**1. Introduction**

Two LysR-type transcriptional regulators (LTTRs) control benzoate degradation by a soil bacterium, *Acinetobacter baylyi* ADP1 [1]. These paralogs, BenM and CatM, have overlapping but distinct functions (Figure 1). BenM controls the initial steps in benzoate consumption by activating transcription of the *benABCDE* operon in response to benzoate and one of its metabolites, *cis,cis*-muconate, hereafter designated muconate [2,3]. CatM, which responds solely to muconate, activates low-level expression from this promoter, PbenA [4]. We sought to create a benzoate-responsive CatM that mimics an unusual characteristic of BenM, namely the ability to activate transcription synergistically in response to two effectors. These studies should improve understanding of the molecular basis of this type of transcriptional activation and, in general, facilitate the engineering of LTTRs to respond to novel effectors when designed for varied biotechnology applications.

BenM and CatM have similar N-terminal DNA-binding domains (DBDs, Figure 2), and they bind to the same regions of PbenA [2,4]. Whereas LTTR-DNA interactions repress basal transcription, conformational changes occur in response to effectors to activate transcription (depicted in Figure S1). Although CatM alone does not activate sufficient transcription for growth on benzoate as the carbon source, mutations can increase CatM-regulated PbenA transcription by augmenting its response to muconate or by enabling transcription without an effector [3,4]. Two amino acid replacements in

the effector-binding domain (EBD) of CatM each enable BenM-independent growth on benzoate, ye<sup>t</sup> neither of these, nor any other known CatM variant, responds to benzoate.

**Figure 1.** BenM and CatM regulate the *ben* and *cat* genes from four promoter regions (diamonds). The *benABCDE* operon is primarily regulated by BenM from PbenA, and the *catBCIJFD* operon is primarily regulated by CatM from PcatB. The encoded enzymes are used for the degradation of benzoate.

**Figure 2.** (**A**) Ribbon representation of BenM-DBD and the adjacent "linker" helix bound to DNA (PDB ID 4IHS). Green side chains show differences between BenM and CatM. Labeled residues are discussed in the text. (**B**) Nine amino acids differences in the DBDs (bold), including those at positions 18 and 38 (highlighted). The Helix (α2)-Turn-Helix (α3) (HTH) motif is involved in DNA recognition.

The structures of BenM-EBD and CatM-EBD are highly conserved (Figure S2) [5,6]. Although benzoate binds in a hydrophobic pocket of BenM, this compound has not been detected in structures of the corresponding region of CatM [5,6]. This pocket is distinct from an inter-domain cleft in BenM and CatM that binds muconate and serves as the typical effector-binding site in LTTRs [6]. The EBDs of LTTRs usually assume the conformation of a periplasmic-binding protein [7–9]. However, BenM is the only LTTR known to have a secondary effector-binding site that enables synergistic activation of transcription with different metabolites [2]. In an effort to understand this synergism, we engineered amino acid replacements to make CatM more similar to BenM.

Two residues in the hydrophobic binding pocket of BenM are critical for benzoate-activated transcription, R160 and Y293 [5,6]. When these amino acids are replaced with those at the comparable positions of CatM, H160 and F293, BenM fails to activate transcription in response to benzoate as a sole effector or in combination with muconate [5]. Moreover, benzoate inhibits muconate-activated gene expression in these BenM variants [3,5]. As described here, the converse changes in CatM (H160R and F293Y) did not initially generate a benzoate-responsive CatM. However, efforts to alter CatM were expanded to increase understanding of these representative members of the LTTR family, the largest group of homologous transcriptional regulators in bacteria [10]. After multiple attempts, CatM variants were isolated that respond to benzoate and that activate increased transcription in response to both effectors. As discussed below, changes were required in both the DBD and EBD regions of CatM for this new functionality.

