Three strains of symbiotic V. fischeri were experimentally evolved under various pH concentrations: V. fischeri, ES114 (version: ES114-L; host Euprymna scolopes, from Kaneohe Bay, Hawaii), V. fischeri ET401 (host E. tasmanica, from Townsville, QLD, Australia), and V. fischeri EM17 (host E. morsei from Tokyo Bay, Japan) at pHs of 6.5, 6.8, 7.2, 7.6, and 8 for 600 generations (Table 1). Based on a pilot study, V. fischeri at pH 5.5 and 6 were unable to survive in vitro and were not examined further. pH levels for this study were chosen to represent acidic conditions of the light organ (6.5–7.6) and a control pH of 8.0 (seawater pH). For biological significance, 10 independent lines of clonal isolates from each strain were experimentally evolved simultaneously. Chromosomal insertion of the chloramphenicol resistance marker Cam^R into the genomes of V. fischeri ET401 and EM17 were made with a mini-Tn7 cassette. The Cam^R cassette was used as a selective agent for discerning between ancestral (WT-control, no Cam^R) and evolved strains (carrying Cam^R). V. fischeri JRM200, an isogenic strain of V. fischeri ES114-L with the mini-Tn7 Cam^R, was obtained from earlier studies (McCann et al., 2003; Soto et al., 2012). Briefly, E. coli strain CC118λpir carrying pEVS104 and another strain of E. coli BW23474 carrying pUX-BF13 (Visick and Ruby, 1997; McCann et al., 2003) were combined with the recipient V. fischeri strain in a tri-parental mating-bacterial conjugation step and incubated for ~12 h at 28°C. The conjugation mixture was then re-suspended in SWT [seawater tryptone; per liter composition: 700 mL of seawater (30 g of Instant Ocean and 10 g of Marine Mix, 300 mL of dH₂O), 5 g tryptone, 3 g yeast extract, 3.75 mL 80% glycerol] media and spread onto SWT-agar plates (15 g agar/L) with chloramphenicol (20 μg/mL). V. fischeri JRM200, ET401 (Tn7 Cam^R), and EM17 (Tn7 Cam^R) were then streaked onto SWT agar plates media with chloramphenicol (20 μg/mL), then incubated at 28°C for 18–24 h. Individual clonal isolates (single colonies) were transferred to 5 mL of SWT with chloramphenicol (20 μg/mL) liquid medium with the selected pH levels and placed in a 28°C shaking incubator set at 225 rpm for 8–12 h. When growth reached OD_600nm = 0.5, 10 μL of each culture was transferred (1:500 dilution) to 5 mL of fresh SWT media with chloramphenicol (20 μg/mL) at their respective pH. Experimental evolution was sustained in 18 × 150 mm glass test tubes kept in an Innova® 43R incubator shaker (New Brunswick, Enfield, CT) at 28°C, 225 rpm.

Previous studies using experimental evolution on E. coli (Wielgoss et al., 2013) illustrated how most mutations under selective pressure occur during the exponential phase of bacterial growth; therefore, all 10 lines of the clonal isolates from the three strains at all pH levels were transferred at the end of log phase. For consistent timing and ease of calculations, culture transfers were completed every 12 h at a 1:500 dilution, correlating to the end of log phase, but prior to stationary phase. Generations of evolving bacteria were determined by calculating the specific generation times of V. fischeri JRM200, ET401 (Tn7 Cam^R), EM17 (Tn7 Cam^R). Calculations were completed assuming that bacterial growth is a first-order chemical reaction, where μ = (lnN–lnN₀) (t–t₀) and g = ln2/μ, where μ represents the growth rate, N is the number of cells at time t, N₀ is the number of cells at t₀, and g is the generation time (Soto et al., 2009). Using the previously calculated generation time of each strain for all five pH levels, evolutionary “frozen fossil records” were generated by freezing down cultures with 20% glycerol final volume every 100 generations, and storing each in 2 mL cryo-vials (Wheaton, NJ) at −80°C. To check for contamination and experimental progress, 50 μL of each culture that was frozen was also spread on SWT-Cam agar plates and examined for the viability of strains through growth studies the following day. All cultures were continually transferred to fresh SWT-Cam (20 μg/mL) at their respective pH levels. Frozen fossil records were collected at 0, 100, 200, 300, 400, 500, and 600 generations. For the experimental calculations we used three replicates from each of the 10 lines of frozen fossil records of the evolved strains from each 100th generation. Each pH treatment had 90 samples (three replicates × 10 lines × three strains).

