Introduction

The gastrointestinal tract is home to a dynamic collection of trillions of microorganisms that exerts significant impacts on health, including maintenance of immunity, metabolism, and intestinal barrier regulation1,2. Dysregulation of the gut microbial composition, such as disruption in the abundance and diversity of the microbiota compared to healthy hosts, is correlated with disease and infection and can be induced by diet or drug treatment3,4. Efforts to modify the gastrointestinal microbial community for treatment or disease prevention have long involved the use of probiotic organisms, including the lactic acid bacteria family of Lactobacillaceae5,6,7. Lactobacillaceae species can confer beneficial effects on health including modulation of the immune system, inhibition of pathogens by colonization resistance, and secretion of antimicrobial factors in the gut8,9. Among the many nutrients that gut microbes must compete for are essential metals, including calcium, iron, and zinc, which when limited or in excess, can drive changes to the gut microbiota composition that may be correlated with disease or infection10,11,12,13. Human studies show that Lactobacillaceae abundance declines in diarrhea-dominant, irritable bowel syndrome (IBS), type 1 diabetes, and multiple sclerosis (MS) patients14. Studies also show that Lactobacillaceae abundance increases with conditions such as Chrohn’s disease, but there are conflicting reports for abundance changes of Lactobacillaceae in humans with obesity and type 2 diabetes14. Lactobacillaceae are often among the organisms increased or decreased with changes in the bioavailability of metal nutrients, although the underlying molecular mechanisms are not yet well-defined11,12,13. Furthermore, the efficacy of probiotics like Lactobacillaceae has in some cases been correlated with the presence of different metals or can impact host uptake of metals, suggesting an important and probably species- or strain-specific metal-microbe-host interplay11,13,15.

Among several essential metals, calcium is an abundant inorganic ion that is widely used as a cellular signaling ion in eukaryotes for regulating cellular processes ranging from the cell cycle to gene expression and metabolism16,17. In prokaryotes, however, the requirements and roles of Ca2+ ions are less well defined18,19. There is, however, growing evidence that calcium is involved in a wide range of cellular activities in bacteria, including cell division and development, motility, stress response, and host-pathogen interactions. Several bacteria are capable of maintaining calcium homeostasis20,21,22,23, including low intracellular levels of the ion, and calcium binding proteins have been identified in bacteria, although most have not been functionally characterized24,25,26,27,28,29. For example, the first prokaryotic proteins found to contain canonical EF-hand motifs include calerythrin from Saccharopolyspora erythraea (which is presumed to function as a calcium buffer)27, calmodulin-like proteins proposed to be involved in calcium sensing and transport in Bacillus subtillis26, and CcbP from Anabaena sp. PCC 7120, which is a calcium-binding protein that may be involved in heterocyst differentiation28. Animal studies and clinical trials show that changes in dietary calcium levels influence the composition of the gut microbiota, suggesting variable calcium requirements and roles in the gastrointestinal ecosystem30,31,32,33,34,35,36,37,38,39. There is also growing evidence that calcium can promote the viability of intestinal Lactobacillaceae species30,31,32,33,36,37,40,41,42.

Bacteria inhabit a large variety of environments ranging from food and soil to host organisms. Correspondingly, the Ca2+ levels encountered by different bacteria vary significantly, contributing to our incomplete understanding of the roles of Ca2+ in prokaryotes. Furthermore, Ca2+ requirements for bacterial growth are not well-understood, and the commonly studied model organisms Escherichia coli and Bacillus subtilis can be grown in minimal medium lacking added Ca2+43. Trace amounts of Ca2+ may be present, however, to facilitate growth, which is supported by chemical chelator studies20,44. Among the lactic acid bacteria, early studies suggested limited Ca2+ requirements for some species (e.g. Lactobacillus arabinosus, which is now known as Lactiplantibacillus plantarum)45,46 and growth-promoting effects for others (e.g. Lacticaseibacillus casei)45,47,48,49. Recent work found that the Ca2+ requirements of L. plantarum were strain-specific and could impact intraspecies competitive fitness50. Bioinformatic approaches suggest the presence of Ca2+ uptake and efflux channels in bacteria, but few have been experimentally verified and none have been studied for Lactobacillaceae species18.

In some microbes, Ca2+ is a known regulatory signal and can have extracellular roles. Ca2+ ions can stimulate biofilm formation in Vibrio, Xylella fastidiosa, and Pseudomonas aeruginosa51,52,53,54, or play structural roles, including binding to and stabilizing the cell wall55,56. Vibrio vulnificus also harbors a matrix-associated Ca2+ binding protein that is required for biofilm maturation rather than initial attachment or adhesion57. In B. subtilis, the presence of Ca2+ enhances biofilm formation by forming CaCO3 and strengthening the extracellular matrix58,59. In the host environment, Ca2+ ions can induce a transition from a planktonic lifestyle to surface adhesion and biofilm community formation, improving the survival of invading pathogens19, but Ca2+ ions could play a similar role in the survival of commensal organisms and probiotic efficacy. Strong Ca2+ interactions with bacterial cell structures and surface proteins, including pili, fimbriae, teichoic acids, adhesins (including S-layer proteins), lipopolysaccharides, and exopolysaccharides (EPS) contribute to surface adhesion, cell-cell aggregation, and biofilm formation19,53,57,60,61,62,63. Few studies, however, have investigated the regulatory or structural roles of Ca2+ in commensal and probiotic Lactobacillaceae. Early work showed that the presence of Ca2+ during growth of Lactobacillus acidophilus ATCC 4356 facilitates a morphology transition from filamentous to bacilloid rods, and improves resistance to freezing, although the molecular mechanism was not investigated64. In another study, Ca2+ promoted the production of EPS in a strain of L. plantarum without affecting bacterial growth kinetics65.

