Dynamics of Thioalkalivibrio species in a co-culture under selective pressure of ampicillin

Abstract Haloalkaliphilic chemolithoautotrophic sulfur-oxidizing bacteria belonging to the genus Thioalkalivibrio are highly abundant in microbial communities found in soda lakes and dominant in full-scale bioreactors removing sulfide from industrial waste gases. Despite certain soda lakes being remote and unaffected by anthropogenic activities, haloalkaliphilic microorganisms, including Thioalkalivibrio strains, possess various antibiotic resistance genes. In this study, we investigated the impact of the antibiotic ampicillin on a co-culture of two Thioalkalivibrio species, Tv. thiocyanoxidans ARh2T and Tv. versutus AL2T, both experimentally and through in silico analysis of antibiotic resistance. Cell growth dynamics were monitored over time at increasing ampicillin concentrations using rep- and qPCR. Within ten days after the addition of ampicillin, the co-culture transitioned from a Tv. thiocyanoxidans ARh2T-dominated to a stable Tv. versutus AL2T-dominated culture. This shift was attributed to Tv. versutus AL2T displaying a lower susceptibility to ampicillin, making it more competitive. These results emphasize the potential implications of antibiotic pressure on microbial communities, where a resistant species can outcompete a stable co-culture. This study presents the first evidence of such dynamics in haloalkaliphilic chemolithoautotrophs. By understanding the antibiotic resistance and the competitive dynamics of haloalkaliphilic bacteria like Thioalkalivibrio, we can gain insights into their behaviour and stress response.


Introduction
Soda lakes are unique stable haloalkaline environments harbouring div erse micr obial comm unities dominated by pr okaryotes (Sorokin et al. 2014, Sorokin et al. 2015 ).These lakes can be found worldwide in areas with arid climates and with groundwater rich in sodium bicarbonate.High e v a por ation r ates tr ansform these lakes into highly sodium carbonate/bicarbonate-buffered, stable alkaline systems able to r eac h salt saturation (Jones et al. 1977, Grant et al. 1990, Jones et al. 1998 ).Despite these extr eme conditions, the micr obial comm unities ar e well ada pted and flourish in these environments (Jones et al. 1998, Ochsenreiter et al. 2002, Mesbah et al. 2007, Lanzén et al. 2013 ), in whic h the c hemolithoautotr ophic sulfur-oxidizing bacteria of the genus Thioalkalivibrio are abundantly present (Sorokin et al. 2013, Vav ourakis et al. 2016, Edw ar dson and Hollibaugh 2017, 2018, Zorz et al. 2019 ).In addition to soda lakes, Thioalkalivibrio strains are also important members of the microbial communities in the so-called "Thiopaq pr ocess", in whic h they sustainabl y r emove sulfide from industrial waste gas through oxidation to elemental sulfur under haloalkaliphilic conditions (Sorokin et al. 2008(Sorokin et al. , 2012 ) ).Recently, the antibiotic and metal "resistome" of the microbial communities of Lonar Lake, an anthropogenicpolluted alkaline lake in India, was described (Chakraborty et al. 2020a ,b ), for which Thioalkalivibrio has been signifi-cantly associated with antibiotic resistance genes (Chakraborty et al. 2020a ).
The increased occurrence of antibiotic resistant bacteria and their associated genes in waterbodies due to anthropogenic pressure is an alarming phenomenon (Chakraborty et al. 2020b, Zhang et al. 2009, Vaz-Mor eir a et al. 2014, Yang et al. 2017, Su et al. 2020 ).Ho w e v er, antibiotic r esistance as r esult of long-term evolution, existed long before anthropogenic pressure (D´Costa et al. 2011, Perry et al. 2016 ), protecting bacteria against their own antibiotic production and enabling them to outcompete other microbes in the ecosystem (Fajardo et al. 2009, Granato et al. 2019 ).Applying sub-inhibitory concentrations in humans, livestock or the environment poses a major threat to bacterial disease treatments with antibiotics, as it leads to an increase of antibiotic resistance amongst pathogens (Andersson and Hughes 2012 ).Alternativ el y, an antibiotic susceptible strain can be swept out of the population by an emer ging r esistant str ain, whic h initiall y composed onl y a small fraction of the population, if an antibiotic concentration between the minimal inhibitory concentration (MIC) of the susceptible strain and the MIC of the resistant strain is applied.This concept is described as the "mutant selection window" hypothesis (Negri et al. 1994, 2000, Drlica and Zhao 2007 ).
Even though certain soda lakes are situated in remote regions without anthropogenic pressure, various haloalkaliphilic microor-ganisms from those environments possess a diverse spectrum of different antibiotic resistances (Pikuta et al. 2003, Boldareva et al. 2008, 2009a,b , Wu et al. 2010, Zargar et al. 2010, Kompantse v a et al. 2012, Liu et al. 2012, Kiplimo et al. 2019 ).This has also been shown for strains of Thioalkalivibrio (Chakraborty et al. 2020a ).With the interest to explore the antibiotic resistance capacity within Thioalkalivibrio , we studied here the dynamics of two Thioalkalivibrio species growing as a co-culture at increasing concentrations of ampicillin over a period of four w eeks follo w ed b y four weeks without antibiotics .T he dynamics in the co-culture wer e monitor ed with r e petiti v e-element PCR (r ep-PCR), whic h is a molecular fingerprint technique able to distinguish between the two Thioalkalivibrio species, and with quantitative PCR (qPCR) targeting strain-specific genes.