All ten lines of evolved bacterial strains at 100, 200, 300, 400, 500, and 600 generations and the ancestral (control) strains were grown in SWT liquid media with pH levels adjusted to of 6.5, 6.8, 7.2, 7.6 and 8 by acetic acid (99.7+% A.C.S reagent, Sigma-Aldrich, St. Louis, MO) and measured with an Orion PerpHecT pH Meter (Model 330—Thermo Scientific, Waltham, MA). pH had no negative affect on the antibiotic used throughout this study for all evolved isolates.

Cultures were shaken at 225 rpm at 28°C overnight, then sub- cultured at a 1:500 dilution into new SWT media for all strains at all pH levels measured. All samples were grown to OD_600nm of 0.3 at 28°C and shaken at 225 rpm in an Innova® 43R incubator shaker (New Brunswick, Enfield, CT). Cells were inoculated in 5 mL SWT media at 1:10 dilution in triplicate (3 independent bacteriological test tubes (glass, 18 × 150 mm) per each time point/pH levels) for a 12-h growth study. Reading of all triplicate samples were taken every hour by Spectramax spectrophotometer (Molecular Devices, Sunnyvale, CA). The term Relative Light Unit-RLU for bioluminescence measurement is used since luminometers typically do not yield a measurement directly in units of photons. During the measurements, the Luminoskan Ascent luminometer (Thermo Scientific, Waltham, MA) was set to one light unit and all other measurements were made relative to the set value. Bioluminescence (Relative light units, RLUs) was measured simultaneously every hour with a Luminoskan Ascent luminometer (Thermo Scientific, Waltham, MA). Growth rates of samples were determined by calculating the generation time of each bacterial strain during each growth study (Soto and Nishiguchi, 2014). The factors of Relative Light Unit (RLU) measurement per generation at specific pH levels for each strain were used in a two-way ANOVA analysis for statistical analysis of the bioluminescence assay. GraphPad Prism (8.0.0) software was used for all analysis (p < 0.001, N = 10).

For growth assays, the two-way ANOVA analysis (GraphPad Prism, 8.0.0) were conducted using values of the growth rate per strains (evolved and ancestral bacteria) in various pH levels at different generations (p < 0.001, N = 10).

To determine whether pre-adaptation to low pH impacts animal host colonization compared to non-evolved strains, infection assays were performed as previously described (Nishiguchi, 2000; Chavez-Dozal et al., 2014). For single and competition infection experiments, overnight cultures of the 600 generation wild-type (control) ES114, EM17, ET401 and evolved JRM200, ET401-Tn7, EM17-Tn7strains were regrown in 5 mL of fresh SWT medium at their respective evolved pH until they reached an OD_600nm of 0.3. Cultures were then diluted to ~1 × 10³ CFU/mL in 5 mL of sterile seawater and added to glass scintillation vials where newly hatched juvenile squids were placed (one individual/vial). Three independent plates of the same inoculation seawater with each respective strain were plated for all samples, and CFUs were counted the next day for the accuracy of the initial inoculum. Seawater (pH = 8) was changed with fresh non-sterile seawater (without bacteria) every 12 h over a period of 48 h. Animals were maintained on a light/dark cycle of 12/12. After 48 h colonization (48 h = standard colonization assay time), animals were sacrificed and homogenized, and the diluted homogenate was plated onto SWT agar plates for both wild-type and evolved V. fischeri, and SWT with chloramphenicol (20 μg/mL) for the V. fischeri evolved strains. A second set of animals were selected for competition studies, where juvenile squids were co-infected with the ancestral strain (WT-control) and their respective evolved strain, sacrificed after 48 h, homogenized, and the diluted homogenate plated onto SWT and SWT with chloramphenicol (20 μg/mL) agar plates. The concentration of CFUs was the same as the prior set of studies as well as in the competition studies, with the combined total being the same CFUs as the single infections (5000 CFUs total). Colony forming units (CFUs) were counted the next day to determine colonization efficiency of each strain. For calculation accuracy, the plates containing SWT with chloramphenicol (20 μg/mL) were subtracted to quantify evolved strains colonies from wild-type (control). This was due to the fact that only the evolved strains have the capacity to grow on agar plates containing chloramphenicol (20 μg/mL) whereas both ancestral and evolved will grow on SWT without chloramphenicol. A total of 15 animals/strain were used for each competition assay, and five non-infected (aposymbiotic) juveniles were used as negative controls for each sample run. We chose evolved strains at pH 7.2 since these had the most significant change compared to all other pH concentrations measured, and due to our limited supply of animals for the study. Results of single colonization and competition study were analyzed using two-way ANOVA analysis with GraphPad Prism (8.0.0). The factors of analysis for this study are calculated CFU/Light Organ values per strains for both single colonization and competition studies (p < 0.001 and N = 15).