Given the importance of Lactobacillaceae as health-promoting microbiota members and probiotics that are often influenced by Ca2+ ions, we aimed to systematically evaluate the effects of calcium on planktonic growth and biofilm formation in two distinct species, L. plantarum ATCC 14917 and L. acidophilus ATCC 4356. L. plantarum ATCC 14917 and L. acidophilus ATCC 4356 are two relatively well-studied beneficial species among several that are considered to permanently colonize the gastrointestinal tract66,67. The Ca2+ requirements of Lactobacillaceae have not been well-defined, but the studies described above suggest that they vary among different strains of L. plantarum and that Ca2+ is beneficial for the viability, possibly via cell surface stabilization, of L. acidophilus ATCC 4356. Systematic studies, however, have not been carried out and the intra- and extracellular roles of Ca2+ ions are not clear in these bacteria. Aside from these knowledge gaps, we chose to study the impacts of Ca2+ ions on biofilm formation in these two species because they have distinct cell surface-localized adhesion-promoting proteins. For example, L. acidophilus harbors S-layer proteins (SlpA) while L. plantarum has the mannose-binding protein (Msa) and both species have mucus-binding proteins, which are thought to promote host adhesion68,69,70. Our work reveals differential effects for calcium modulation of planktonic growth and biofilm formation in these species. Ca2+ induces the growth of both species and strongly promotes biofilm formation of L. acidophilus ATCC 4356 but not L. plantarum ATCC 14917 (Fig. 1). Inclusion of an extracellular chelator for Ca2+ diminishes biofilm enhancement but does not affect planktonic growth of L. acidophilus ATCC 4356. Microscopy analyses of cells grown in the presence and absence of Ca2+ and EGTA are consistent with the colorimetric biofilm quantitation results and further show a change in morphology from filamentous-like to bacilloid rods with addition of Ca2+ for L. acidophilus ATCC 4356. A bioinformatic analysis of potential Ca2+-binding surface proteins for each species reveals that S-layer proteins, which are specific to L. acidophilus, and a putative surface protein from L. acidophilus ATCC 4356 have or are likely to have Ca2+ binding sites that may facilitate cell-cell interactions involved in biofilm formation. Overall, this work reveals a new role for Ca2+-induced biofilm formation in some Lactobacillaceae species and identifies a possible mechanism of action for future studies.

Fig. 1
figure 1

Schematic illustration of calcium effect on Lactobacillus acidophilus.

Results and discussion

Calcium promotes the growth of L. acidophilus ATCC 4356 and L. plantarum ATCC 14917

Previously, we quantified metal uptake in L. acidophilus ATCC 4356 and L. plantarum ATCC 14917 grown in rich (MRS) medium and a metal-limited synthetic medium using inductively coupled plasma-mass spectrometry (ICP-MS)71. In that work, we focused on the effects of zinc, iron, and manganese on bacterial growth kinetics, but we also found that both species accumulated considerable levels of calcium (3.5–11 × 10− 10 nmol/CFU compared to 2.6–2.9 × 10− 10 nmol/CFU for E. coli)71. Given this result and other limited prior work examining the Ca2+ requirements and effects on viability of L. plantarum and L. acidophilus, we first aimed to evaluate the effects of Ca2+ on the growth kinetics of these species in various media.

