Aerococcus viridans, Staphylococcus haemolyticus and Corynebacterium bovis: sub-inhibitory exposure of lactic acid and cross-resistance to β-lactams antibiotics

TL;DR

Exposure to low levels of lactic acid can increase bacterial tolerance to the acid and cross-resistance to beta-lactam antibiotics in certain species.

Contribution

The study reveals that sub-inhibitory lactic acid exposure can induce cross-resistance to beta-lactam antibiotics in Aerococcus viridans and Staphylococcus haemolyticus.

Findings

Aerococcus viridans isolates showed a 2.3–7.5-fold increase in lactic acid tolerance during sub-inhibitory exposure.

Aerococcus viridans and two Staphylococcus haemolyticus isolates developed cross-resistance to beta-lactam antibiotics.

Corynebacterium bovis showed no changes in tolerance to lactic acid or antibiotics after repeated exposure.

Abstract

Lactic acid (LA) is a commonly applied post-milking teat disinfectant to prevent bovine mastitis. Subsequent to disinfection, microorganisms are often subjected to sub-inhibitory concentrations of biocide (residues), which can promote resistance to the applied biocide and cross-resistance to various antibiotics. In this study, control strains and 20 field isolates of three bacterial species, including Staphylococcus haemolyticus (n = 5), Aerococcus viridans (n = 9) and Corynebacterium bovis (n = 6) were exposed in-vitro to sub-inhibitory concentrations of LA. In the 30-day-study-period LA supplemented growth medium was changed every 48 h for Staphylococcus haemolyticus and Aerococcus viridans and every 72 h for Corynebacterium bovis. Changes of susceptibility towards LA during sub-inhibitory treatment were assessed by recording the first non-turbid concentration, named MICA. Changes of tolerance towards antibiotics were assessed before and after LA exposure by determining antibiotic MIC through microdilution. Results showed that all Aerococcus viridans isolates exerted an increase of LA MICAs during sub-inhibitory exposure, which was 2.3–7.5-fold. Staphylococcus haemolyticus reflected a 1.3–1.8-fold rise of LA MICAs. Aerococcus viridans showed a rise of oxacillin, cefoxitin, cefazolin and ceftiofur MICs after sub-inhibitory exposure to LA. Two isolates of Staphylococcus haemolyticus showed a rise of β-lactam-MICs. Corynebacterium bovis did not show any changes of tolerance towards LA or any tested antibiotics following 9 repetitions of sub-inhibitory treatment. The findings suggest that Aerococcus viridans and Staphylococcus haemolyticus have potential to develop tolerance towards LA and cross-resistance to β-lactams antibiotics following sub-inhibitory exposure to LA.

Figures

ATrend of LA MIC changes for *A. viridans* during sub-inhibitory treatment: isolates that did not reach a plateau.Fig 1A dummy alt textFig. 1BTrend of LA MIC changes for *A. viridans* during sub-inhibitory treatment: isolates that reached a plateau.Fig 1B dummy alt text

Trend of LA MIC changes for *S. haemolyticus* during sub-inhibitory treatment.Fig 2 dummy alt text

Oxacillin MIC for *A. viridans* before and after sub-inhibitory exposure to LA.Fig 3 dummy alt textFig. 4Cefazolin MIC for *A. viridans* before and after sub-inhibitory exposure to LA.Fig 4 dummy alt textFig. 5Cefoxitin MIC for *A. viridans* before and after sub-inhibitory exposure to LA.Fig 5 dummy alt text

Correlation of LA and ꞵ-lactam antibiotic* tolerance between pre- and post-exposure in *A. viridans.*Tolerance compared in measured MIC steps (+1). One MIC-step equals to a 2-fold change in antibiotic MICs and a 1.3-fold change in LA MICs. *Cefquinome, penicillin + novobiocin, and amoxicillin + clavulanic acid showed no changes at all.Fig 6 dummy alt text

Correlation of LA and ꞵ-lactam antibiotic* tolerance between pre- and post-exposure in *S. haemolyticus*.Tolerance compared in measured MIC steps (+1). ▼ = 1 step decrease from the initial MIC. One MIC-step equals to a 2-fold change in antibiotic MICs and a 1.3-fold change in LA MICs. *Cefquinome, penicillin + novobiocin, and amoxicillin + clavulanic acid showed no changes at all.Fig 7 dummy alt text