Evolved V. fischeri strains JRM200, ET401-Tn7, and EM17-Tn7 at 600 generations demonstrated significantly different growth rates at pH 6.5, 6.8, and 7.2 when compared to their respective ancestral strains (T = 0) ES114-WT, ET401-WT, and EM17-WT at the same pH (Figures 1A–F). The two-way post hoc ANOVA analysis demonstrated significant effects for pH levels 6.5, 6.8, and 7.2 among all bacterial strains. The post-hoc ANOVA interaction analysis between various pH levels for different generations of bacterial strain shown significant differences (p < 0.001, N = 10). pH 7.2 shown significant differences at 400, 500, and 600 generations in comparison to pH 6.5, 6.8, 7.2, and 8. pH 7.6 and pH 8 shown no significant differences among each other, but there were significantly different with other pH levels. pH 6.5 and 6.8 were shown no significant differences among each other, but there were significantly differ among other pH levels at different generations (p < 0.001, N = 10).

Bioluminescence increased from 300 to 600 generations with significantly higher luminescence at 600 generations in relative light units for evolved V. fischeri strains JRM200, ET401-Tn7, EM17-Tn7 at pH 6.5, 6.8, and 7.2 when compared to their ancestral strains at the same pH (Figures 2A–C). The two-way post hoc ANOVA analysis demonstrated significant effects for all pH levels among all bacterial strains with respect to the Relative Light Unit (RLU) of bioluminescence. The post-hoc ANOVA interaction analysis between various pH levels for different RLU measurements per generation of each bacterial strains showed significant differences. pH 7.2 showed significant differences at 400, 500, and 600 generations in comparison to pH 6.5, 6.8, 7.2, and pH 8. pH 7.6 and 8 showed no significant difference among each other, but there were significantly different with other pH levels. pH 6.5 and 6.8 had no significant difference among each other, but they were significantly different among other pH levels at different generations (p < 0.001, N = 10).

Bioluminescence results show that relative light units (RLUs) of low pH adapted V. fischeri strains ES114, ET401, and EM17 at pH 6.5, 6.8, and 7.2 increased from 400 to 600 generations (Figure 2). These results suggest that adaptation to lower pH may also enhance the bioluminescence activity of evolved V. fischeri strains, which would entail changes in lux operon functionality and alternations to the metabolic changes that influence input/output from associated light producing proteins (Wier et al., 2010; Wilson, 2013). Bacteriogenic bioluminescent animals such as squid and fish have evolved organs specifically dedicated to host luminescent bacteria (e.g., V. fischeri, V. logei), whose bioluminescence is used during certain behaviors such as hunting and counterillumination (Jones and Nishiguchi, 2004; Haddock et al., 2010; Bose et al., 2011). The observed increase in bioluminescence activity of evolved strains in this study suggests that maintenance of this important symbiotic feature gives the cell an evolutionary advantage over ancestral strains (Stabb and Visick, 2013). Given that luminescence is important in establishing and maintaining the symbiosis, it is not unlikely that biofilm production is also affected by this phenomenon. Recent proteomic studies on biofilm production between free-living and symbiotic states in V. fischeri have detected a bioluminescence regulatory protein which is responsible for regulating the quorum sensing cascade and subsequently increasing exopolysaccharide production in the biofilm matrix (Chavez-Dozal et al., 2015). Although differences in biofilm formation between ancestral and evolved strains were not examined in this study, biofilm production may be affected similarly to bioluminescence and other symbiotic loci when V. fischeri is evolved at low pH. Previous research has demonstrated that the day-night oscillation of V. fischeri's metabolic pathways shift from aerobic respiration to fermentation, and the concomitant expression of genes from catabolic cycles suggest variation from a neutral to an acidic pH level in the light organ crypt (Wier et al., 2010; McFall-Ngai, 2014). Thus, enhanced bioluminescence activity that is dependent upon high bacterial cell density in the light organ can be impacted when V. fischeri is resistant to low pH (Jones and Nishiguchi, 2006; Guerrero-Ferreira and Nishiguchi, 2010). Future studies will focus on the regulation of bioluminescence and biofilm related proteins both in ancestral and evolved strains to determine how adaptation to low pH influences these specific symbiotic loci. Additionally, understanding the synergism of low pH adapted genes to other regulatory networks responsible for bacterial physiological behavior during infection and colonization will further our knowledge on the genetic diversity of V. fischeri to such conditions in the wild.