We used three types of subculture growth media to investigate the effects of calcium on the growth kinetics of each species in polypropylene clear round-bottom microplates (Supplementary Scheme S1 and Methods section). Both rich (MRS) and chemically defined minimal (CDM) media were used as well as CDM medium supplemented with the mucin glycoprotein (mCDM). Mucin makes up much of the gastrointestinal mucus layer, binds metals including calcium, and is a surface to which bacteria can adhere72,73,74,75,76,77. CDM medium was prepared using Chelex to remove metals (see Methods). After growing starter cultures in MRS medium, each species was subcultured separately in MRS, CDM, and mCDM. After 20–24 h, all subcultures were washed with CDM and then incubated in CDM supplemented with a trace mineral formulation containing all essential metals except Ca2+. Added CaCl2 was then varied from 0 to 25 mM and the Ca2+-dependent growth curves for each subculture were collected. This range of CaCl2 concentrations was chosen based on reported intestinal Ca2+ concentration ranges78,79. The growth of both L. acidophilus ATCC 4356 and L. plantarum ATCC 14917 MRS and CDM subcultures was promoted by added Ca2+ (Fig. 2), although the growth promoting effect was stronger at lower Ca2+ concentrations (1 and 5 mM Ca2+) for L. acidophilus ATCC 4356 compared to L. plantarum ATCC 14917. CaCl2 decreases the lag time and increases growth rate as well as max OD of all subcultures (Fig. 2d-f and j-l). Mucin subcultures for each species exhibited the least impact of added Ca2+ on growth. For L. acidophilus ATCC 4356 mucin subcultures, all added Ca2+ conditions resulted in growth curves similar to those for added Ca2+ in MRS and CDM subcultures. For L. plantarum ATCC 14917, calcium induces the growth of MRS subcultures and to some extent the CDM subcultures, but not the mucin subcultures. It appears that the commercial mucin used in this work is improving the viability of the bacteria when subsequently subjected to calcium starvation. Mucin is known to bind metal ions and there are structures now reported showing specific binding sites for Ca2+ and Cu+/2+ in different locations on the intestinal mucin glycoprotein, MUC276,77. To gain more insight to whether mucin is acting as a source of Ca2+ and/or other nutrients, we used ICP-MS to measure the Ca2+ concentration in mCDM for comparison to CDM and MRS media (Supplementary Fig. S1). Based on the amount of calcium found in the mucin analyzed using ICP-MS (29 mmol Ca/g mucin), we calculated that for 0.1% mucin in CDM medium, the calcium concentration from the added mucin is ~ 29 mM. Mucin could, therefore, be acting as a significant source of calcium in mCDM medium. In subcultures grown in mCDM, cells may be able to accumulate enough Ca2+ to minimize the impact of subsequent varied Ca2+ on cell growth in CDM.

Fig. 2
figure 2

Calcium-dependent growth kinetics of L. acidophilus ATCC 4356 (a–f) and L. plantarum ATCC 14917 (g–l) subcultures in CDM supplemented with 1% Wolfe trace mineral solution. MRS (a,g), CDM (b,h), and mCDM (c,i) subculture growth curves as monitored by OD600. All error bars represent the standard deviation from three biological replicates with ≥ 2 technical replicates. Corresponding growth parameters for each condition: lag time (d,j), growth rate (e,k), and max OD (f,l) represent the mean ± SEM of three biological replicates with ≥ 2 technical replicates. All growth parameters were calculated using nonlinear regression curve fitting to a 4-parameter Logistic equation with GraphPad Prism 9 software. Significance was determined by one-way ANOVA with Tukey multiple comparison test: ns (not significant), *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001, ****p ≤ 0.0001.

Calcium increases biofilm formation of L. acidophilus ATCC 4356 but not L. plantarum ATCC 14917

Next, we aimed to assess whether Ca2+ could affect biofilm formation of L. acidophilus ATCC 4356 and L. plantarum ATCC 14917 using a static assay for biofilms formed on microplates with polystyrene flat bottom wells. As for the growth kinetics experiments, we assessed biofilm formation of each species under varied calcium concentrations (0–25 mM CaCl2, Scheme S1). Subcultures of each species in MRS, CDM, and mCDM were prepared as described above and in the Methods section, then incubated in CDM supplemented with trace minerals (not including Ca2+) and grown in the presence of varied CaCl2 without shaking. After washing, total biofilm was quantified by colorimetric crystal violet (CV) staining at 24 and 48 h of incubation. Here, increasing Ca2+ induces L. acidophilus ATCC 4356 biofilm formation to a small extent at 24 h and to a large extent at 48 h (Fig. 3a-b). The most significant effects occurred from 5 to 25 mM added Ca2+ in most subcultures. Conversely, only low added Ca2+ (5–10 mM) promoted biofilm formation of L. plantarum ATCC 14917 in all subcultures at 24 h. At 48 h, however, low added Ca2+ only promoted biofilms in MRS and mCDM subcultures, but not the CDM subculture (Fig. 3c-d). In addition, biofilm formation at 48 h in all L. plantarum ATCC 14917 subcultures was lowest at 25 mM Ca2+, whereas the total biofilm formation of L. acidophilus ATCC 4356 was highest at this condition. Both species grew well at this condition in planktonic form (Fig. 2), suggesting that there could be different extracellular roles for calcium in modulating biofilms of L. acidophilus ATCC 4356 and L. plantarum ATCC 14917. To examine the extent to which biofilms could interfere with the growth kinetics measurements presented in Fig. 2 (measured by OD600), we quantified total biofilms formed for MRS subcultures in the same microplates grown under the same conditions as the growth kinetics experiments (polypropylene clear round-bottom wells and incubating with shaking). Supplementary Fig. S2 shows similar trends of biofilms increasing with increasing added Ca2+ in the round-bottom plates as for the static assay in flat bottom plates (Fig. 3).

Fig. 3
figure 3

Effect of calcium on total biofilm formation for L. acidophilus ATCC 4356 (a-b) and L. plantarum ATCC 14917 (c-d). Biofilms formed by MRS, CDM, and mCDM subcultures of each species grown in CDM supplemented with Wolfe trace minerals solution were quantified by crystal violet (CV) staining as measured by UV-visible absorbance at 595 nm. All error bars represent the standard deviation of three biological replicates, each with ≥ 2 technical replicates. Significance was determined by one-way ANOVA with Tukey multiple comparison test: ns (not significant), *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001, ****p ≤ 0.0001.

