The first phase of our research was to determine which genes, if any, demonstrated a clear effect on the frequency of IS insertion into bgl upon deletion. Figures 2A,B show data gathered from several single deletion strains over a 10-day period. The full list of deletion strains tested is tabulated in Supplementary Table S1. Genes were selected for study based on the DNA-binding abilities of their gene products and were loosely separated into two groups: the global histone-like DNA-binding proteins and operon-specific DNA-binding proteins with known binding sites within the bgl control region. Colony PCR was performed on the mutant colonies of each strain, using primers that flank the Pbgl regulatory region into which IS1 and IS5 are known to insert (Supplementary Table S2). Of the deletion strains tested, ∆ihfA was selected for further study, since it caused the greatest change in the frequency of Bgl⁺ mutant appearances compared to WT (Figure 2A). Further confirmatory Bgl⁺ mutation assays were performed on WT and ∆ihfA to precisely ascertain the difference in mutant incidence between them. These results again showed an approximately eight-fold drop in total Bgl⁺ mutants in ∆ihfA, and a more than 11-fold drop in insertional mutants (Figures 2C,D). Deletion of ihfB, encoding IHF’s other subunit IhfB, showed the same effect as ∆ihfA (Figures 2C,D). To determine whether the ∆ihfA mutation prevented cell multiplication but not necessarily insertion, IS5 was inserted upstream of bgl in its usual position in the ∆ihfA mutant. This strain grew as quickly on salicin media as WT Bgl⁺ cells. This indicates that ∆ihfA acts specifically to lower the IS insertion frequency rather than hinder the growth of Bgl⁺ mutants after insertion takes place. These observations led us to consider the mechanism of IHF’s role in IS insertion, as well as its specificity for the bgl operon.

The ∆ihfA background was transferred to the backgrounds Ptet-G and Iq-G, which constitutively express bglG at the intS locus, leaving the native bgl operon intact. Our objective was to observe how IHF’s observed effect would interact with increased BglG levels, which has been previously shown to increase the rate of Bgl⁺ insertional and non-insertional mutations (Zhang et al., 2022). The results presented in Figures 3A,B show that the stimulatory effect of increased bglG expression on insertion frequency counteracts ∆ihfA’s negative effect on the same. As with the WT background, deletion of ihfA in either background led to a greater than 10-fold drop in insertional mutants. In the case of Ptet-G, the total mutants fell by four-fold when ihfA was deleted, and a more than 10-fold decrease was observed among insertional mutants. These data further solidify BglG’s effect on insertional mutation rate and confirm our results on ihfA deletion, which appears to mitigate both the insertional and non-insertional mutant incidences in the Ptet-G strain, but not the Iq-G strain.

Since the IHF protein has a significantly positive effect on IS insertion upstream of the bgl promoter, we decided to determine if changing the amounts of both IhfA and IhfB subunits would have a different effect. PK01, a strain expressing both the ihfA and ihfB genes in the chromosome using the constitutive Ptet promoter, was therefore constructed and subjected to a mutation assay as previously described. Interestingly, a decrease in both total and insertional mutants was observed after 10 days (Supplementary Figures S1A,B). We consider several possible reasons for why this occurred (see Discussion).

PK01_R was constructed, containing the tetR gene. tetR encodes the Ptet repressor protein TetR, which is most effective in the absence of its inducer, any one of several tetracycline derivatives. Our goal was to see if different levels of Ptet induction would reveal a similar trend as that already observed with normal PK01. Supplementary Figures S1C,D show the results of a mutation assay as described, but with the addition of several different concentrations of the Tet inducer anhydro-tetracycline (aTC). As expected, the increase of IhfA and IhfB led to more Bgl⁺ mutations. The number of mutants that appeared was like the previous assay at levels of maximal aTC induction (~30 μM), supporting the results of Supplementary Figure S1B.

To create a strain expressing both IHF subunits at even higher levels, a pZA31 plasmid containing Ptet-driven ihfA and ihfB was constructed and then transformed into wild type BW25113 and its counterpart containing the Tet repressor gene tetR. These new strains, called PK02 and PK02_R, respectively, were compared to each other and an isogenic control (WT Rf) in a mutation assay. All three strains behaved very similarly (Figures 4A,B), suggesting that the increased transcription of both IHF subunits did little to influence the insertion rate. To confirm this, PK02_R was plated on M9 salicin media containing differential levels of the Tet promoter’s inducer aTC (Figures 4C,D). The amount of inducer present did not appear to cause a corresponding change in total or insertional Bgl⁺ mutants, as consistent with the findings in Figures 4A,B.