Cultiv a tion and adaptation
Cultures of Tv. thiocyanoxidans ARh2 T with low numbers of Tv. versutus AL2 T wer e gr own in 10 ml batch cultures at 100 rpm and 30 • C with incr easing concentr ations of ampicillin for four weeks follo w ed b y four w eeks without the addition of ampicillin.During the ampicillin adaptation phase, the culture was changed at each transfer to fresh medium with a doubled concentration in ampicillin, but it was also tr ansferr ed a gain to fr esh medium with the same ampicillin concentration.This was in case the culture would not be able to grow at the increased ampicillin concentration and needed more time to adapt at the lo w er concentration.The cultures started at 0.1 μg/ml ampicillin and were able to grow until 12.8 μg/ml.No growth could be obtained with 25.6 μg/ml.The experimental tr ansfer sc hedule of the complete experiment can be found in Table S1 and is illustrated for the adaptation phase in Fig. S1 .For each new culture, a 1:200 dilution of the previous culture was applied during the serial transfers .T he medium was composed of 17.5 g/l Na 2 CO 3 , 13.9 g/l NaHCO 3 , 6.1 g/l NaCl, 1 g/l K 2 HPO 4 , 0.2 g/l MgCl 2 , 40 mM Na 2 S 2 O 3 , 5 mM NH 4 Cl and 1:1 000 trace metals (Pfennig and Lippert 1966 ), and the pH was adjusted to pH 9.8.Before inoculating the bacteria, a sterile solution of ampicillin (Roc he Dia gnostics GMBH, Mannheim, German y) was added to the medium.Cultures with added ampicillin were prepared in triplicates and one culture without ampicillin was used as a r efer ence .T he MIC of both species was tested in triplicate using the same culture medium and the same ampicillin concentrations as for the adaptation experiment.Growth of the bacteria w as follo w ed daily b y measuring the optical density in triplicates at 600 nm, and the av er a ge and the standard deviation of the three measur ements wer e used in the data anal ysis.

Repetiti v e-element PCR
Samples were taken in the exponential phase at the end of each adaptation step and were pelleted by centrifugation at 15 000 × g for 2 min.The supernatant was r emov ed and the pellets wer e stor ed at −20 • C until further pr ocessing.Genomic DN A w as extracted with the DNeasy Blood & Tissue Kit (Qiagen, Venlo, The Netherlands) following the supplier's instructions for Gramnegative bacteria.
Rep-PCR genomic fingerprinting was performed using the BOX A1R (5 -CTA CGGCAA GGCGA CGCTGA CG-3 ) primer (Versalovic et al. 1994 ).The PCR protocol included an initial denaturation step at 94 • C for 3 min, follo w ed b y 30 c ycles of tw o denaturation steps of 94 • C for 3 sec and 92 • C for 3 sec, an annealing step of 46.4 • C for 1 min, and an elongation step at 72 • C for 8 min, and a final elonga-tion step at 72 • C for 8 min (Foti et al. 2006 ).The reaction included in a total reaction volume of 25 μl, the Taq PCR Master Mix Kit 2x (Qiagen, Venlo, The Netherlands), 0.5 μM primer, and 90 ng of template DNA.Rep-PCR products were visualized on a 1.5% (w/v) a gar ose gel stained with 0.6 μg/ml ethidium br omide, whic h r an at 60 V for 8 h at 4 • C. The GeneRuler DNA Ladder Mix (Thermo Fisher Scientific, Waltham, MA, USA) was used as a marker.