Next, to gain a more comprehensive understanding of the dynamics between planktonic growth and biofilm formation of these species, we performed viability assays for planktonic cells and biofilms (Methods, Supplementary Fig. S3). Briefly, at 24 and 48 h, planktonic cells were aseptically removed from each well of microplates for serial dilution to determine numbers of viable cells [expressed as colony forming units (CFU)/well]. After planktonic cells were removed, biofilms in each well were aseptically washed and resuspended in phosphate buffered saline (PBS), followed by serial dilution to determine viability (expressed as CFU/well). After 24 h, viable planktonic cell counts for L. acidophilus ATCC 4356 (Supplementary Fig. S3a) were slightly higher with 10 and 25 mM added Ca2+ compared to 0 mM added Ca2+. By comparison, the growth curve for L. acidophilus ATCC 4356 as measured by OD600 exhibited a similar, albeit more pronounced, trend (Fig. 2a). After 48 h and when the growth curves for all conditions represent the stationary phase (Fig. 2a), viable planktonic cell counts of L. acidophilus ATCC 4356 grown with 0 mM added Ca2+ reached the highest level measured and the viability of planktonic cells grown with 10 and 25 mM Ca2+ declined (Supplementary Fig. S3b). Viable planktonic cell counts of L. plantarum ATCC 14917 after 24 h with 0, 10, and 25 mM Ca2+ were at similar levels (Supplementary Fig. S3e), although the corresponding growth kinetics experiments (Fig. 2g) showed higher maxOD for 10 and 25 mM added Ca2+ compared to lower concentrations. After 48 h, viable counts for L. plantarum ATCC 14917 at each condition were higher than those at 24 h, and the cell viabilities at 0 mM Ca2+ were highest among the different Ca2+ conditions. It is possible that the highest viable planktonic cell counts were measured at 0 mM Ca2+ after 48 h because both species have the slowest growth rates with no added Ca2+ compared to all other added Ca2+ conditions (Fig. 2e and k). Therefore, more cells should be viable compared to conditions under which cells are further into the stationary phase and undergoing increased cell death. Based on these results, the higher biomass measured by OD600 for both species at 10 and 25 mM added Ca2+ after 24 h of growth (Fig. 2) could be due to the accumulation of total planktonic cells (viable and nonviable) and biofilm formation at the bottom of the wells (Supplementary Fig. S2).

In parallel with the above viability assays for planktonic cells in each culture, we assessed the viabilities of biofilms present in the same cultures. Here, similar viabilities were measured between 0 and 10 mM added Ca2+ for L. acidophilus ATCC 4356 biofilms at 24 h, and biofilm cells with 25 mM added Ca2+ had slightly higher viabilities. After 48 h, the viability of biofilms formed with 25 mM added Ca2+ declined significantly and the viability of biofilms formed with 0 and 10 mM added Ca2+ slightly declined compared to biofilms at 24 h. The viabilities of L. plantarum ATCC 14917 biofilms were similar for 0 and 10 mM added Ca2+ (as for L. acidophilus ATCC 4356), but the viability was slightly lower with 25 mM added Ca2+ at 24 h. After 48 h, the viabilities of L. plantarum ATCC 14917 biofilms all declined compared to 24 h biofilms. The viability trends of 24 h biofilm cells for both bacteria (Supplementary Fig. S3c and g) resemble the trends observed for total biofilms at 24 h as quantified by CV (Fig. 3a and c). The decreased viabilities of 48 h biofilm cells, however, for L. acidophilus ATCC 4356 with 25 mM added Ca2+ and for L. plantarum ATCC 4356 at all Ca2+ concentrations suggests that the number of viable cells in these biofilms are less than the total biofilm formed. These observations warrant further investigation into the relationship between quantity and viability of biofilms formed with varied calcium, particularly at timescales beyond 24 h. Previously, a poor correlation was reported between the CV assay and viable cell counts for biofilms of L. plantarum WCFS1 (a strain similar to L. plantarum ATCC 14917 used in our work) and L. plantarum food isolates (FBR 1–6) formed in meat based medium with supplements at 24, 48, and 72 h80. Here, most biofilms had the highest viable cell counts within 24 h and remained constant up to 48 h, but decreased after 72 h. The authors hypothesized that higher amounts of biofilms were quantified with the CV stain compared to viable cell counts because CV could bind to dead cells within the biofilms and to biofilm matrix-associated eDNA, proteins, and EPS. Overall, our findings suggest that calcium may impact biofilm formation of L. acidophilus ATCC 4356 at 25 mM Ca2+ via an extracellular mechanism, although the specific roles of viable versus nonviable cells in the biofilms, such as in their contributions to biofilm structure and contents over an extended period of time, will warrant future studies involving both static assays and flow conditions. Static biofilms and biofilms grown under flow conditions can also vary in viability, as previously found for L. plantarum WCFS1 and several L. plantarum food isolates (BR1-6), which formed biofilms that under flow conditions had higher resistance to disinfectants, in terms of viability, compared to static biofilms81.