#### **2. Materials and Methods**

#### *2.1. Bacterial Strains and Growth Conditions*

*A. baylyi* strains, derived from the wild-type ADP1 [11,12] are listed in Table S1. *Escherichia coli* DH5α (Thermo Fisher Scientific, Waltham, MA, USA) and XL-1 blue (Agilent Technologies, Santa Clara, CA, USA) were used as plasmid hosts. Bacteria were grown on Luria-Bertani (LB) medium at 37 ◦C [13]. In some cases, *A. baylyi* strains were grown in minimal medium [14,15] with succinate (10 mM), pyruvate (20 mM), benzoate (2 mM), muconate, (2.5 mM), or anthranilate (1.5 mM) as the carbon source. Antibiotics were added when needed at the following final concentrations: ampicillin, 150 μg/mL, kanamycin, 25 μg/mL, spectinomycin, 13 μg/mL, and streptomycin, 13 μg/mL. Growth was monitored by turbidity (OD600).

#### *2.2. Site-Directed Mutagenesis and Strain Construction by Allelic Replacement*

Site-directed mutagenesis of plasmid DNA was conducted with mutagenic primers and methods based on the QuickChange II protocol (Agilent Technologies, Santa Clara, CA, USA [5]). The primers, and the mutations they introduce, are listed in Tables S2 and S3. In some cases, plasmids were constructed using splicing by overlap extension PCR (SOEing) [16]. Linearized plasmid-borne alleles were used to replace chromosomal genes in *A. baylyi* recipient strains by homologous recombination [17,18]. Transformants were identified by phenotypic changes in antibiotic resistance, carbon source utilization or loss of the *sacB* marker (in the presence of 10% sucrose and no NaCl in the medium) [5,18]. The genotypes of mutant strains were confirmed by PCR analysis and DNA sequencing (Genewiz laboratories, South Plainfield, NJ, USA) of the chromosomal regions where changes were introduced.

#### *2.3. Selection for BenM-Independent Growth on Benzoate*

Strains that form colonies on plates with benzoate as the sole carbon source were defined as Ben<sup>+</sup>. Spontaneous Ben<sup>+</sup> mutants arising from strains lacking BenM were isolated as described [3,4]. Chromosomal *catM* DNA was recovered from Ben<sup>+</sup> strains using the gap-repair method [18,19]. Briefly, cells were grown on benzoate medium to mid-log phase, mixed with linearized pBAC184 (Table S2) and plated on LB medium. Transformants with circularized plasmids, resulting from homologous recombination, were selected in medium with ampicillin. Drug-resistant cells were pooled, and plasmid DNA was extracted and used to transform *E. coli*. Recovered *A. baylyi* DNA was tested for the ability to confer a Ben<sup>+</sup> phenotype to recipient strains (without BenM) by allelic replacement. Mutations were identified by DNA sequencing (Genewiz laboratories, South Plainfield, NJ, USA).

#### *2.4.* β*-Galactosidase (LacZ) Assays*

Transcriptional *lacZ* fusions were constructed as described [3,4]. For cultures grown on LB, effectors were added at final concentrations of 500 μM of benzoate or muconate, or when added together, 250 μM of each. Some cultures were grown with pyruvate (20 mM) or muconate (3 mM) as the carbon source. Effectors added to pyruvate-grown cultures were added at the following concentrations: 65 μM benzoate or muconate, or 32.5 μM of each when added together. Growth was measured by optical density (OD600), and assays were done when cultures reached late-exponential phase as

described [4,5]. Directions from the FlourAce β-galactosidase reporter kit (BioRad, Hercules, CA, USA) were followed. The hydrolysis of the substrate, 4 methylumbelliferyl-galactopyranoside (MUG) to the product 4-methyllumbelliferone (4MU) was detected with a TD-360 miniflourometer (Turner Designs, San Jose, CA, USA). A standard curve was used to quantify 4MU.