Quantitati v e PCR
The qPCR was performed in an ABI 7500 Real-Time PCR system (Applied Biosystems, Waltham, MA, USA) with strain-specific primers on the same template DNA as were used for the rep-PCR.The primers were designed with the Primer3 pr ogr am ( https: //primer3.ut.ee ) and are listed with their target strain and genes in Table 1 .The specificity of the primers was assessed in silico by BLAST and confirmed by a PCR on genomic DNA of Tv. versutus AL2 T , Tv. thiocyanoxidans ARh2 T and a 1:1 mixture of both species.The PCR protocol included an initial denaturation step at 94 • C for 3 min, follo w ed b y 35 c ycles of a denaturation at 94 • C for 30 sec, an annealing at 60 • C for 30 sec and an extension at 72 • C for 15 sec, and a final extension at 72 • C for 10 min in 25 μl of reaction mix containing Taq PCR Master Mix Kit 2x (Qiagen, Venlo, The Netherlands), 0.5 μM primer, and 10 ng DNA.PCR products were revealed by gel electr ophor esis ( Figur e S2 ).The qPCR pr otocol included an initial denaturation at 95 • C for 10 min, follo w ed b y 40 cycles of a denaturation at 95 • C for 15 sec, an annealing at 60 • C for 30 sec and an extension at 72 • C for 30 sec.The reaction mix contained Maxima SYBR Green qPCR Master Mix (2x) (Thermo Fisher Scientific, Waltham, MA, USA), 0.3 μM of forw ar d and r e v erse primers, and 2 ng DNA in a total reaction volume of 25 μl.ROX solution (T hermo Fisher Scientific , Waltham, MA, USA) was used to correct for w ell-to-w ell v ariation and melting curv es wer e anal ysed for all measured samples to exclude non-specific PCR products.Each qPCR plate contained a reference sample (0 μg/ml-Ref) and a r efer ence gene (16S rRNA) to ov ercome plate effects.For eac h individual sample, three technical replicates were measured, and the av er a ge and the standard err or of the thr ee measur ements were used in the data analysis.
Individual amplification efficiencies (E) were calculated with LinRegPCR v.2018.0(Ramakers et al. 2003, Ruijter et al. 2009 ) and were found between 1.8 and 2.0 with the exception at very low target numbers .T he data was baseline corrected and threshold cycle (C t ) values were calculated using LinRegPCR.Relative gene quantification was calculated using the initial culture before adaptation (T 0 ) as r efer ence sample and the 16S rRNA as r efer ence gene according to the 2 − Ct method (Livak and Schmittgen 2001 ).This type of quantification compares the changes in the target gene´s copy number in a sample of interest to those in a r efer ence sample (Livak and Schmittgen 2001 ).Negative control samples, in which the DN A template w as replaced b y w ater, did not sho w amplification.

In silico analysis on ampicillin resistance genes
The genomes of Tv. versutus AL2 T (NZ_MVAR00000000) and Tv.thiocyanoxidans ARh2 T (NZ_ARQK00000000) wer e pr e viousl y sequenced and annotated.The gene sequences of both species were obtained from the NCBI RefSeq FTP serv er.Blastp anal yses wer e performed to detect the presence of beta-lactamase genes in the genome of Tv. versutus AL2 T and Tv.thiocyanoxidans ARh2 T .For this, the beta-lactamase protein gene sequences of TEM, SHV, OXA and AmpC fr om other Gamma pr oteobacteria wer e used as r efer ences (Unipr ot: Q6SJ61, Q0QXE5, Q5MD99, P0A9Z8, C7S9D0, Table 1.Primers used in the qPCR.P0A3M3, Q5EET9, D6NSM8).Mor eov er, the KEGG pathway map for beta-lactam resistance was analyzed for both strains and genes encoding for multidrug efflux pumps, RND family efflux transporters, AcrAB-TolC efflux pumps and MFS transporters were retrie v ed for analysis from the genomes of both species.