Our next goal was to establish a mechanism by which IHF exerts its effect on the expression of bgl. If ∆ihfA’s effect on the frequency of Bgl⁺ insertional mutations is due to the inability of IHF binding to the bgl upstream region, then LacZ assays could show a difference in transcriptional activity when IHF is present versus when it is absent. The constructs we devised for determining this are shown in Supplementary Figure S2. The native bgl operon was left intact to maintain normal induction by β-glucosides. A second, altered bgl promoter and its upstream region were positioned in front of the native lacZ gene in place of its usual promoter. This new construct, Pbgl-Z, can measure the promoter activity of this second bgl operon. A similar construct, Pbgl-G-Z, contained the bglG gene between Pbgl and lacZ for the purpose of determining operon activity. Figures 5A,B show the lacZ activity results, which demonstrated a ~ 2-fold change in promoter activity but no significant change in operon activity between WT and ∆ihfA. We repeated these experiments using a BglB assay, which functions similarly in practice to the LacZ assay. However, this assay uses PNP-Glucoside (PNPG) as its substrate and measures activity of the bgl operon directly, rather than having to rely on additional constructs and the lacZ gene. The results of the BglB assay confirmed those of the LacZ experiments, showing no increase or decrease in bgl activity (Figure 5C). We suspected that any potential change may not be detectable due to the natural low levels of bgl activity; to investigate this possibility, the ∆Pbgl-G strains were constructed. ∆Pbgl-G strains are similar to the strains originally used for BglB but exhibit a 200-fold increase in operonic activity due to the removal of the bglG gene, both of its flanking terminator sequences, and the downstream H-NS binding site. To see if ∆Pbgl-G would lead to observable differences between strains, BglB assays were run again using several previously used strains now in a ∆Pbgl-G background. No large change was observed across the board, leading us to reject the notion that specific binding of IHF significantly affects transcription of bgl (Figure 5D).

Since our LacZ data did not reveal a substantial change in transcriptional activity within the context of bgl, we hypothesized that ∆ihfA does not affect bgl directly but may exert its effect on insertion upstream of bgl by a nonspecific means. If IHF, being nonspecific, is important for IS insertion in general, then it may be involved in the upstream process of IS excision from other locations in the genome. In this situation, other operons into which IS1 and IS5 insert would also experience a change in insertion frequency upon deletion of ihfA. To explore this possibility, a ∆ihfA strain was tested alongside an isogenic WT strain in mutation assays corresponding to several other operons where preferential IS insertion into their respective SIDD sites have been previously documented (McCalla et al., 1978; Chen et al., 1989; Whiteway et al., 1998; Zhang et al., 2010, 2017; Humayun et al., 2017). Agar plates were made with the intention of promoting insertion into the glpFK, fucAO, and flhDC operons, and were inoculated with a number of cells specific to the paradigm tested. The results for these three operons are presented in Figures 6A–C, and all three showed a significant decrease in insertional frequency, just as observed for bgl.

As when testing bgl, these experiments compared ∆ihfA to the WT strain. However, the parameters of each experiment differed depending on the specific operon being tested. Interestingly, all the operons examined showed a significant decrease in the insertional mutant appearance rate of the ∆ihfA counterpart, suggesting that the dependency of this process on IHF affects several operons into which insertions occur, and consequently, it may affect an early step (IS excision or transposition) of the IS1 or IS5 element itself.