To determine whether the biofilm enhancement measured for L. acidophilus ATCC 4356 was exclusive to calcium, we investigated the effect of Mg2+ (0–25 mM) on the growth and biofilm formation of both Lactobacillaceae species. The luminal concentration of Mg2+ in the small intestine is estimated to be in the range of 1–5 mM depending on host diet and conditions82. Using MRS, CDM, and mCDM subcultures of each species, biofilm assays were carried out similarly to those for Ca2+ (as detailed in the Methods section) and total biofilm was quantified at 48 h of incubation (Supplementary Fig. S4). Here, increasing Mg2+ shows no effect on biofilm formation of L. acidophilus ATCC 4356. For L. plantarum ATCC 14917, biofilm formation was decreased with increasing Mg2+ up to ~ 5 mM and then remained constant at higher Mg2+ concentrations (10 and 25 mM). Magnesium supplementation up to 1 mM modestly enhanced the growth of MRS and CDM subcultures and significantly enhanced the growth of mCDM subcultures for both species (Supplementary Fig. S5-6). These results support a clear Ca2+-specific role in biofilm enhancement for L. acidophilus ATCC 4356.

Presence of an extracellular chelator induces distinct effects on the growth and biofilm formation of L. acidophilus ATCC 4356 and L. plantarum ATCC 14917 supplemented with calcium

Several studies in different bacteria have demonstrated that calcium can bind to cell surface macromolecules, including adhesin proteins, and facilitate aggregation and biofilm formation19,53,57,60,61,62,63. In consideration of this possibility, we examined the effects of the extracellular chelator, ethylene glycol-bis(β-aminoethyl)-N,N,Nʹ,Nʹ-tetraacetic acid (EGTA), on biofilm formation in the presence and absence of Ca2+ for both Lactobacillaceae species. In the gastrointestinal tract, calbindins, calmodulin, and other calcium binding proteins as well as peptides, amino acids, and other small molecules can bind to Ca2+ and play important roles in intestinal calcium absorption83,84,85. Here, we used EGTA, a well-established synthetic calcium chelator that does not readily cross cellular membranes and binds to calcium with higher affinity (apparent dissociation constant, Kd = 67 nM at pH 7.4) than magnesium and other divalent metals86. EGTA can control or buffer the labile (free or unbound) Ca2+ concentration in growth medium with added Ca2+. By adding EGTA to the bacterial growth medium, we could evaluate the ability of bacteria to compete with the chelator for Ca2+ ions, which could then be used for growth and/or biofilm formation. Specifically, we included EGTA for each condition at concentrations that would reduce the available Ca2+ ion concentration to < 0.3 µM (50 mM EGTA added to 0 mM Ca2+; 80 mM EGTA added to 10 mM Ca2+; 100 mM EGTA added to 25 mM Ca2+). Labile Ca2+ concentrations in the presence of EGTA were calculated using MaxChelator to ensure that the added EGTA would be sufficient for Ca2+ chelation under each condition (Table S1)87. Growth kinetics and biofilm formation were monitored in MRS subcultures of each species (Fig. 4). In the presence of chelator, even at high concentrations of calcium (10 and 25 mM added Ca2+), calcium no longer induces biofilm formation of L. acidophilus ATCC 4356 (Fig. 4a). For planktonic growth, however, in the presence of excess chelator and at 10 and 25 mM Ca2+, L. acidophilus ATCC 4356 was able to recover its growth kinetics. These findings suggest that L. acidophilus ATCC 4356 may be equipped with efficient calcium uptake transporters that can compete with EGTA for Ca2+ to support planktonic cell growth under these conditions, but not biofilm formation. These results also suggest an extracellular role for Ca2+ in L. acidophilus ATCC 4356 in increasing biofilm formation, potentially by facilitating cell adhesion through binding to bacterial surface proteins. In the presence of chelator and no added Ca2+, growth was severely restricted, although the cells eventually grew after a long lag time.

Intriguingly, EGTA had less impact on L. plantarum ATCC 14917 than L. acidophilus ATCC 4356, although addition of EGTA to 10 and 25 mM Ca2+ resulted in a modest increase in biofilm mass for L. plantarum ATCC 14917. Under the calcium and metal starvation condition (0 mM added Ca2+ with 50 mM EGTA), L. plantarum ATCC 14917 growth was able to recover after a moderately delayed lag time compared to L. acidophilus ATCC 4356 growth, which was severely delayed at this condition. Previously, we showed that L. plantarum ATCC 14917 and L. acidophilus ATCC 4356 took up similar amounts of Ca2+ when grown in MRS and L. plantarum ATCC 14917 acquired only slightly less Ca2+ than L. acidophilus ATCC 4356 when grown in metal-limited CDM71. These results suggest that L. plantarum ATCC 14917 has lower Ca2+ requirements in Ca2+-limited medium and/or can more efficiently take up calcium needed for growth than L. acidophilus ATCC 4356. On the other hand, the presence of EGTA did not impact the biofilm formation of L. plantarum ATCC 14917 (Fig. 4b) compared to L. acidophilus ATCC 4356.