Rev ealing ampicillin-dri v en species shifts in the co-culture
The rep-PCR patterns of a culture treated with increasing ampicillin concentrations and of the r efer ence gr own in the absence of ampicillin w ere follo w ed in time (Fig. 1 ).The rep-PCR pattern of Tv. thioc yanoxidans ARh2 T w as pr esent for the initial cultur e and for the culture containing 0.1 μg/ml ampicillin.A mix of both species' r ep-PCR finger print was seen for the cultur e at 0.2 μg/ml and at 0.4 μg/ml.The r ep-PCR finger print then switc hed completely to the fingerprint of Tv. versutus AL2 T at ampicillin concentrations of 0.8 μg/ml and higher.Ho w e v er, the r efer ence cultur e incubated without ampicillin, did not experience a shift in pattern and maintained its Tv.thiocyanoxidans ARh2 T pattern.Characteristic bands with the BOX A1R primer were found for Tv.thiocyanoxidans ARh2 T at ∼470 bp, ∼520 bp, ∼850 bp, ∼950 bp, ∼1100 bp, ∼1850 bp and ∼2200 bp, whereas for Tv.versutus AL2 T they were found at ∼320 bp, ∼510 bp, ∼650 bp, ∼960 bp, ∼1300 bp, ∼1550 bp, and ∼2300 bp.
The r elativ e abundance of Tv. thiocyanoxidans ARh2 T and Tv.versutus AL2 T in the culture was monitored using qPCR amplification of strain-specific genes (Fig. 2 ).The gene encoding for thioc y anate: c ytoc hr ome c oxidor eductase (TcDH) was selected for Tv.thiocyanoxidans ARh2 T and the gene encoding the peptidemethionine (S)-S-oxide reductase (PMOR) for Tv.versutus AL2 T as they are single-copy genes and not present in the r espectiv e other species .T he 16S rRNA gene is also only present as a single-copy gene in both species and was used as a r efer ence for normalisation.The specificity of all three primers was e v aluated as good after testing them in silico as well as by PCR using the genomic DNA of both species and a 1:1 mixture of DNA of both species ( Figure S2 ).
The r efer ence cultur e, whic h was incubated without ampicillin, maintained its initial composition of Tv. thiocyanoxidans ARh2 T in the qPCR.Ho w e v er, cultur es with incr easing ampicillin concentration started to shift their composition from a Tv.thiocyanoxidans ARh2 T -dominated culture as seen with 0 and 0.1 μg/ml ampicillin to a mixture of Tv. thiocyanoxidans ARh2 T and Tv.versutus AL2 T when the ampicillin concentr ation incr eased to 0.2 and 0.4 μg/ml between day three and eight.The original species Tv. thiocyanoxidans ARh2 T was fully outcompeted by Tv. versutus AL2 T after an increase to 0.8 μg/ml ampicillin sampled at da y 10.T he initial composition with Tv. thiocyanoxidans ARh2 T did not r ecov er after ten successive incubations without ampicillin over a period of four weeks.

Sensitivity of pure cultures to ampicillin
To understand the shift in species composition after incubation with the increasing concentrations of ampicillin, the sensitivity of Tv. versutus AL2 T and Tv.thioc yanoxidans ARh2 T w as tested using the same concentrations as in the adaptation experiment ( Figure S3 ).Tv. versutus AL2 T was able to grow with 0.1 μg/ml after two days and with 0.2 and 0.4 μg/ml after three days lag period whereas Tv. thiocyanoxidans ARh2 T showed only limited growth with ampicillin with 0.1 μg/ml after only three da ys .From a concentration of 0.8 μg/ml and of 0.2 μg/ml onw ar ds, Tv. versutus AL2 T and Tv.thiocyanoxidans ARh2 T , r espectiv el y, wer e onl y able to grow after four days of incubation, which is likely due to degradation of ampicillin in the growth medium.T herefore , the respective MIC for ampicillin for Tv.versutus AL2 T and Tv.thiocyanoxidans ARh2 T are 0.8 μg/ml and 0.2 μg/ml.Chemical degradation of ampicillin in the culture medium was tested by inoculation of Thioalkalivibrio after one, two, three, or four days.Independently of when Thioalkalivibrio was added, visible growth only started after the fourth day when ampicillin was added to the medium (data not shown).In the absence of ampicillin, both species sho w ed gro wth within one da y.T he OD 600 peak at the start of each Thioalkalivibrio cultur e r esulted thr ough the pr esence of elemental sulfur pr oduced by the oxidation of thiosulfate.Its decrease after one day is linked to the depletion of sulfur due to complete oxidation to sulfate.

