-DBD with the Site 1 region of PbenA or PcatB can provide a framework for considering LTTR-DNA interactions.

#### *4.3. Interactions Between DBDs and Operator–Promoter Regions of PbenA and PbenA5146*

The I18F replacement in CatM-DBD alters transcription su fficiently to enable Ben<sup>+</sup> growth. Crystal structures of BenM-DBD-PbenA-Site 1 and a similar structure, CbnR-DBD-PcbnA-Site indicate that this residue is in helix α2, where it is involved in indirect readout [23,31]. Indirect readout refers to the e ffects of local nucleotides that cause sequence-dependent deformations of the phosphate backbone to control binding. In BenM, F18 is grouped with residues whose main chain amide N

atoms form hydrogen bonds and van der Waals contacts with DNA. The equivalent residue in CbnR, M18, is involved in sugar–phosphate recognition and contributes to DNA-binding strength rather than promoter specificity [31]. CatM binding to PbenA is weaker than for BenM (Table 2), which may be due in part from I18 distorting the packing of the helix against the DNA. While binding affinity was not tested for CatM(I18F), the CatM(I18F,K38N) was evaluated (Table 2). Consistent with our interpretation, this variant had an affinity for PbenA that was intermediate between that of CatM and BenM. The entire BenM-DBD not only further increased the affinity of CatM variants for PbenA, but this affinity actually surpassed that of the native BenM (Table 2). These results indicate that binding affinity is not determined solely by the DBD and also sugges<sup>t</sup> there are important variations in the oligomeric conformation of these regulators.

Residue 38 in BenM and CatM is in the recognition helix (α3) of the HTH motif. N38 in BenM can form a hydrogen bond with the DNA phosphate backbone, although this type of non-specific interaction is unlikely to confer specificity for PbenA. Instead, the importance of this residue may derive from a dynamic interaction with R34, which interacts directly with DNA and provides specificity in DNA recognition. The surrounding residues orient the side chain of R34. An interaction network between these residues and others, including E41 and Q37, might contribute to a conformational switch controlling promoter recognition or RNAP activation at PbenA [23]. K38 in CatM would form an electrostatic interaction with the phosphate backbone of PcatB and would project deeper into the major grove if a purine were positioned at the interaction point, e.g., nucleotide A36 in the BenM-DBD-*catB* DNA structure. However, the methyl-group of thymine, as found in this nucleotide position of PbenA, would impede this direct interaction. Consistent with this possibility, CatM(K38N) increased muconate-activated transcription of PbenA better than the wild-type CatM (Figure 4). Similar interactions are observed in other LTTRs, such as CbnR (A38, R34, and D42) [31], MetR (S38, S34, H35, and Q42) [32], DntR (R43, N39, T46 and A47) [33] and Tsar (Q38, D42, and S34) [34].

DNA interactions with residue 38 in the DBD may influence effects at PbenA5146. Binding at Site 2 of this promoter (which has a T-to-A transversion) by CatM will be enhanced compared to PbenA due to K38 interactions with the partial negative charge on the N7 atom of the adenine base. Interactions of BenM with PbenMA5146 may not be negatively impacted because a water molecule can bridge to the N7. Thus, the interaction of CatM with PbenMA5146 may be stabilized at Site 2 but not Site 3, thereby disrupting the equilibrium between the protein binding to Site 2 (stabilized by effectors) and Site 3 (stabilized by lack of effectors). This alteration could result in the de-repression of basal activity observed at this promoter for CatM and help explain transcriptional activation in the absence of effectors for the CatM variants with F293Y and/or H160R replacements in the CatM-EBD (Figure 3A).

#### *4.4. Interactions Between DBDs and Operator–Promoter Regions of PcatB*

Factors affecting K38 interactions with PbenA5146 may also affect recognition at PcatB. Between the two half sites of dyad symmetry in PcatB Site 2, a G corresponds to the position of the T-to-A mutation of PbenMA5146 (Figure 3C). In general, the sequence ATAC-pyrimidine-N5-purine-GTAT will favor binding at PcatB by CatM. This pattern is observed for Sites 1 and 2. At PbenA, Site 2 swaps the arrangemen<sup>t</sup> of the purine/pyrimidine pair, a situation that may partially explain why wild-type CatM shows significantly reduced responses at PbenA.

BenM activated transcription relatively strongly at PcatB despite a different pattern of response to benzoate compared to its activity at PbenA (Figure 5). As measured by EMSAs, the affinity of BenM for PbenA and PcatB is comparable. While the native CatM has a higher affinity for its cognate PcatB promoter than for PbenA, all proteins with a native BenM-DBD have a comparable affinity for both promoters (Table 2). It is not ye<sup>t</sup> possible to discern the features that control such variation in binding and transcriptional control, and a better understanding of LTTR interactions with RNAP is needed.