Full text

Introduction

1

Mastitis is one of the most prevalent health problems in dairy herds (Abebe, Hatiya, Abera, Megersa & Asmare, 2016; Krishnamoorthy, Goudar, Suresh & Roy, 2021). Based on the presence or absence of clinical symptoms, it can be categorized into clinical and sub-clinical mastitis (SCM) (Cobirka, Tancin & Slama, 2020). The global prevalence of SCM (1988–2019) per total of investigated cows ranges from 5 % to 90 % and meta-analysis showed a mean cow-prevalence of 46 %. (Çelik Gürbulak & Akçay, 2024). Coagulase-negative staphylococci [often also referred to as Non-aureus staphylococci (NAS)], Aerococcus (A) viridans, and Corynebacterium (C) spp. are reported to be important etiological agents of SCM in cows (Levison et al., 2016; Lücken, Woudstra, Wente, Zhang & Krömker, 2022; Sun et al., 2017). SCM causes significant economic losses by reducing milk production (Pfützner & Ózsvári, 2017) and reproductive performance of cows (Bouamra, Ziane, Akkou, Bentayeb & Titouche, 2024), necessitating management of SCM from both health and economic perspectives. Post-milking teat disinfection is regarded as one of the most efficacious strategies to prevent SCM (Wicaksono et al., 2019). For that purpose, a number of disinfectants are used as post-milking teat disinfectants, such as iodophors, quaternary ammonium compounds, chlorhexidine, chlorine dioxides, and lactic acid (LA) (Fitzpatrick, Garvey, Flynn, O'Brien & Gleeson, 2021). Among these, LA has been documented as the second most applied post-milking teat disinfectant in a survey conducted among 31,430 cows of 43 Hungarian dairy operations (Ózsvári & Ivanyos, 2022). It is also reported as a safe (Kovalenko et al., 2024), biobased, and biodegradable broad-spectrum biocide (Boomsma et al., 2015). To note, many biocides exhibit a significant reduction in efficacy upon interaction with organic materials, which likely occurs on the teat, or when the product becomes contaminated through environmental exposure (Coutinho, Medeiros, Silveira, Silva & Mota, 2012). Subsequent to the disinfection process, microorganisms are often subjected to residues of biocidal agents and thus exposed to sub-inhibitory concentrations (Tezel & Pavlostathis, 2015). Numerous investigations have indicated that exposure to sub-inhibitory concentrations of biocides promotes the development of resistance to these agents, and may additionally result in cross-resistance to various antibiotics (e.g.; β-lactams, aminoglycosides) (El Behiry, Schlenker, Szabo & Roesler, 2012; Mc Cay, Ocampo-Sosa & Fleming, 2010; Wieland et al., 2017). Adkin et al. (2022) found S. aureus increased tolerance towards oxacillin after sub-inhibitory exposure to chlorhexidine. We could not identify any study about cross-resistance selected by LA specifically in udder pathogens when searching the literature except Schwenker et al. (2022), who found an upward shift of CH-MICs due to use of LA as a post-milking teat disinfectant. For foodborne pathogens, Komora et al. (2017) found that multiple-antibiotic-resistant Listeria (L) monocytogenes were less susceptible to LA than antibiotic-susceptible strains from food and clinical origin and suggested further investigation to understand the possible correlation between antibiotic resistance and LA stress. Al-Nabulsi et al. (2015) found increased resistance in L. monocytogenes (collected from processed meat and dairy products) to multiple antibiotics including penicillin and ampicillin after having been stressed with 85 % LA for 30 min. Dawan et al. (2024) found increased tolerance of Salmonella Typhimurium and Staphylococcus (S.) aureus towards ciprofloxacin, gentamicin and tetracycline after having been adapted in sub-inhibitory sodium chloride (48 h at 37 °C) and subsequently exposed to an actually lethal dose of LA (24 h at 37 °C), which suggests a potential cross-resistance mechanism. By contrast, in the same study they also found that Salmonella Typhimurium and S. aureus decreased tolerance towards ciprofloxacin, gentamicin and tetracycline after having been adapted in sub-inhibitory LA and then exposed to a lethal concentration of LA (Dawan et al., 2024), which suggested potential co-lateral sensitivity. In a study meant to investigate an association between LA and benzalkonium chloride tolerance in Escherichia coli, Castro et al. (2023) found that genes related to biofilm formation significantly increased resistance to benzalkonium chloride, but reduced resistance to LA (Oguadinma, Mishra, Juneja & Dev Kumar 2022). Such findings denote an inconsistency about cross and co-resistance of LA and other antimicrobials, which is also reported by a relevant review of Liao et al. (2020). So, extending research on LA is deemed important in order to clarify the mechanisms behind these inconsistencies. In this study, we investigated the changes of tolerance and cross-resistance to standard mastitis antibiotics in reaction to in-vitro sub-inhibitory treatment with LA. For that purpose, we focused on A. viridans, which is being addressed as a nascent emerging mastitis pathogen (Liu et al., 2019; Sun et al., 2017), S. haemolyticus as one of the most frequently identified NAS with multi-drug-resistance potential (Parashar, 2012) and C. bovis, which had been found to be selected by LA in a previous study (Schwenker et al., 2022).