Fig. 4
figure 4

Effect of chelator on Ca2+-dependent total biofilm formation and growth kinetics of L. acidophilus ATCC 4356 (a,c) and L. plantarum ATCC 14917 (b,d). Total biofilm levels (a,b) were determined by crystal violet staining (CV assay) as measured by A595 after 48 h. Calcium only effects from (a) and (b) were presented in Fig. 3b and d. Growth kinetics (c,d) were measured by monitoring the OD600. Measured data are shown with error bars as SD of three biological replicates, each with ≥ 2 technical replicates. Growth parameters for each condition: lag time (e), growth rate (f), and maxOD (g) represent the mean ± SEM of three biological replicates with ≥ 2 technical replicates. The growth parameters were calculated using nonlinear regression curve fitting to a 4-parameter Logistic equation with GraphPad Prism 9 software. Significance was determined by one-way ANOVA with Tukey multiple comparison test: ns (not significant), *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001, ****p ≤ 0.0001.

We performed control experiments to observe the biofilm formation and growth kinetics of these two species when grown in metal-limited CDM with no added calcium and with 50, 80, or 100 mM EGTA (Supplementary Fig. S7). The results revealed that EGTA at these high concentrations suppressed biofilm formation of both species when there was no added Ca2+ in the media (Supplementary Fig. S7a-b). Additionally, L. acidophilus ATCC 4356 could still grow in the presence of 50 mM EGTA (with no added Ca2+), but growth was completely suppressed in the presence of 80 and 100 mM EGTA (Supplementary Fig. S7c). Although EGTA was also detrimental to the growth of L. plantarum ATCC 14917, the growth was not entirely suppressed (Supplementary Fig. S7d-f). Given that EGTA can bind several essential divalent metals88, the suppression effects of excess EGTA without added Ca2+ on growth and biofilm formation of both species could be consistent with a reduction in the availability of other metal nutrients.

To better understand the effect of EGTA on cell viability, we conducted viability assays for planktonic and biofilm cells as described earlier and in the Methods for samples in the absence of EGTA (Supplementary Fig. S3). At 24 h, the viabilities of L. acidophilus ATCC 4356 planktonic cells grown in the presence of EGTA (50 mM EGTA added to 0 mM Ca2+; 80 mM EGTA added to 10 mM Ca2+; 100 mM EGTA added to 25 mM Ca2+) were lower than the cells grown with 0, 10, and 25 mM Ca2+, and there were no viable planktonic cells at 0 mM Ca2+ with 50 mM added EGTA (Supplementary Fig. S3a). After 48 h, the viable planktonic cell counts of L. acidophilus ATCC 4356 for these conditions increased compared to those at 24 h (Supplementary Fig. S3b). These results agree with the growth kinetics measurements using OD600 (Fig. 4c). For L. plantarum ATCC 14917, the viabilities of planktonic cells grown in the presence of 0, 10, and 25 mM added Ca2+ with 50, 80, and 100 mM EGTA, respectively, were slightly lower than the viable cell counts at these conditions without added EGTA (Supplementary Fig. S3e). After 48 h, the viabilities of planktonic L. plantarum ATCC 14917 cells increased for the condition with EGTA added to 0 mM Ca2+ and remained similar for 10 and 25 mM added Ca2+ with EGTA conditions (Supplementary Fig. S3f). These L. plantarum ATCC 14917 planktonic cell viabilities (Supplementary Fig. S3e-f) are also consistent with the corresponding growth kinetics measurements made using OD600 (Fig. 4d).

Viable biofilm cell counts, on the other hand, for L. acidophilus ATCC 4356 at 24 and 48 h were reduced for all conditions with added EGTA compared to biofilm cell counts for varied Ca2+ without EGTA (Supplementary Fig. S3c-d). These results are consistent with total biofilm formation as quantified by CV (Fig. 4a). The viabilities of L. plantarum ATCC 14917 biofilm cells, however, at 24 h with 10 and 25 mM added Ca2+ and EGTA were higher than with 10 and 25 mM added Ca2+ and no EGTA, although these viabilities slightly dropped at 48 h (Supplementary Fig. S3g-h). Overall, the effects of the chelator on biofilms are consistent between the viability and CV quantitation assays. These results suggest that EGTA is more detrimental to the viability of L. acidophilus ATCC 4356 planktonic and biofilm cells than L. plantarum ATCC 14917 cells, possibly due to the reduction of labile Ca2+ available for interacting with the L. acidophilus ATCC 4356 cell surface, which may be required for improving cell-cell adhesion for biofilm formation.

Microscopy analysis reveals a calcium effect on the aggregation of L. acidophilus ATCC 4356 but not L. plantarum ATCC 14917 biofilms