#### *4.5. A Model of PbenA and PcatB Promoters with RNAP and the DBDs of BenM and CatM*

Aspects of the model described above for protein interactions with PbenA also apply to PcatB. However, some important di fferences stem from DNA sequence variation. PcatB di ffers in the region of the 3' half-site of symmetry in Site 2 from that of PbenA. The sequence of PcatB in this region (TCTTTT, Figure 3) does not match the −35 consensus sequence of σ70 promoters. Further, the 3' half-site of Site 2 (TTTA) lacks similarity to the canonical binding 3' half-site shared by BenM and CatM, GTAT (PbenA, GTGT). This sequence divergence suggests that the regulators at PcatB may play a more direct role in transcription initiation by replacing the -35 region DNA interactions with σ70 with compensating LTTR-RNAP contact. Analysis of the electrostatic surfaces of σ70 and BenM-DBD at the *benA* promoter show a strongly negative surface created by D23 in BenM (E23 in CatM) and a strongly positive surface defined by residues K593, R596, and K597 of σ70 which are part of the conserved regulatory region 4.2 of the sigma factor. Clearly, rotation of the RNAP holoenzyme toward BenM by moving forward a few base pairs along PcatB will position the RNAP against BenM or CatM in a complementary fashion. The amino acids in this region 4 of sigma have been mapped as binding residues in other sigma regulatory molecules [35]. Between the half-sites of symmetry of Site 2 in PcatB, there is a mix of A and T nucleotides that could act as an UP-element to stabilize αCTD interactions, as observed in PbenA. This combination of σ70 and αCTD interactions would make the PcatB a mixed class I–class II promoter. However, the UP-element sequence di ffers from that of PbenA because of the flanking C/G nucleotide pair at the 5' half-site of Site 2. Alternatively, if CatM and BenM bind only to the 5' half-site sequence, the 3' area of Site 2 might be occupied by σ70 (perhaps at the intervening TTGTT) and allow some conformational flexibility for the previously mentioned electrostatic interactions to be favored.

#### *4.6. Broader Implications and Conclusions*

Although LTTRs have been studied for more than 30 years, many aspects of their structure and function remain unclear [8]. Because of their similarities and the overlap in their control of a complex regulon, BenM and CatM provide unique opportunities for comparative investigation [1]. Furthermore, their role in aromatic compound catabolism holds promise for biotechnology applications, including lignin valorization [18]. Because of such applications, these and related regulators are receiving renewed attention. For example, transcriptional regulators that respond to aromatic compounds are useful as biosensors [36]. In a recent study, high throughput methods were developed to combine protein domains for DNA binding and those for e ffector responses with the aim of creating benzoate-responsive biosensors [37]. This approach builds on the modular nature of bacterial transcriptional regulators.

Our studies demonstrate that regulatory alterations can be engineered to integrate multiple signals by enabling the simultaneous response to more than one e ffector molecule. Although it took repeated e fforts, we obtained CatM variants that recognize and respond to benzoate. In some variants, benzoate increased the transcriptional response to muconate, thereby approximating the synergistic transcriptional regulation mediated by BenM. This type of rapid signal amplification has potential uses in metabolic engineering and synthetic biology. While it is not ye<sup>t</sup> possible to interpret some characteristics of the variant regulators that were studied, our results lay the foundation for continued investigation. LTTRs are more than the sum of their parts; both the DBDs and the EBDs a ffected promoter specificity as well as transcriptional responses to e ffectors. The abundance of this family of regulators and the importance of LTTR-regulated processes underscore the value of continued research.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2073-4425/10/6/421/s1, Figure S1. Regulatory model for PbenA. Figure S2. Structural comparisons of BenM-EBD and CatM-EBD. Figure S3. Changes in BenM-DBD a ffect relative *benA* expression. Table S1. *Acinetobacter baylyi* strains. Table S2. Plasmids. Table S3. Primers.

**Author Contributions:** Conception and design: M.P.T.-V., C.M., and E.L.N.; Data collection and analysis: M.P.T.-V., C.M., E.L.N; Plasmid and strain construction: M.P.T.-V. and N.S.L.; Lab experiments: M.P.T.-V.; Manuscript writing: M.P.T.-V., E.L.N and C.M.

**Funding:** This research was supported by NSF grants MCB1024108 (to C.M. and E.L.N) and MCB1615365 (to E.L.N.).

**Acknowledgments:** We thank Ajchareeya Ruangprasert and Melesse Nune for assistance in protein purification and electromobility shift assays. We also thank Curtis Bacon for help with plasmid and strain construction. Jenifer Morgan, Chelsea Kline, James Valle and Walker Whitley assisted with the selection of *A. baylyi* strains.

**Conflicts of Interest:** The authors declare no conflict of interest.