Methods

2

Selection of bacterial isolates

2.1

The test isolates were selected from previously collected milk samples by Schwenker et al. (2022). Isolated bacteria were identified by MALDI TOF MS. Then, their minimum inhibitory concentration (MIC) for LA was determined by broth macrodilution following the method of the German Veterinary Society (DVG, 2017). In total, 20 identified bacterial isolates were chosen [(A. viridans (n = 9), S. haemolyticus (n = 5) and C. bovis (n = 6)] for sub-inhibitory treatment with LA in triplicates. The isolates were chosen according to their determined initial MICs, covering the lower and upper range of initial MICs. Together with the field isolates, reference strains from of Deutsche Sammlung von Mikroorganismen und Zellkulturen (DSMZ, Braunschweig, Germany) were run in triplicates (DSMZ 20,340 for A. viridans, DSMZ 20,263 for S. haemolyticus and DSMZ 20,582 for C. bovis).

Sub-inhibitory exposure to lactic acid

2.2

Sub-inhibitory (0.75 and 0.56-fold of MIC) treatment and determination of susceptibility changes during adaption (MIC adaption procedure = MIC_A)_ for LA of the tested isolates were accomplished simultaneously by broth macrodilution following the method of the German Veterinary Society (DVG, 2017) in a glass flask instead of a tube. Deviating from the DVG-protocol, the inoculum was only adjusted for the first exposure, but not during all later steps of the adaption procedure in order to not lose adapted cells; here, 50 µl of highest concentrated turbid overnight culture was mixed with 50 µl of the first non-turbid concentration (MIC in the first step, MIC_A_ in all others) and diluted 1:200fold. The first non-turbid concentration arising from that procedure was named MIC_A_, to distinguish it from classical MICs as assessed by the DVG-protocol.

Preparation of LA solution: The dilution of LA was prepared in water of standardized hardness (WSH) as a percentage of the recommended concentration of LA (3.5 %, w/w; relative density of 1.06 g/mL ± 0.02; LactiFence, DeLaval, Tumba, Sweden), considering 3.5 % LA as 100 %. To prepare 1000 mL of WSH, 6 mL of solution A (19.84 g of magnesium chloride and 46.24 g of calcium chloride, dissolved in 1000 mL of deionized water and autoclaved) and 8 mL of solution B (35.02 g of sodium hydrogen carbonate, dissolved in 1000 mL of deionized water and sterile filtered through a 0.22-μm membrane) were mixed in 700 mL of deionized water and filled up to 1000 mL. The pH (7.0 ± 0.2) was adjusted with 1 mol/L sodium hydroxide. The dilution series was prepared depending on the determined MIC /MIC_A_ of LA for the test isolate, which included 1.33 x MIC (D1), MIC (D2), 0.75 x MIC (D3) and 0.56 x MIC (D4).

Preparation of inoculum: On the first day, the bacterial test suspension was prepared according to the Clinical and Laboratory Standards Institute (CLSI) (CLSI, 2015). For this purpose, fresh overnight cultures were used, which were inoculated at 37 °C for 18 ± 2 h (h) on CBA. For the cultivation of fastidious species of the genus C. bovis, 100 µl of Tween 80 was previously spread on the CBA. Using a sterile swab (ROTILABO, Carl Roth GmbH & Co. KG, Karlsruhe, Germany), some colony material was collected and approximately 1.5 × 10^8^ cfu/mL cell density was adjusted in 0.9 % sodium chloride. The bacterial cell number at different McFarland standard units (MFU) was initially counted by plate count for the DSMZ strain of C. bovis, and A. viridans to estimate the required optical density of bacterial solution in MFU for preparing a 1.5 × 10^8^ cfu/mL cell density solution. This ensured the required optical density (as measured in a commercial densitometer) as 2.3 MFU for C. bovis (cultured on CBA with Tween 80) and 2.5 MFU for A. viridans. For S. haemolyticus, the standard recommendation of 0.5 MFU was applied. So, for S. haemolyticus, C. bovis and A. viridans a solution of 0.5 MFU, 2.3 MFU and 2.5 MFU, respectively, was prepared in 0.9 % sodium chloride. The prepared solution was diluted 1:10 in tryptone sodium chloride to ensure final bacterial inoculum concentration of 1.5 × 10^7^ cfu/ml, which corresponds to a final concentration of 1.5 × 10^5^ after having added 100 µl to 10 mL broth.