Since our view of IHF’s effect proved to be relevant to other operons, the question remained: What is the mechanism of the IHF effect on transposition? First, we tested whether the DNA-binding ability of IHF was the cause of the observed multi-operon effect. A mutation assay was performed comparing WT and ∆ihfA to two other strains. These strains have Ptet-driven ihfA at the intS locus, but here the ihfA gene product contains a G-E substitution at the 62nd position (referred to as G62E). The product of this ihfA mutant loses its DNA-binding function but is still able to dimerize with IhfB (Granston and Nash, 1993; Hales et al., 1994). The ihfA_G62E construct was placed into WT as well as the deletion strain for ihfA at its native position (∆ihfA), yielding PK10 and PK11. If binding to DNA was important for IHF’s effect, then PK11 would show similar insertion rates as ∆ihfA, while PK10 would have binding and nonbinding IhfA and therefore would not experience as drastic of a change. Figures 7A,B show the results of a bgl mutation assay, conducted as previously described. The strains acted according to expectations, with PK10 having fewer mutants than WT, but more than its ∆ihfA counterpart PK11.

To confirm these results, the G62E constructs were also used in mutation assays for the glpFK and fucAO operons as previously described; the results of these are in Figures 7C–F. The G62E strains act similarly with respect to other operons as with bgl; in all operons tested, the G62E mutant generated around the same number of mutations as the WT if the native ihfA gene was intact. However, the G62E mutation in the ∆ihfA deletion mutant still caused a drastic drop to the same level of ∆ihfA or lower.

Research conducted by other laboratory groups had demonstrated that IHF binding sites are present at both ends of IS1 and on one end of IS5 (hereafter referred to as IS5C; IS5A is the end without an IHF binding site) (Gamas et al., 1987; Prentki et al., 1987; Muramatsu et al., 1988). We therefore decided to test whether the presence and/or binding of IHF causes a change in the level of transcription of either or both IS elements. This was accomplished by placing the promoter regions for IS1 and both ends of IS5 (containing the Inverted Repeats to which IHF binds) directly before the native lacZ gene (Figure 8A). This construct was placed in the WT, ∆ihfA, Ptet-driven ihfA at the intS locus, and Ptet-driven ihfA G62E at the intS locus (containing the G-E substitution at the 62^nd position that abolishes DNA-binding ability of IHF). The results for each reporter are presented in Figures 8B–D. Overall, no significant change was observed, regardless of the genetic background. Since IHF is known to bind to either end of IS1 and to IS5C, these results strongly suggest that IHF exerts its effect via direct binding to IS1 and IS5 without altering the expression level of either IS element.

To test whether Crp (when activated by cAMP) affects the rate of insertional mutations in the upstream region, two types of mutation assays were conducted. Simply testing a ∆crp mutant is not an option in this case, as Crp is required for expression of the bgl operon. If insertion takes place in a crp deletion mutant, we would be unable to observe it since the colony would not grow. Therefore, we first decided to perform mutation assays using WT on M9 salicin plates with minimal cAMP and compare mutation rates to those growing on plates with an excess of cAMP. The results of this experiment are shown in Figure 9A. They demonstrate no change in insertional mutants as the amount of extracellular cAMP increases.

Next, we compared these WT results to those of a cyaA deletion mutant. cyaA encodes adenylate cyclase, and without it, the cells are unable to produce cAMP and must rely on extracellular cAMP provided in the growth media. In Figure 9A, we observe an increase in insertional mutants as the level of extracellular cAMP increases. Interestingly, the number of total colonies was like those observed for the WT strain, suggesting that the lack of cyaA has a significant effect on IS insertion specifically. The next set of experiments used cpdA to control the levels of intracellular cAMP. cpdA encodes a phosphodiesterase which degrades cAMP to 5’-AMP, and in its absence, cAMP levels increase. Similarly, if cpdA is expressed at higher levels, the amount of cAMP, and therefore the amount of active Crp, should decrease. A cpdA deletion mutant was constructed and was given an “empty” pZA31 plasmid for isogeneity, producing ∆cpdA Rf. A pZA31 plasmid containing Ptet-cpdA was electroporated into WT cells, yielding PK20. All three strains were subjected to the bgl mutation assay as previously described. The results are presented in Figures 9B,C. The ∆cpdA mutant showed a more than two-fold increase in insertional mutants compared to the WT strain, and the cpdA overexpression strain PK20 showed a similar number of total colonies as WT, but with only half as many insertional mutants. Together, these data suggest that Crp, when activated by cAMP, is a positive regulator of insertion upstream of the bgl operon. In the absence of cAMP via deletion of cyaA or via increased degradation of cAMP, the insertion rate decreased 2–3-fold.