Given the differential effects of Ca2+ and EGTA on biofilm formation of Lactobacillaceae, we used bright-field microscopy to examine the effects on cell morphology and cell density or extent of cell-cell aggregation. Here, we used MRS subcultures of each species to grow biofilms on treated microscope slides in CDM medium with no added (0 mM) Ca2+, 10 mM Ca2+, 25 mM Ca2+, 0 mM Ca2+ + 50 mM EGTA, 10 mM Ca2+ + 80 mM EGTA, and 25 mM Ca2+ with 100 mM EGTA. Images of L. acidophilus ATCC 4356 show that cell-cell aggregation increases with increasing Ca2+ concentration (Fig. 5a-c). The extent of cell-cell aggregation remains similar between the no added Ca2+ and no added Ca2+ with chelator conditions (Fig. 5a and d) but decreases noticeably for each of the added Ca2+ and chelator conditions (Fig. 5b-c and e-f). Corresponding zoomed out images and three biological replicate microscope images are shown in Supplementary Fig. S8 and S9 respectively. These results are consistent with those obtained in the colorimetric biofilm assay (Fig. 4a). We also noticed that, as had been previously reported64, the morphology of L. acidophilus ATCC 4356 cells transitions from filamentous-like without added Ca2+ (Supplementary Fig. S8a and S9a) to bacilloid rods with 25 mM Ca2+ (Supplementary Fig. S8c and Fig. S9c). With addition of EGTA to 10 or 25 mM added Ca2+ conditions (Supplementary Fig. S8e-f and Fig. S9e-f), we note the return of a filamentous-like rod morphology although with addition of EGTA in the presence of no added Ca2+, the cells are mostly present as bacilloid rods (Supplementary Fig. S8d). L. plantarum ATCC 14917 cell density was low for 0 and 25 mM Ca2+, but increased at 10 mM Ca2+, like the trend observed in the colorimetric biofilm assay (Figs. 4b and 6a-c). The inclusion of EGTA chelator either does not clearly change the apparent cell-cell aggregation (0 and 10 mM Ca2+, Fig. 6d-e) or leads to an increase in cell density (25 mM Ca2+, Fig. 6f). Corresponding zoomed out images and three biological replicate microscope images are shown in Supplementary Fig. S10 and S11. This result is relatively consistent with the colorimetric biofilm assay, where an increase in biofilm was detected in the 10 mM Ca2+/EGTA and 25 mM Ca2+/EGTA samples (Fig. 4b). There are no apparent morphology changes for the L. plantarum ATCC 14917 cells, which remain in the form of bacilloid rods under all conditions. As for the colorimetric biofilm assay results, the microscopy analysis supports an extracellular role for Ca2+ in cell-cell aggregation of L. acidophilus ATCC 4356, but the microscopy analysis does not show a clear Ca2+-dependent trend for L. plantarum ATCC 14917.

Fig. 5
figure 5

Bright-field microscopy images showing calcium and chelator effects on biofilm formation at 48 h of L. acidophilus ATCC 4356. CDM medium was supplemented with 0, 10, or 25 mM CaCl2 (a–c) and 0 mM CaCl2 + 50 mM EGTA, 10 mM CaCl2 + 80 mM EGTA, or 25 mM CaCl2 + 100 mM EGTA (d–f). The scale bar (2 μm) shown on image (a) applies to all images (af). Corresponding zoomed out images are shown in Supplementary Fig. S8.

Fig. 6
figure 6

Bright-field microscopy images showing calcium and chelator effects on biofilm formation at 48 h of L. plantarum ATCC 14917. CDM medium was supplemented with 0, 10, or 25 mM CaCl2 (a–c) and 0 mM CaCl2 + 50 mM EGTA, 10 mM CaCl2 + 80 mM EGTA, or 25 mM Cl2 + 100 mM EGTA (d–f). The scale bar (2 μm) shown on image (a) applies to all images (a–f). Corresponding zoomed out images are shown in Supplementary Fig. S10.

Estimating an apparent binding affinity of calcium to L. acidophilus ATCC 4356 using biofilm quantitation

Our results thus far suggest that Ca2+ is facilitating L. acidophilus ATCC 4356 biofilm formation through an extracellular mechanism. To gain further insight, we carried out a Ca2+ titration using biofilm formation to estimate an apparent affinity of Ca2+ for L. acidophilus ATCC 4356. This analysis assumes that extracellular Ca2+ binding is correlated with the amount of biofilm formed. Using CDM medium supplemented with trace minerals, we quantified total biofilm formation at 48 h in the presence of increasing Ca2+ concentrations from 2 to 23 mM (Fig. 7). The absorbance values at 595 nm from the crystal violet staining assay were plotted against the corresponding added CaCl2 concentrations. The resulting binding isotherm was best fitted to a single site binding model using DynaFit to estimate an apparent dissociation constant (Kd) of 46 ± 0.7 µM89,90. The fitting script is shown in Supplementary Fig. S12. Ca2+ binding affinities for proteins vary widely, but tend to be stronger for cytosolic intracellular proteins (Kd ~10− 6-10− 8 M) and weaker for extracellular proteins (Kd ~10− 3-10− 6 M)91,92. This result suggests a possible surface Ca2+-binding affinity that would be reasonable for an extracellular protein, but further work is needed to determine the specific Ca2+-binding site(s) that may be involved in biofilm formation.

Fig. 7
figure 7

Total biofilm formation (a) of L. acidophilus ATCC 4356 grown with varied calcium concentrations in CDM medium supplemented with 1% Wolfe trace mineral solution and 50 mM EGTA. Total biofilm was quantified by crystal violet staining (CV assay) as measured by A595. Data used for fitting are technical replicates from three biological replicates. Error bars are shown as SD of these four representative replicates. Data in (a) was fitted by DynaFit with a 1:1 binding model (b) to give Kd = 46 ± 0.7 µM. The fitting script is shown in Supplementary Fig. S12.