Sub-inhibitory exposure: For this purpose, 5 mL of respective dilution of LA (D1, D2, D3 and D4) and 5 mL of a double-concentrated tryptic soy broth (2xTSB) (Carl Roth GmbH & Co. KG, Karlsruhe, Germany) were added to 20 mL glass flasks. Following Watts and Rossbach (2000), a supplement of 1 % Tween 80 (VWR International GmbH, Randor, Pennsylvania, United States) was added to the 2xTSB when testing the fastidious species C. bovis. Each dilution was triplicated for every test isolate. 100 µl of the prepared inoculum (50 µl + 50 µl, see below) was added to each test tube. A dilution series of teat disinfectants without inoculation of bacterial test suspension served as a negative control. In addition, a growth control of each isolate (without disinfectant) was included as a positive control. A reference DSMZ strain of the respective species was tested in parallel with each species. All tubes were kept in a shaking incubator at 37 °C and a shaking frequency of 100 RPM/minute.

Determination of MIC_A_: After 48 h (72 h in case of C. bovis), all tubes were sub-cultured on CBA and kept in an incubator at 37 °C without shaking (due to limited access, and to follow the DVG instructions). After 48 h, bacterial growth was recorded to determine the new MIC_A_ and grown dilutions were plated in order to check cultural purity and cfu count. The MIC_A_ value was determined by observing the lowest concentration of disinfectant that hinders the growth of bacteria (non-turbid flask, confirmed by plate counts lower than 10^4^ cfu/mL). The MIC_A_ was calculated considering the further dilution of the applied concentration of disinfectant with the volume of broth and bacterial solution and was converted to (wt/v) mg/L from percentage of recommended concentration of LA.

After 48 h (72 h in case of C. bovis), the procedure above was repeated to prepare the mixture of 5 mL 2xTSB and 5 mL LA dilution. The inoculum was applied from 100 µl (50 µl of D3+50 µl of D2) of the broth culture to all respective test tubes which ensure range of bacterial concentration between 10^4^ and 10^7^ cfu/mL. If the bacteria were found to have adapted to higher dilution, the dilution series was upgraded with the next higher concentration (1.33-fold of D1), in reaction to the new MIC_A_. The procedure was repeated for 30 days in a 48 h interval (15 cycles) for A. viridans and S. haemolyticus and a 72 h interval (10 cycles) for C. bovis. All microbiological work was performed in a laminar flow workbench.

Determination of antibiotic susceptibility

2.3

The susceptibility of test isolates towards standard mastitis antibiotics was tested before and at the endpoint of sub-inhibitory exposure to LA (day 30) by microdilution using Micronaut-S veterinary (Merlin, Bruker, Munich, Germany) antibiotics-coated plate, following the manufacturer's protocol. The Micronaut-S plates consist of a doubled layout (allowing to test 2 isolates per plate) coated with serial concentrations (µg/mL) of 14 different antibiotics (penicillin G, penicillin/novobiocin, ampicillin, cefazolin, cefoperazon, cefoxitin, cefquinome, ceftiofur, oxacillin, pirlimycin, erythromycin, amoxicillin/clavulanic acid, kanamycin/cephalexin and marbofloxacin), including two wells for growth control. To perform this test, a suspension was prepared equivalent to 0.5 MFU for S. haemolyticus and 2.5 MFU density for A. viridans in 2 mL of 0.9% sodium chloride from a pure culture in CBA with 5% sheep blood using sterile swabs (ROTILABO, Carl Roth GmbH & Co. KG). Then, in case of A. viridans and S. haemolyticus 100 µl of prepared suspension was transferred to 11.5 mL of cation-adjusted Mueller Hinton broth (Thermo Fisher Scientific, Dreieich, Germany) kept in a 20 mL glass test tube and homogenized well by vortexing. In the case of the fastidious bacterium C. bovis, 200 µl of 2.3 MFU suspension of test bacteria was transferred to 11.5 mL Micronaut-H medium (Sifin diagnostics GmbH, Berlin, Germany) and homogenized well by vortexing. The prepared broth medium with bacterial suspension was placed in a 2-channel reservoir and collected manually using the 8-channel pipette (Eppendorf Research, Hamburg, Germany). Then, 100μl of the suspension was pipetted into each well of the Micronaut-S plate. After filling, the panels were closed with unperforated cover film or covered with another panel to prevent evaporation during incubation. The plates were then incubated at 37°C for approximately 18-24 h. The results were determined photometrically (photometer Tecan infinite F50, Tecan Trading AG, Switzerland) and visually. The MIC values were determined visually by observing cloudiness or button formation (white dot at bottom). In case of doubts at visual examination, photometric values of ≥0.3 were interpreted as bacterial growth.

Statistics

2.4

Paired comparison (Wilcoxon matched-pairs signed rank test) of both LA and antibiotics MIC / MIC_A_ for the tested isolates was performed in order to compare MIC-values before and after the sub-inhibitory treatment. For that, we used the number of MIC-steps between the pre- and post-exposure value in a (+1) transformation. The nonparametric spearman correlation (r value) was determined between the number of these MIC-steps of LA (1.3-fold) and of every tested antimicrobial (2-fold) to reveal a possible correlation of LA- and antimicrobial MIC changes.