To observe how loss of ihfA affects mutation frequencies along with differing cAMP-Crp complex levels, the ∆cpdA Rf and PK20 backgrounds were transferred to the ∆ihfA strain. The results of this set of mutation essays are shown in Figures 10A,B. In the absence of IHF, increased Crp-cAMP by deleting cpdA led to a more than 2-fold elevation of Bgl⁺ mutations. Similarly, decreased Crp-cAMP by overexpressing cpdA led to a 3-fold reduction of Bgl⁺ mutations in the absence of IHF. Combining both IHF and Crp, these two proteins affect Bgl⁺ mutations by up to 33-fold (Figures 9B, 10A).

PK01, which replaced the native promoters of ihfA and ihfB with the constitutive Ptet promoter, had the surprising effect of lowering the insertion rate into bgl by nearly two-fold compared to WT. Additionally, the titratable strain PK02_R, which grows similarly to WT, grew only around half as many colonies when saturated with its inducer aTC. We here provide a few possible explanations for these phenomena. The first and most likely possibility is that the native promoters for ihfA and ihfB are already strong. Since IHF is a global protein with several functions in the cell and with genes in locations adjacent to other important genes, promoters for both subunits may be very strong (Pozdeev et al., 2022). It may be that the promoters driving these subunits are already operating at a higher rate of expression than Ptet itself. Therefore, in an effort to measure the mutational response of a cell by increasing expression, it is possible that replacement by Ptet, while strong, could actually have expressed ihfA and ihfB at a lower rate than the native promoter. Our data on the titratable strains PK01_R and PK02_R are in support of this hypothesis, because even at maximal levels of inductions, the mutation rate remains lower than WT.

Another possibility is that this strain is in fact expressing IHF at levels higher than WT, but that this has an overall negative effect on mutation frequency. For example, these increased levels of expression may be toxic to the cell, or the increased binding of IHF to the multitude of its recognized binding sites may have pleiotropic effects that lead to lower growth rates, and consequently less insertion.

A final possibility is related to the fact that IHF levels increase greatly when the cell is starving (Bushman et al., 1985; Ditto et al., 1994). It may be that this increase is regulated by transcription factors affecting the native ihfA/ihfB promoters under starvation conditions such as during the mutation assay. If we change the promoters to Ptet, we may well have inadvertently removed the cell’s normal ability to respond to starvation by upregulating ihfA and ihfB expression, which would also bring IHF levels to a lower-than-expected value.

cAMP-activated Crp, which is known to bind in bgl’s promoter region and positively regulate transcription, was also shown to positively regulate the rate of IS element transposition upstream of bgl. Results of a mutation assay comparing WT with an adenylate cyclase deletion mutant (∆cyaA) showed a decrease in the incidence of insertional colonies, although the overall Bgl⁺ colony counts remained comparable. Similarly, in backgrounds overexpressing or deleting the cAMP-degrading phosphodiesterase gene, cpdA, the rate of insertion decreased or increased, respectively. Thus, these results reveal a new role for Crp within the context of bgl as a regulator of transposition as well as transcription. Interestingly, previous data from our group have suggested that in the glpFK operon, cAMP-Crp has a negative effect on upstream IS insertion. It appears that while Crp maintains its status as a positive transcriptional regulator in the vast majority of operons which it influences, its role with respect to mutational changes in the bgl operon may be responsive to the sequences and/or DNA structures surrounding its binding site.

Crp has a known binding site in bgl’s promoter region, and this site is close to or overlapping the projected SIDD site into which IS elements insert (Humayun et al., 2017). This novel effect of cAMP-activated Crp presented here likely results from the increased energetic favorability for insertion to take place due to its binding. However, advances in computational programs that analyze energetics would be helpful to confirm this, as the current programs do not account for the binding of proteins to the DNA.

The IS1-lacZ reporter activity in section 2.7 proved to be low, even when compared to the two orientations of the IS5 reporter. After a thorough examination of the sequence of the construct, no mutation was found, leading us to consider other possibilities as to why IS1 exerted such a low operon transcriptional activity compared to IS5. We submit that either the growth conditions were not sufficient to increase IS1’s promoter to higher levels, or the region upstream of the IS1-lacZ gene construct lacks a site for a postulated positive transcription factor for the IS1 promoter. If so, it must be present in the other native locations in the genome where IS1 exists. Both considerations may independently or together explain why IS1’s reporter showed such low activity.