Genomic explor a tion of ampicillin resistance genes in Thioalkalivibrio
Analysis of the genome sequences of Tv. versutus AL2 T and Tv.thioc yanoxidans ARh2 T w as performed to r etrie v e their ampicillin and general antibiotic resistance capacity and to reveal differences.No beta-lactamase protein gene sequences of TEM, SHV, OXA and AmpC could be detected in either of the Thioalkalivibrio genomes using BLASTp against known protein sequences of other Gamma pr oteobacteria.T he KEGG pathwa y map for beta-lactam resistance sho w ed limited presence of genes in both species, but included penicillin-binding proteins and multidrug efflux pumps and Tv.versutus AL2 T only possesses one supplementary gene (B0684_RS10505), which could not be found in Tv. thiocyanoxidans ARh2 T ( Figure S4 , Figure S5 , Table S2 ).Both genomes were also screened for genes annotated as putative multidrug efflux pumps, putativ e RND famil y efflux tr ansporters, putativ e AcrAB-TolC efflux pumps and putative MFS transporters ( Table S2 ).Most genes could be r etrie v ed in both genomes, but four genes are unique to Tv. thiocyanoxidans ARh2 T , whereas eight to Tv. versutus AL2 T .For Tv. thiocyanoxidans ARh2 T , those genes are a multiple antibiotic resistance protein MarC (G372_RS0106695), multidrug resistance efflux pump (G372_RS0103685), multidrug efflux pump subunit AcrB (G372_RS0103690) and an outer membrane Figure 2. Dynamics of the co-culture followed over time by qPCR with TcDH, which is specific to Tv. thiocyanoxidans ARh2 T and PMOR, which is specific to Tv. versutus AL2 T .Adapted culture 1, 2, and 3 represent triplicates from the same starting culture and were k e pt se parate from each other throughout the experiment.The reference is a culture from the same starting culture as the adapted cultures, but without addition of ampicillin.The error bars depict the standard error of the a verage .