Results

3

The results of 20 field isolates (in 3 replicates) from three bacterial species, including A. viridans (n = 9), S. haemolyticus (n = 5) and C. bovis (n = 6) and 1 DSM reference strain per species were considered to analyze the effect of sub-inhibitory treatment with LA towards both LA and antibiotic tolerance in tested bacteria. All three replicates of different tested isolates showed similar pattern of MIC changes within the respective isolates without any evident difference (except 1 to 2 cycles earlier or later adaption). Thus, in order to ease the visual access, we did not include error bars in the figures.

Effect of sub-inhibitory exposure in test bacteria on lactic acid tolerance

3.1

Results elucidated an increase of LA tolerance among all A. viridans isolates after sub-inhibitory exposure. The change of tolerance ranged from a 2.3-fold to a 7.5-fold rise of the initial LA-MIC (Fig. 1A, 1B and Table 1) and the differences between before and after LA exposure were found statistically significant (P = 0.004) in a paired comparison, while comparing the determined MIC.Table 1. Summary of LA MIC changes for Aerococcus viridans and Staphylococcus haemolyticus.Table 1: dummy alt textID. of IsolateStarting MIC (mg/L)End MIC = Max MIC (mg/L)Day to first increaseDays to reach Max. MIC2723AV440.81028.58163592AV330.61836.68183701AV238.81028.56222915AV330.61377.510223919AV238.81377.58183865AV440.81377.510303743AV183.71377.58283472AV440.81377.510203120AV330.61377.51430DSMZ 20,340440.81377.510303958SH1377.52442.716223164SH1377.52442.712124137SH2442.72442.7No ChangeNo Change4447SH1836.62442.728283934SH1836.62442.71212DSMZ 20,2631377.51836.622

After sub-inhibitory exposure, the LA-MIC of 6 out of 9 isolates rose >4-fold. One of these isolates (3743AV) reached a 7.5-fold increase in MIC and for 2 isolates (3592AV and 3919AV), MICs increased nearly 6-fold. In total, 8 out 9 isolates lustered a >3-fold rise of tolerance towards LA compared to their initial MIC. Most of the isolates (8 out of 9) adapted to a 1.3-fold of the initial MIC within five cycles of exposure to sub-inhibitory concentration, among which 5 started getting adapted after four cycles and 1 after three cycles (Supplementary file). In addition, 4 out of 9 isolates (including the DSMZ reference strain) did not reach a plateau even after 14 repeated sub-inhibitory exposures (Fig. 1.A) and 5 out of 9 isolates reached a plateau, which sustained until the end of the trial, thus plateauing for 5 to 8 cycles of sub-inhibitory exposure (Fig. 1.B).Fig. 1ATrend of LA MIC changes for A. viridans during sub-inhibitory treatment: isolates that did not reach a plateau.Fig 1A dummy alt textFig. 1BTrend of LA MIC changes for A. viridans during sub-inhibitory treatment: isolates that reached a plateau.Fig 1B dummy alt text

In case of S. haemolyticus, MICs started at higher initial values than in A. viridans, and 4 out of 5 field isolates showed a further rise of tolerance towards LA ranging from 1.3 to 1.8-fold of the initial MIC (Fig. 2), which was statistically non-significant (P = 0.125). The change of tolerance was seen in five isolates including the reference strain, whereas 1 field isolate (4137SH) remained static at the initial MIC, which was already 1.8-fold higher than the lowest and 1.3-fold higher than the highest initial MIC of other S. haemolyticus isolates. The DSM reference strain reached a plateau already at the second cycle, and 2 field isolates reached a plateau after 5 cycles of sub-inhibitory exposure (Supplementary file).Fig. 2. Trend of LA MIC changes for S. haemolyticus during sub-inhibitory treatment.Fig 2 dummy alt text

Tolerance of C. bovis towards LA, which was already 4–10-fold higher compared to A. viridans and 1.3-fold higher than S. haemolyticus in the beginning, did not alter for any tested isolate following 9 cycles of sub-inhibitory treatment (data not shown).

Effect of sub-inhibitory exposure of lactic acid on antibiotic susceptibility

3.2

A. viridans reflected changes of MICs of oxacillin, cefoxitin, cefazolin, penicillin and ceftiofur tolerance after being treated with sub-inhibitory concentration of LA. No other changes were observed.