As stated in the Results section, both ends of IS1 and at least one end of IS5 contain IHF binding sites (Gamas et al., 1987; Prentki et al., 1987; Muramatsu et al., 1988). However, LacZ assays measuring the activities of the IS1 and IS5 promoters showed no clear change when the level of IHF binding was altered. Despite this, our other data indicate a non-specific effect of IHF on transposition of IS1 and IS5. Thus, the effect exerted by IHF on these IS elements must be of a nature other than transcriptional. Both IS1 and IS5 have DDE catalytic regions (Ohta et al., 2002; Curcio and Derbyshire, 2003), and may therefore act similarly for the purposes of replicative excision from the E. coli genome. Our IHF mutation assay data supports this; if IHF only affected one element’s transposition and not the other, then only a 2-fold change at best would be expected, and all the insertional colonies would contain the same element, instead of the roughly equal levels of IS1 and IS5 insertion that has been classically observed (Schnetz and Rak, 1988; P.K., unpublished data). Work on other transposons with DDE activity has suggested that IHF plays an important, but not essential, role in the efficient transposition of Tn10 and the Mu bacteriophage (Kobryn et al., 2002; Swinger and Rice, 2004; Gueguen et al., 2005). The role of IHF involves assisting in the conformation of the transposon via binding so that the transposase protein can effectively bind to the transposon ends and then dimerize (or tetramerize in the case of Mu). It has been shown that when IHF is not present, DNA supercoiling has a compensatory effect, allowing for some transposition to occur even when an important conformational factor is not present (Chalmers et al., 1998). Interestingly, Tn10 in vitro experiments mimicking in vivo supercoiling levels in the absence of IHF observed a ~ 10-fold decrease in transposition frequency (called the “basal level” of 27 excision), which is a similar decrease to what we observed in the ihfA deletion mutant (∆ihfA) for the Bgl⁺ mutation assay (Chalmers et al., 1998; Figure 2A). We utilized in vivo models to support past findings in relation to other transposable elements and draw further similarities between IS elements and other transposons in the DDE transposase family.

The promoter region (−55 to +30 relative to the insA translation initiation site) driving the IS1 transposase gene insAB1 was amplified using IS1p-Xho-F and IS1pBam-R from BW25113 genomic DNA. This region contains all of the upstream region (including the left-hand IR) and the first 10 residues of insA (plus a stop codon TAA). The amplified products were digested with XhoI and BamHI and cloned into the same sites of the default integration vector, pKDT (Klumpp et al., 2009), yielding pKDT-PIS1. The region carrying the km^r, rrnBT and PIS1 (km^r:rrnBT:PIS1) was PCR amplified using oligos IS1- Z-P1 and IS1-Z-P2 (Supplementary Table S2) and then integrated into the chromosomal default strain EQ42 (Klumpp et al., 2009) to replace the lacI gene and the lacZ promoter. The resultant strain carries the km^r:rrnBT:PIS1 cassette followed by lacZ’s ribosomal binding site (RBS) and the lacZ structural gene within the lac locus. After being confirmed by PCR and sequencing, the promoter reporter, PIS1, driving lacZ expression (that is, PIS1-lacZ) was transferred into BW25113 and various genetic backgrounds by P1 transduction. This yielded the IS1 promoter reporter strains IS1-Z, ΔihfA_IS1-Z, PK14, and PK15, respectively.

IS5 carries three open reading frames (ins5A, ins5B, and ins5C). ins5A encodes the main transposase, and it is transcribed from its own promoter located close to the left-hand IR while the divergent ins5B and ins5C genes may form an operon that is driven by another promoter located close to the right-hand IR (Sawers, 2005). The ins5A promoter region (−68 to +30 relative to the ins5A translational initiation site) and the ins5CB promoter region (−207 to +30 relative to the ins5C translational initiation site) were individually cloned into pKDT, yielding pKDT_PIS5A and pKDT_PIS5C, respectively. As for PIS1 described above, PIS5A and PIS5CB each carries the first 10 amino acids from the target gene followed by a stop codon TAA. These promoters were integrated into the lac site on BW25113 chromosome as for PIS1-Z, yielding strains IS5A-Z and IS5C-Z, respectively. Using P1 transductions, these new IS5 promoter lacZ reporters were transferred to other genetic backgrounds by P1 transduction. This yielded the IS5 transposase promoter (PIS5A) reporter strains ΔihfA_IS5A-Z, PK16, and PK17, as well as the IS5C promoter reporter strains ΔihfA_IS5C-Z, PK18, and PK19.

