Materials and methods
Bacterial strains and strain construction. Bacterial strains were selected from replicate populations started with strains REL606 and REL607, which are isogenic except for a neutral arabinose marker and a mutation in recD that also appears to be neutral (Tenaillon et al. 2016). Populations were evolved in Davis-Mingioli (DM) minimal media supplemented with different presentations of glucose (175 μg/ml) and lactose (210 μg/ml) (Cooper and Lenski 2010). The evolution environments included DM supplemented with lactose (Lac), daily fluctuations of glucose and lactose (G/L), or long-term switching from glucose to lactose (G_L) or from lactose to glucose (L_G) every 2,000 generations. Each population was evolved for 8,000 generations, except for one G/L population, which was evolved for 7,000 generations (G/L2). Six replicate populations were evolved in each environment, the replicate number is indicated by the number following the evolution environment designation. Clones with the lacI- mutation were selected from populations based on their growing as a blue colony on indicator plates that contained X-gal (5-bromo-4-chloro-3-indolyl-beta-D-galactopyranoside) and glucose (TGX plates) (Fig. 1; Quan et al. 2012). On TGX plates, the blue phenotype is indicative of a strain constitutively expressing the lac operon. A total of nine clones were selected, one from each of three lactose only populations, a G_L population, and five G/L populations. In all clones the lacI gene was amplified and sequenced to verify thatlac operon constitutive expression was caused by a mutant LacI repressor (Fig.1). Amplification was carried out using the primers: 5’-GCGGAGCTGAATTACATTCC-3’ (11-F) and 5’-GGGTGCCTAATGAGTGAGCT-3’ (12-R).
To determine the effect of substituted lacI- mutations, we first constructed lacI + derivatives of our focal evolved strains. To do this, we PCR amplified the ancestral allele using primers 11-F and 12-R, ligated this product into pCR2.1 using a TA cloning kit (Invitrogen, CA USA). The plasmid was used to transform TOP10F’ cells, which were blue/white screened to identify transformants having a plasmid insert. pCR2.1::lacI + plasmids were purified, and the lacI + fragment excised and cloned into pDS132 using enzymes SacI and XbaI. The resulting plasmid, pDS132::lacI+ was used to transform MFDpir cells (Ferrières et al. 2010). MFDpir(pDS132::lacI +) was separately conjugated with each target evolved strain by mixing donor and recipient at a 1:2 ratio, respectively. Conjugation was carried out on LB agar plates supplemented with 2,6-diaminopimelic acid (30 μM), which MFDpir needs for growth, for 3-4 hours at 37 ° C. The conjugation mixture was resuspended in 200 μL DM medium and plated onto minimal glucose agar (MG) supplemented with chloramphenicol (Cm; 20 μg/mL) to select for recipient strains with the pDS132 plasmid successfully integrated into the chromosome (Philippe et al. 2004). After overnight incubation, six colonies were restreaked onto MG + Cm agar, again incubated overnight, and then restreaked a second time to the same medium. A colony descended from each of the original six was resuspended in 500 μL DM base liquid media and plated onto sucrose plates supplemented with X-gal to select for excision of the pDS132 plasmid (Philippe et al. 2004) and screen for clones that had retained the introduced lacI + allele. One blue colony and one white colony were selected from each plate and restreaked twice onto sucrose + X-gal plates. White colonies indicate clones that successfully integrated thelacI + allele. Blue colonies, which retained the originallacI- allele were used as negative controls to test for the presence of additional mutations that occurred during the allele replacement process. All clones were tested for proper pDS132 plasmid excision by checking for chloramphenicol sensitivity. The lacI + insertion was verified by sequencing.
To evaluate the possibility of secondary mutations being inadvertently added during construction of lacI + strains, we measured the fitness of control strains that went through the conjugation process but did not retain the introduced lacI+ allele in competition with an otherwise isogenic strain with a distinct neutral ara marker. Theara marker strains were constructed either by pairwise conjugations between the evolved strain and MFDpir (pDS132::ara -) (same method as above except X-gal was not used in the media), or by plating 100 μL of overnight LB cultures onto minimal arabinose agar (MA) and selecting for spontaneous ara + mutants. Indistinguishable fitness between competing strains was interpreted as indicating the absence of fitness-effecting secondary mutations.