A 2- to 16-fold rise in oxacillin tolerance was observed in 6 out of 9 A. viridans isolates (Fig. 3), which was statistically significant (P = 0.031) in a paired comparison (Wilcoxon matched-pairs signed rank test) between before and after exposure to LA. Two isolates (3865AV and 3743AV) asserted a 2-fold rise of cefazolin-MIC while 1 isolate had a 2-fold decrease (Fig. 4). In S. haemolyticus, 2 isolates (3701AV and 3865AV) exerted a 2-fold increase of cefoxitin-MIC (Fig. 5). One isolate (3701AV) reached a 4-fold rise of penicillin and ceftiofur-MIC.Fig. 3. Oxacillin MIC for A. viridans before and after sub-inhibitory exposure to LA.Fig 3 dummy alt textFig. 4Cefazolin MIC for A. viridans before and after sub-inhibitory exposure to LA.Fig 4 dummy alt textFig. 5Cefoxitin MIC for A. viridans before and after sub-inhibitory exposure to LA.Fig 5 dummy alt text

While S. haemolyticus displayed a 2-fold rise of cefoperazon for 2 field isolates (4137SH and 3934SH) plus the DSMZ reference astrain, there was also a 1 step decrease for 1 isolate (3958SH). In addition, there was a 16-fold rise of penicillin and ampicillin-MICs for 1 isolate (4447SH) after adaption to LA (Figure 7). No other antibiotic MICs were changed. C. bovis did not show any alteration of sensitivity to the tested antibiotics.

Correlation of lactic acid tolerance and antibiotics tolerance

3.3

Focusing on a possible correlation between LA tolerance and antibiotic tolerance post-exposure, no significant correlation was found. We observed that 6 out of 9 A. viridans isolates exhibited a >4-fold increase of LA tolerance after 15 cycles of sub-inhibitory exposure among which 5 isolates showed increased tolerance of at least one and maximum three β-lactam antibiotics, while 1 isolate did not show any change in any antibiotic MIC (Fig. 6). DSMZ 20,340 showed a lower than 4-fold increase of LA tolerance, and no changes for any antibiotics (Fig. 6).Fig. 6. Correlation of LA and ꞵ-lactam antibiotic* tolerance between pre- and post-exposure in *A. viridans.*Tolerance compared in measured MIC steps (+1). One MIC-step equals to a 2-fold change in antibiotic MICs and a 1.3-fold change in LA MICs. *Cefquinome, penicillin + novobiocin, and amoxicillin + clavulanic acid showed no changes at all.Fig 6 dummy alt text

Also for S. haemolyticus, no significant correlation was found in the Spearman rank test. Two field isolates and DSMZ 20263 showed a 1.3-fold increase of LA tolerance and showed increased tolerance to at least one β-lactam antibiotic, ranging from 2 to 16-fold (Figure 7). One isolate did not show any change in LA tolerance but showed a 2-fold increase in cefoperazon tolerance (Figure 7). The remaining 2 isolates showed a 1.7-fold increase in LA tolerance but did not exhibit any changes in antibiotic tolerance (Figure 7). C. bovis did neither show changes of LA MIC nor of any antibiotic MIC, so no correlation was tested.Fig. 7. Correlation of LA and ꞵ-lactam antibiotic* tolerance between pre- and post-exposure in S. haemolyticus.Tolerance compared in measured MIC steps (+1). ▼ = 1 step decrease from the initial MIC. One MIC-step equals to a 2-fold change in antibiotic MICs and a 1.3-fold change in LA MICs. *Cefquinome, penicillin + novobiocin, and amoxicillin + clavulanic acid showed no changes at all.Fig 7 dummy alt text

Discussion

4

The results of the experimental study elucidate a significant increase of in-vitro tolerance of A. viridans and a non-significant increase for S. haemolyticus towards LA after adaption, while no adaption was seen for C. bovis. Focusing on other research, Rajkovic et al. (2009) found that the repetitive inactivation of L. monocytogenes with LA yielded a higher resistance in comparison to the parental culture. In the cited study, the adapted culture showed a 1.28-log10 reduction of the LA-adapted strain when treated with 2.5 % LA, compared to 2.15-log10 reduction for the parental strain, implying that prior exposure to LA can affect the tolerance for LA in bacteria. To note, both values are well below the reduction needed to consider a disinfection as successful (≥ 5log10) (Reichel, Schlicht, Ostermeyer & Kampf, 2014). In our study, we assessed MIC-values, not log10 reduction, though we can roughly relate both measures at the MIC, since we proved absence of growth by agar plating (100 µl), which means an approx. reduction of replicable bacteria ≥ 4.18 log10 (1.5 × 10^5 to <10 per mL). The DVG protocol for assessing log-reduction differs from the protocol for determining the MIC, since in the latter protocol, bacteria may just have been inhibited instead of killed (since 100 µl of the disinfectant at MIC are transferred to the agar plate). However, we did not aim to explore increasing reduction at sub-inhibitory MICs, but the increase of MICs, instead, which is easier to perform, and is more suitable to assess adaption to borderline-inhibitory concentrations. In that respect, it has to be noted that MIC-assessment during adaption (MIC_A_) differs from the initial MIC in terms of inoculum preparation. We do not assume that this deviation introduced major bias into our adaption protocol, since a mere inoculum effect would have got visible from the very beginning, while none of the isolates except one DSMZ (S. haemolyticus) showed any visible signs of adaption during the first cycles of sub-inhibitory exposure. A second modification between assessing MIC and MIC_A_ was the incubation period for A. viridans and S. haemolyticus, which was 48 h instead of 72 h (DVG-protocol). This was modified due to the fact that none of the isolates had shown any MIC-changes between 48 h and 72 h in the initial MIC-testing, and the shorter incubation period allowed to speed of the experimental procedure and to end up with more experimental cycles. Original times were kept for C. bovis, due to fastidious growth generally hindering susceptibility testing (Mitchell & Markantonis, 2025).