Escherichia coli reporter strains were cultured in 4 mL of LB contained in glass test tubes (1.5 cm in diameter × 15 cm in length) with shaking at 37°C for 8 h. An amount of 30 μL of LB cultures were used to inoculate 3 mL of M63 minimal media in smaller glass tubes (1.2 cm × 12 cm), and the tubes were shaken at 37°C overnight. The carbon sources were 0.5% glycerol, 0.5% salicin, or both. The tubes were rotated at 250 rpm and 37°C, and cell densities (OD₆₀₀) were measured with a Bio-Rad spectrophotometer. During the exponential growth phase, four samples were collected in the range of OD₆₀₀ from 0.1 to 1. The samples (roughly 0.3 mL for promoter reporter strains, 0.9 mL for the IS1/5 promoters), and 0.6 mL for operon reporter strains were immediately frozen at −20°C prior to β-galactosidase assays. To measure β-galactosidase activities in bgl promoter reporter strains, 0.8 mL of Z-buffer containing β-mercaptoethanol (2.7 μL/mL) and sodium dodecyl sulfate (SDS) (0.005%) was mixed with 0.2 mL of sample and 25 μL of CHCl₃ in test tubes. Alternatively, for bgl operon reporter strains, 0.5 mL of Z-buffer was mixed with 0.5 mL of the sample. The tubes were vortexed twice (each time for 10 s at a constant speed) and incubated in a 37°C water bath until temperature equilibration. A 0.2 mL aliquot of O-nitrophenyl galactoside (ONPG) substrate (4 mg/mL) was then added to each test tube. When a yellow color developed, the reaction was stopped by adding 0.5 mL of 1 M Na₂CO₃ followed by vortexing. Reaction mixtures were centrifuged (15,000 rpm, 3 min), and the absorbance values of the supernatants were measured at 420 nm and 550 nm. A control tube was run in parallel using M63 salts instead of the test sample. β-galactosidase activity (Miller units) = [(OD₄₂₀–1.75 × OD₅₅₀)/(sample volume in mL × time in min)] × 1,000 (Miller, 1972). For a given test strain, the slope of OD₆₀₀ values versus β-galactosidase activities was referred to as the promoter activity or the operon activity.

To test the effect of IHF on bglGFB transcription, the bgl promoter reporter (Pbgl-Z) and the bgl operon reporter (Pbgl-G-Z) (Tran et al., 2022) were individually transferred to an ihfA deletion background. This yielded strains ∆ihfA_Pbgl-Z and ∆ihfA_Pbgl-G-Z.

β-glucosidase assay was performed as described in our previous study (Lam et al., 2022). Briefly, 20 μL of LB culture was transferred to M63 minimal medium with 0.66% casamino acids (CAA). After overnight growth at 37°C with shaking, the culture was diluted into 6 mL of M63 + 0.66% CAA +0.5% salicin minimal medium at a starting OD₆₀₀ of 0.025. The cells were grown at 37°C with shaking. Five samples with 0.8 mL each were collected during the late exponential growth phase when the bgl operon was fully induced with an OD₆₀₀ of above 1.5. The samples were centrifuged at a speed of 5,500 rpm for 2.5 min and the cells were suspended with 1 mL of Z-buffer with 50 μg/mL chloramphenicol.