#### *2.5. Purification of BenM and CatM and Variant Proteins*

Plasmids, pBAC433 and pBAC430, were used to express full-length regulators with C-terminal histidine tags, BenM-His and CatM-His, respectively [2]. Plasmids were made to encode variants, pBAC1027 (BenM-DBDCatM-His), pBAC1045 (CatM(I18F,K38N)-His), and pBAC1086 (Ben-DBDCatM(H160R,F293Y)-His). BenM-His was purified as described [20]. CatM-His and CatM variants were purified similarly but were eluted in a di fferent bu ffer (30 mM Tris, 500 mM NaCl, 30% glycerol (*v*/*v*), 500 mM imidazole, and 10 mM β-mercaptoethanol (pH 7.9)). Fractions with pure CatM were pooled and dialyzed against a bu ffer (20 mM Tris-HCl, 250 imidazole (pH 7.9), 500 NaCl, 10% (*v*/*v*) glycerol) to increase solubility, and then were concentrated to 2–10 mg·mL−1. Protein concentrations were determined by the method of Bradford with bovine serum albumin as the standard [21]. Proteins fractions were frozen with liquid nitrogen and stored at −70 ◦C until use.

#### *2.6. Electrophoretic Mobility Shift Assays (EMSAs)*

Operator-promoter DNA (PbenA and PcatB) was PCR amplified with 5'-6-carboxyfluorescein (6-FAM) labeled primers (Table S3). PCR products, approximately 150–250 bp, were extracted with a gel DNA recovery kit (Zymo Research, Irvine, CA, USA). For EMSAs, 1 nM DNA was incubated with di fferent concentrations of protein (0, 2.5 nM, 5 nM, 10 nM, 20 nM, 40 nM, 80 nM, 160 nM, 320 nM, 640 nM and 1.28 μM) for 1 h at 37 ◦C with or without muconate, benzoate or both. E ffectors in the reaction were present at a total concentration of 1.6 mM individually or at a concentration of 800 μM each when combined. DNA-protein samples were resolved in 6% polyacrylamide gels. Electrophoresis was performed in bu ffer (40 mM Tris (pH 7.6), 20 mM acetic acid, 1 mM EDTA) for 1 h at 185 volts at 4 ◦C. When indicated, e ffectors were added to this bu ffer at the same concentrations described above. Fluorescently labeled bands were detected using the Amersham Typhoon PhosphorImager system (GE Healthcare Bio-Sciences, Pittsburgh, PA, USA) at 526 nm using the short-pass emission filter. The bound DNA relative to the unbound DNA was quantified by Gel-Pro analyzer (Media Cybernetics, Rockville, MD, USA). Values were fitted into a saturation curve using the equation for one site binding with total accounting for ligand depletion to determine the equilibrium constant (Kd) (Prism 8 software, GraphPad, San Diego, CA, USA) [22].

#### *2.7. Modeling the Transcription Activation Complex at PbenA and PcatB*

An atomic model of PbenA was created by merging the structures of BenM-DBD bound to DNA [23] and a transcription initiation complex of *E. coli* RNA polymerase (RNAP) [24] using the X3DNA suite [25] with the DNA atoms as alignment guides. Since the Site 1 DNA of PcatB has the most sequence identity with *benA* Site 2, the structure of BenM-DBD complexed with PcatB Site 1 (PDB ID 4IHS, chains A, B, E, and F) was used to model atoms locally at *benA* Site 2. The structure of BenM-DBD complexed with PbenA Site 1 (PDB ID 4IHT, chains A, B, E, and F) was used to model *benA* Site 1. The model for RNAP (PDB ID 4YLN) was positioned assuming that the TTGAAC downstream from PbenA Site 2 corresponds to the *E. coli* σ70 binding site with the sequence TTGACA in the 4IHT complex. DNA residues in the three structures were changed and renumbered to match the PbenA sequence with the program *mutate\_bases*. A composite DNA backbone was generated by calculating the helical parameter sets (shear, stretch, stagger, buckle, prop-tw, opening, shift, slide, rise, tilt, roll, and twist) with program *find\_pair* from the three individual DNA double helical segments and then merging the parameters into one file. The two helical parameter sets at the transition regions were averaged. A DNA model was generated from the combined helical parameters using the program *rebuild*. The three atomic structures containing both the DNA and protein residues were aligned on the composite

DNA backbone using *align* command of PyMOL [26]. Similar structures of the PcatB and PbenA5146 transcriptional activation complexes were generated by mutating the *benA* DNA sequences in the composite structure. The EBD domains were not fitted to the model because the close proximity of the two binding sites is not consistent with any full-length structure of an LTTR at this time.