Some studies (Foster & Hall, 1990; Goodson & Rowbury, 1989) explained that an increase of LA tolerance can be due to the acid tolerance response (ATR) phenomenon of bacteria; this means that exposure of bacteria to mild acid can induce increased tolerance to strong acid challenges. The significant in-vitro increase of oxacillin in A. viridans and a non-significant increase of other β-lactam MICs after 15 cycles of sub-inhibitory exposure to LA in A. viridans and S. haemolyticus is indicative of a potential to develop cross-tolerance between LA and β-lactams antibiotics. Reports on such a cross-tolerance between LA and antibiotics in general are scarce and inconsistent (Liao et al., 2020). McMahon et al. (2007) found an association between growth in an acidic medium (acidified with hydrochloric acid) and increases in the MIC to amikacin, ceftriaxone and nalidixic acid (E. coli), gentamicin and erythromycin (S. aureus) and amikacin, ceftriaxone and trimethoprim (Salmonella Typhimurium). Anyhow, in that setting, the growth medium in the AMR-test was acidified, while in our experimental setting, no relevant concentration of LA should have been transferred to the microdilution plate, due to plate passage of the bacteria. Marzoli et al. (Marzoli et al., 2021) found that sub-inhibitory exposure to benzalkonium chloride increased tolerance towards the biocide and induced cross-resistance towards cefoxitin in NAS. LA, benzalkonium chloride and β-lactam antibiotics have a partially common mode of action in bacterial cell membrane disruption (Iyer, 2022; Kampf, 2024; Mani-López, García & López-Malo, 2012), where LA and benzalkonium chloride directly affect cell membrane permeability. By contrast β-lactams act as suicide substrates of the dd-transpeptidase catalytic domain of the penicillin-binding proteins (PBPs) and thereby prevent the last cross-linking step of cell wall assembly (Tipper & Strominger, 1965). Indirectly, this also interferes with cell membrane permeability by weakening the cell wall to make it more susceptible to osmotic stress. Thus, compensatory mechanisms (e.g. production of compatible solutes and uptake of osmoprotectants) stabilizing cell permeability may help bacteria to adapt to biocide stress and beta-lactams (Kempf & Bremer, 1998; Wood et al., 2001). Findings of Wu et al. (2012) confirmed that LA tolerance enhances higher expression of genes (murA and murG) involved in peptidoglycan synthesis in Lactobacillus casei which also might facilitate the survival of the bacteria in acid stress. We could not proof a statistical correlation between the extent of adaption to LA and to antibiotics in terms of x-fold changes of MICs. Anyhow, this does not mean that there is no association at all, since for that specific question, it would have been necessary to maintain a similar number of unexposed controls for 30 days. We could not afford to do so due to a lack of glass ware, incubators, and staff capacity. Initially, we did not expect that all strains will adapt, so we aimed to compare adapted strains with strains that failed to adapt, which, however, where not found for A. viridans, while there was one single unreactive strain of S. haemolyticus. To run unexposed controls would have also helped in order to assess whether the adaption was simply due to prolonged subculture, independent of LA-exposure, since bacteria may change phenotypic features during serial passaging (Somerville et al., 2002). Anyhow, we do not expect the simple number of passages to have introduced major systematic bias, since there were no changes in MICs except LA MICs in the broad majority of cases.