Fitness assays. Fitness effect of the lacI - mutation in each isolated strain was measured in glucose (175 μg/ml) and lactose (210 μg/ml) only environments. Cells were transferred from freezer stocks initially to LB medium, then, after overnight growth, to minimal medium supplemented with glucose or lactose as used in the particular competition assay. Strains were initially acclimated to the assay environment over two 24 hr transfer cycles with a 1:100 dilution occurring between each cycle. Pre-conditioned competitors were mixed at a 1:1 ratio by diluting each competing strain 1:200 directly into the assay environment. Competitions were carried out over two (lactose competitions) or four (glucose competitions) transfer cycles. On the initial and final day of competitions, cells were plated onto TGX plates and incubated at 37 ° C for 18-20 hours in order to distinguish competing genotypes. Relative fitness effect of the lacI - mutation was determined based on the change in density of blue (lacI -) and white (lacI +) colonies on TGX plates using the formula: ln(blue2 × 100t / blue0)/ ln(white2 × 100t / white0), where subscripts indicate the time at which competitor density was estimated, and t accounts for transfer cycles during the competition (Lenski et al. 1991). Test competitions checking for additional mutations occurring during allele replacement procedures or selection of spontaneous ara + mutants were performed as described above, except competitors were distinguished by plating onto tetrazolium arabinose agar (TA).
Expression assays. Expression of the lac operon was measured using a GFP reporter construct controlled by the Plac promoter region including the O1 and O3 LacI operator sequences, and native primary CRP binding site (Quan et al. 2012). This reporter was previously cloned into a mini-Tn7cassette in a suicide-vector (Quan et al. 2012) and was introduced into target strains by tri-parental conjugations between a target recipient evolved strain, a donor strain (MFDpir(pUC18R6KT::Plac -GFP, kanr)), and a helper strain (MFDpir (pTSN2)) (Choi et al. 2005; Ferrières, et al. 2010; Quan et al. 2012). Strains were mixed at a 3:1:1 ratio (recipient : donor : helper) on LB + DAP (30 μM) agar and incubated at 37 ° C for 3 hours. The conjugation mix was resuspended in DM medium and plated onto LB + kanamycin (Km; 60 μg/mL) + isopropyl β-D-1-thiogalactopyranoside (IPTG; 1 mM) agar plates. Kanamycin selects for clones that successfully obtained the Plac -GFP reporter while IPTG induces expression of the reporter allowing identification of clones that stably expressed GFP. After 24-36 hours of growth, six fluorescent colonies were restreaked onto LB + Km (60 μg/mL) plates. Restreaked colonies were streaked a second time on LB + Km + IPTG plates and then tested for absence of the delivery plasmid by spotting colonies on LB + Ap (100 μg/mL) agar. Insertion of the Plac -GFP reporter into theatt Tn7 site was confirmed by PCR (primers: 5’-TAACAGCCAGCACCACGCCG-3’ (120-F) and 5’-CGCGAATCCGATCTGGCGCT-3’ (121-R)). Transposition of the reporter into each recipient strain’satt Tn7 site allows consistent insertion of the reporter into the same region of the genome, minimizing divergent effects on reporter fitness and expression costs.
To measure Plac -GFP reporter expression, reporter strains were grown overnight in LB broth from freezer stocks, diluted 1:10,000 in the assay environment, and allowed to grow for 24 hours. The following day cultures were diluted again 1:10,000 in the assay environment and grown in a VersaMax spectrophotometer (Molecular Dynamics, CA) until mid-log phase (OD450~0.1-0.15) to allow measurement of a steady-state level of Plac -GFP reporter expression. Plac -GFP expression was measured in an Accuri C6 flow cytometer (Becton Dickinson, NJ). Assay environments included DM+175 μg/mL glucose, DM+210 μg/mL lactose.
Statistical analyses. Data was analyzed using R (version 3.5.0) (R Core Team, 2018). Dunnett’s tests for one against many comparisons were performed using the glht function in the multcomp package. Where appropriate, experimental block was included in analyses as a random effect in mixed-model ANOVA performed using the lmer function in the lme4 package. Fixed effects were tested for significance by comparing models fitted with and without the factor of interest using a χ2 test .