To measure β-glucosidase (BglB) activities, test samples were warmed in a 37°C water bath, and 200 μL of p-nitrophenyl-β-D-glucoside (PNPG, 8 mg/mL) was added to the cell suspension in Z-buffer. After a visible yellow color appeared, the reaction was terminated by adding 0.5 mL of 1 M Na₂CO₃ and subsequently vortexing. The reaction mixture was centrifuged, and the absorbance of the reaction mixture was measured at 420 nm and 550 nm. The BglB activity of each sample was calculated using the equation: β-glucosidase (BglB) activity = [1,000 × (OD_420 nm – 1.75 × OD_550 nm) × Dilution factor]/[OD_600 nm × Time of reaction (min) × Volume of sample (mL)]. The activity of the strain was determined by averaging the BglB activities of the samples measured.

β-glucosidase (BglB) assays were conducted in wild type and ∆ihfA backgrounds. To see the IHF effect on the cells with higher operon expression, strains G50 (deleted for the two terminators flanking bglG) and ∆Pbgl-G (deleted for Pbgl and bglG together with two terminators; Lam et al., 2022) was used.

Bgl⁺ mutation assays were performed on minimal M9 agar plates with 0.5% of a β-glucoside (salicin) as the sole carbon source. Strains to be tested (from single fresh colonies) were cultured in LB liquid medium for approximately 7 h at 37°C, washed twice with carbon source-free M9 salts (M9) and applied onto plates (2 × 10⁷ cells/plate). The plates were then incubated in a 30°C incubator and were examined every 2 days for the appearance of Bgl⁺ colonies, with each colony representing a new Bgl⁺ mutation. On these β-glucoside minimal agar plates, any colonies appearing by day 2 were considered to be from Bgl⁺ cells initially applied onto the plates. They were therefore subtracted from the subsequent measurements. IS1 and IS5 have been previously shown to insert into the same upstream bgl promoter region. Colony PCR using primers flanking this region was used to differentiate between insertional and non-insertional mutants. Mutation frequency was determined in the manner described in Cairns and Foster (1991) and Zhang and Saier (2009b).

For the mutation assays which used differential levels of aTC, the growth media was prepared with different amounts of aTC to obtain several ng/mL concentrations as outlined in Results 2.3.

To determine the effects of other carbon sources on Bgl⁺ mutations, mutation assays were performed on minimal M9 agar plates with 0.25% glycerol or 1% propanediol as the sole carbon source. To determine the total populations, the cells were washed off the minimal M9 agar plates at relevant time points, serially diluted, and plated onto LB agar plates. To determine Bgl⁺ populations, appropriate dilutions were applied on M9 + salicin agar plates. The frequencies of Bgl⁺ mutations were determined as described above for Bgl⁺ on M9 + salicin agar plates.

Glp ⁺ mutation assays were conducted on glycerol M9 minimal agar plates as described in Zhang and Saier (2009b). Strains ∆crp and ∆crp∆ihfA were used for the mutation assays.

Using the wild-type and ΔihfA strains, the swarming mutation assays for the appearance of hyper-swarming mutants (outgrowing subpopulations from the inoculated cells) were carried out using the method of Barker et al. (2004). Briefly, overnight cell cultures in LB media were washed once with M9 salts and diluted to an OD₆₀₀ of 1.0 prior to use. Two microliters of the cell suspensions were streaked across the centers of LB semisolid (0.3% agar) plates (diameter = 9 cm) using a plastic transfer loop. The plates were incubated at 30°C. The swarming mutants, represented by outgrowths of motile subpopulations from the streaked cells, were counted. The mutation frequency was normalized as outgrowths (mutations) per 9-cm cell streak. Insertional mutants were verified via colony PCR.

The assay for PPD ⁺ mutations was conducted by applying cell suspensions from fresh overnight cultures onto propanediol (1%) M9 minimal agar plates (~10⁸ cells/plate). The wild-type and other strains were tested over a period of several days as in the Bgl⁺ mutation assay. Total populations and mutation frequency were determined as described above under “Bgl⁺ mutation assays.”

The program R was used for all statistical analyses. p-values were generated to determine statistical significance, with values under 0.05 treated as statistically significant. The strains tested were plated along with known controls for each experimental run. Results shown are from a minimum of 3 separate runs/strain, with each run consisting of several plates. The data reported on the figures are expressed as the mean of a minimum of three runs for each strain at each time point. LacZ activity assay results are expressed as the slope of a line generated from 4 time points/run and a minimum of 2 runs (at least 8 data points total). BglB assay results are expressed as the mean of 12 data points from a minimum of 3 separate runs.