Discussion
Our study supports the concept of the "mutant selection window" hypothesis .T his concept states that a resistant subpopulation, whic h is initiall y onl y pr esent as a minor fraction in the population, can emerge and outcompete a susceptible subpopulation from the population, if an antibiotic concentration between the MIC of the susceptible and the MIC of the resistant subpopulation is applied (Negri et al. 1994, 2000, Drlica and Zhao 2007 ).To prev ent emer gence of r esistance fr om a pr e-existing subpopulation, a "m utant pr e v ention concentr ation" needs to be a pplied, whic h is a concentration higher than the MIC of the most resistant subpopulation (Drlica and Zhao 2007 ).This phenomenon is described based on two mutant subpopulations of a strain.Ho w ever, the same principle can also be applied on two strains of a species, or on two species of a genus with the latter being the case in our study.The "mutant selection windo w" hypothesis w as relativised by the study of Alexander and Maclean ( 2020 ).They predicted that at one-eighth of their MIC, individual resistant cells have only a less than 5% probability of detectable outgrowth.Moreover, they suggest that antibiotic concentrations below the MIC of the resistant strain might be efficient enough to eradicate the bacterial population when resistant mutants only make up a small fraction in the beginning (Alexander and Maclean 2020 ).Ho w e v er, it has been shown in our study as well as in multiple others that the selectiv e antibiotic pr essur e can impact the population composition when a concentration below the MIC of the sensitive strain was applied.This provides a window of opportunity for resistant strains to outcompete the susceptible ones (Negri et al. 2000, Gullberg et al. 2011, Andersson and Hughes 2012, Hughes and Andersson 2012, Day et al. 2015 ).
Screening the composition of the co-culture with rep-and qPCR during the incubation with increasing concentrations of ampicillin r e v ealed a switc h in species dominance of the co-culture.Notabl y, the mor e r esistant species Tv. versutus AL2 T , whic h in the beginning of the experiment was only a minor component of the co-culture and not detectable by qPCR, outcompeted the more susceptible species Tv. thiocyanoxidans ARh2 T .Within ten days and with an eight-fold increase of the ampicillin concentration, the resistant species became dominant in the co-culture and the sensitive species was no longer detectable.Growth profiles with different ampicillin concentrations explained this shift in dominance by a higher MIC for Tv.versutus AL2 T .Competitivity of Tv. thiocyanoxidans ARh2 T in the absence of ampicillin pr essur e as seen in the control culture could be explained on the other hand by the theory that the resistant strain has a higher fitness cost, for whic h r eason the sensitiv e str ain can outcompete the r esistant str ain (Gullber g et al. 2011 ).
The in silico analysis of both species´genomes did not detect any beta-lactamase genes and only revealed a limited number of genes associated with beta-lactam resistance.Ho w ever, multiple putativ e m ultidrug efflux pump, putativ e RND famil y efflux tr ansporters, putativ e AcrAB-TolC efflux pumps and putative MFS tr ansporters, whic h hav e been pr e viousl y associated to antibiotic resistance (Nikaido and Takatsuka 2009, Pérez et al. 2012, Nag and Mehra 2021 ), could be detected in both genomes with four genes specific to Tv. thiocyanoxidans ARh2 T and eight to Tv. versutus AL2 T .It is possible that amongst the specific genes in Tv. versutus AL2 T , a gene could be present allowing its increase resistance to ampicillin compared to Tv. thiocyanoxidans ARh2 T .Howe v er, their specific function and implication in ampicillin resistance still remains to be pr ov en experimentall y.Supplementary transcriptomic and proteomic analysis are needed to gain insights in the molecular mechanisms of ampicillin resistance in Thioalkalivibrio and to gain a better understanding of the increased ampicillin resistance of Tv. versutus AL2 T .
Antibiotic tr eatments ar e associated with the e volution and the spread of resistances within microorganisms (Blair et al. 2014 ), as well as with shifts within the microbial community composition.These dynamics include changes in abundance and extinction of species in a giv en comm unity (Coates et al. 2018, Cairns et al. 2020 ).Already a population composed by a single strain and subjected to bactericidal antibiotics sho w ed population dynamics, but no stochastic dynamic was observed with bacteriostatic antibiotics (Coates et al. 2018 ).In a recent study on the impact of antibiotic perturbations on a complex microbial consortium, Cairns et al. ( 2020 ) found a repeatable community response.In their experiment, moc k comm unities composed of 34 bacterial species wer e c hallenged with a pulsed antibiotic disturbance at differ ent concentrations, with or without species immigration, to study the ecological community response .T hey proposed that this response could be linked to the antibiotic susceptibility and the gr owth r ate of eac h giv en species.Next to species sorting, also antibiotic resistance evolution occurred in their experiment and its magnitude increased with increasing concentration of antibiotics.Ho w ever, the genomic evolution seemed not to be r ele v ant for the shortterm ecological dynamics in the studied community as the experiment sho w ed high repeatability amongst the replicates (Cairns et al. 2020 ).In our study, a gr owing cultur e could not be obtained with ampicillin concentration higher than 12.5 μg/ml, which suggests a lack of rescuing genetic mutations in the final Tv. versutus AL2 T -dominated co-culture.
Our r esults demonstr ate how a stable population, which is dominated by a susceptible species can be exchanged by a stable population composed by a resistant species due to antibiotic pr essur e.To our knowledge it is the first time that the application of the "mutant application window" hypothesis is shown for haloalkaliphilic c hemolithoautotr ophs.

Figure 1 .
Figure 1.Rep-PCR patterns showing the dynamics of a co-culture of Tv. thiocyanoxidans ARh2 T and Tv.versutus AL2 T incubated with increasing ampicillin concentrations over time .T he reference is a culture from the same starting culture as the adapted cultures, but without addition of ampicillin.NoAmp5 and NoAmp10 r epr esent the adapted cultures after five and ten transfers without ampicillin, respectively.M represents the DNA marker.