Symbiotically competent V. fischeri that are free-living in the environment have evolved to form a close and persistent association with host squids (McFall-Ngai, 2014; Schwartzman and Ruby, 2016). Likewise, these symbiotic V. fischeri must be able to coexist with the host and be recognized as “beneficial” by the squid immune system, while maintaining features that allow a successful colonization (Sachs et al., 2011). The evolution of specific recognition, colonization, and persistence mechanisms both in the symbiont and host are therefore a combination of environmental and host specific factors that enable the success of such beneficial associations (Sachs et al., 2011; Soto et al., 2012, 2014; Guerrero-Ferreira et al., 2013). For example, cooperation between Rhizobia and leguminous plants occurred due to the benefit of nitrogen fixation, which is a very costly process (Sachs and Simms, 2008). In the squid-Vibrio model system, similar phenomena occur when the bacteria produce bioluminescence for the host, while in return the host provides a nutrient rich environment for growth (Wier et al., 2010; Wernegreen, 2013; McFall-Ngai, 2014; Soto and Nishiguchi, 2014; Pepper et al., 2015). The squid host maintains the symbiosis and controls cell density by venting 95% of its Vibrio symbiont to the surrounding environment at dawn. Thus, the host has evolved a mechanism to cull and maintain only the “best fit” bacteria that can produce luminescence for counterillumination (Jones and Nishiguchi, 2004; Wier et al., 2010; Wernegreen, 2013; McFall-Ngai, 2014; Soto and Nishiguchi, 2014; Pepper et al., 2015). V. fischeri that have evolved to outcompete and survive this diel cycle of venting have been “selected” through multiple generations as in an experimental evolution scenario. Since light production is a costly metabolic process, oxygen levels in the light organ slowly decrease throughout the night causing this drop in pH (Schwartzman et al., 2015; Schwartzman and Ruby, 2016). Whether pH adapted vibrios are better able to accommodate this change, thereby impacting the repopulation efficiency to positively influence the number of pH adapted symbionts may explain how environmental conditions complement V. fischeri's success as a beneficial microbe. Since our pH adapted V. fischeri strains were shown to be more fit than their ancestral lines both in vitro and in vivo, this may be indicative of specific loci that are beneficial for both pH accommodation and host colonization and persistence. Additionally, symbiotic loci in V. fischeri have been found to be more conserved than homologs in free-living strains (Howard et al., 2015). One example is the mannose sensitive hemogglutinin proteins (MshA) that are often induced by environmental factors. MshA is important for light organ colonization (Llorca, 2008; Ariyakumar and Nishiguchi, 2009; Chavez-Dozal et al., 2014) but also for inducing cell-cell communication at low pH (Llorca, 2008). Perhaps proteins such as MshA from evolved V. fischeri strains have been impacted due to experimental evolution at low pH, thereby changing the colonization efficiency such that evolved strains outcompete ancestral strains during host competition studies. Future studies will examine msh loci and other symbiotic loci of both ancestral and evolved strains to determine whether subtle genetic changes are correlated to the evolution at low pH.