The experiment showed that bacterial response was both species- and strain-specific. All strains of A. viridans increased tolerance towards LA during continuous exposure to a sub-inhibitory concentration of LA, mainly beginning after 5 cycles of exposures. This is also plausible in light of Schwenker et al. (2022) who found a significant increase of LA-MICs for all tested bacteria within 5–6 days of an intervention where LA was applied to teats of lactating cows. Despite having run the study for 30 days (15×48 h cycles), there remains a potential of a further increase in tolerance towards LA, as almost half of the tested isolates did not show any plateau. The other A. viridans isolates showed a plateau within 8 to 10 repeated cycles of sub-inhibitory exposure with LA, but that is not necessarily at the endpoint of tolerance development, since 2 isolates showed a further increase of tolerance after having been at constant MICs for four (3865AV) and six (3592AV) cycles of sub-inhibitory exposures. In case of C. bovis, the incubation cycle in broth media was longer (72 h) as they have thick waxy cell walls like Mycobacterium which causes slower growth (Burkovski, 2013; Lewin & Sharbati-Tehrani, 2005; Watts & Rossbach, 2000). Focusing on the C. bovis response after sub-inhibitory exposure to LA, we found it static for LA tolerance as well as antibiotic tolerance. So, while all strains of A. viridans and all but one strain of S. haemolyticus adapted, none of the C. bovis strains did. At the same time, C. bovis started at higher initial MICs compared to A. viridans. This may, again, be due to the special type of cell membrane structure of C. bovis, having an additional arabinogalactan and mycolic acid layer like Mycobacterium, leading to a higher intrinsic tolerance compared to other Gram-positive bacteria (Benz, 2016; Burkovski, 2013). Other studies also reported about differences in intrinsic bacterial tolerance (Schwaiger et al., 2014), which are especially associated to cell wall structure features (Poole, 2002). Anyhow, though offering high intrinsic tolerance, the cell wall structure of C. bovis might also be responsible for the non-adaption to sub-inhibitory exposure of LA, as a high cell wall stability is usually associated with high generation times (Lewin & Sharbati-Tehrani, 2005), and a further adaption towards cell wall stability may be detrimental for bacterial growth. This may explain why not only the initial MIC, but even more the adaption behavior was different in C. bovis, compared to A. viridans. The rather strain-specific results for S. haemolyticus, which adapted to LA at lower initial MICs, while no further adaption was seen for the strain that started at 2500 mg/ already, may also point towards the fact that adaptability is not unlimited. Anyhow, this cannot be finally judged for S. haemolyticus, since the last adaption in a field isolate happened only two cycles before the experiment had to end after 30 days (as it was planned due to limited staff resources).

Practically, post-milking teat disinfectants are used for a longer period as a regular milk hygiene. In dairy environments, teat skin can be frequently contaminated with organic residues such as manure, milk proteins, and bedding material. These organic loads can lead to reduce concentrations of biocide at the skin surface causing sub-inhibitory exposure. The presence of organic matter (Fontes, Da Cunha, Fausto, Souza & Cerqueira, 2023), bacterial attachment abilities (LeChevallier, Cawthon & Lee, 1988) and traits that are involved in forming extracellular polymers (Brown & Gilbert, 1993) can affect the in-vivo tolerance of bacteria, which we could not estimate in this study. Thus, it is difficult to forecast long-term effects of sub-inhibitory exposure from an in vitro experiment for practical application, but the presence of organic matter and routine application (duration of contact) are rather expected to increase than to mitigate the effects. This is also illustrated by the fact that four out of nine A. viridans isolates did not reach a plateau / endpoint of adaption within fifteen cycles / 30 days.

Conclusions

5

The findings suggest that A. viridans and S. haemolyticus have the potential to develop tolerance towards LA as well as cross-resistance to β-lactams antibiotics, while there was no indication of potential co-lateral sensitivity (increased susceptibility to antibiotics). C. bovis has intrinsic tolerance towards LA, likely due to its special type of cell wall, which was stable throughout the experiment. The cross-resistance potential of LA and β-lactams antibiotics in A. viridans and maybe S. haemolyticus may be further investigated by identifying the genetic and functional mechanisms behind. To get a better phenotypic overview of tolerance changes, the presence of organic matter may be considered in further research.

Availability of data and materials

The data sets supporting the conclusions of this article are included within the article (and Supplementary file)

Funding

This project was financially supported by the H. W. Schaumann Foundation (Hamburg, Germany) and research fellowship of Livestock and Dairy Development Project (LDDP), Department of Livestock Services, 10.13039/501100015055Ministry of Fisheries and Livestock, Bangladesh.

Ethics statement

We obtained ethical consent from the dairy farm owner, which is Kiel university, and the farm is approved as a research farm according to § 11 of German Animal Welfare Act (Tierschutzgesetz). The owner of the farm gave informed ethical consent to the research, which consisted of a farm routine measure (teat disinfection).

CRediT authorship contribution statement

Md Shahinur Islam: Writing – review & editing, Writing – original draft, Visualization, Project administration, Methodology, Investigation, Formal analysis, Data curation, Conceptualization. Julia Anna Blumenberg: Writing – review & editing, Project administration, Methodology, Formal analysis, Data curation, Conceptualization. Ann-Kathrin Bremer: Data curation. Christina Susanne Hölzel: Writing – review & editing, Writing – original draft, Visualization, Validation, Supervision, Resources, Project administration, Methodology, Funding acquisition, Formal analysis, Data curation, Conceptualization.

Declaration of competing interest

The authors have not stated any conflicts of interest.