Bioinformatic analysis of possible calcium-binding extracellular proteins in Lactobacillaceae

As discussed earlier, Ca2+ ions can interact with a variety of bacterial cell surface macromolecules, including proteins. Our data thus far support an extracellular role for Ca2+ in mediating biofilm formation of L. acidophilus ATCC 4356. Here, we discuss the similarities and differences in the surface compositions, particularly the surface proteins, of L. acidophilus ATCC 4356 and L. plantarum ATCC 14917, which can provide insight to the role of Ca2+ in mediating biofilm formation.

Like many Lactobacillaceae, both L. acidophilus and L. plantarum have mucus-binding proteins, which comprise a series of mucus-binding domains that are covalently anchored to the Gram-positive cell surface69,93. These proteins, among others including S-layer proteins, are thought to promote adhesion to the host and therefore gut retention68,69,70. Some mucus-binding domains have confirmed or proposed Ca2+-binding sites, including those from Limosilactobacillus reuteri 1063 and L. plantarum Lp994,95,96,97. The role of Ca2+ in these proteins has not been confirmed, but it has been proposed that these Ca2+ sites could help modulate binding with host cells97. There are no reports investigating Ca2+ binding in mucus binding proteins from L. acidophilus, but a sequence- and homology model-based analysis of at least one predicted mucus-binding protein from L. acidophilus (LBA1019) reveals that the Ca2+ binding residues found in the L. reuteri Ca2+-bound mucus-binding domain are conserved in some of the LBA1019 mucus-binding domains. L. plantarum also harbors a mannose specific adhesin (Msa), which is a mucus-binding domain containing protein involved in mannose adhesion, but there is no reported evidence for Ca2+ binding in this protein98.

Some Lactobacillaceae, including L. acidophilus, but not L. plantarum, encode for S-layer proteins99. S-layer proteins make up the two-dimensional self-assembling array of protein/glycoprotein subunits present as the outermost component of the cellular envelope in some bacteria. The resulting semi-porous lattice functions as a protective coat for the cell, a molecular sieve and ion trap, and participates in interactions with the extracellular environment70,100,101. In Lactobacillaceae and other Gram-positive bacteria, S-layer proteins bind non-covalently to cell wall components102. The S-layer of L. acidophilus is made up of the predominant glycosylated SlpA protein (UniProt ID P35829) and an auxiliary SlpX protein (UniProt ID Q5FLN0)103,104,105,106. The expression profiles of the two proteins are impacted by growth phase and high salt conditions107. One-third of the C-terminal domain of SlpA serves to anchor the protein in the cell wall, and the N-terminus, which contains 8 of the 9 glycosylation sites, is involved in self-assembly and adhesion to proteins such as mucin, collagen, and fibronectin, or to cells99,105.

S-layer proteins may be particularly notable because they are recognized to impact biofilm formation in other bacteria100,108,109,110. In L. acidophilus ATCC 4356, removal of the S-layer reduces the ability of the bacteria to autoaggregate111. These cell-cell interactions could play a role in mucosal colonization of bacteria112. Previous studies in some bacteria have also revealed that calcium can bind to S-layer proteins, strengthening the crystalline lattice and improving bacterial viability. S-layers can retain high concentrations of metals, which can also protect the cell, especially when in the presence of toxic heavy metals. Based on our results and these studies on other bacteria, it is possible that Ca2+ binding in the S-layer of L. acidophilus ATCC 4356 facilitates aggregation and biofilm formation. To investigate the possible association of the L. acidophilus ATCC 4356 S-layer with calcium, we used Foldseek to find proteins with similar structures to the predominant S-layer protein, SlpA113. There is no full-length crystal structure available, but X-ray crystal structures of individual domains of SlpA are available in the protein databank. For our analysis, we examined the reported structures and sequences representing domains within the N-terminal self-assembly region of SlpA (PDB 7QLE for domain I, residues 32–198, PDB 7QFL for domain II, residues 199–308, PDB 8BT9 for domain II, residues 201–310, and PDB 7QFG for domain III, residues 309–444)114. AlphaFold 3 models predict Ca2+ binding sites for the sequences of domains I and II (PDB 7QLE, 7QFL, and 8BT9) with good confidence intervals of 70–90 as determined using a pairwise representation to predict a modified local difference test (pLDDT, see Methods)115,116. Although a Ca2+ site was also predicted for domain III using the sequence of PDB 7QFG, this model has a low pLDDT confidence (50–70). In the model of domain I (PDB 7QLE sequence), Ca2+ is predicted to coordinate with the side chain of Glu154, a main chain carbonyl oxygen from Thr159, and the main chain amino group of Lys161, albeit with relatively long Ca2+-ligand bond distances (3.1–3.9 Å, Supplementary Fig. S13a-b). AlphaFold 3 models of domain II suggest there may be Ca2+ sites with only one residue nearby (PDB 7QFL sequence) or that represent weak interactions between Ca2+ and nearby residues (3.4–4.7 Å, PDB 8BT9 sequence) (Supplementary Fig. S13c-f). Collectively, these AlphaFold 3 models suggest that domain I has a higher probability of having a Ca2+ binding site compared to domains II and III. In addition to using AlphaFold 3 to predict possible Ca2+ binding sites in SlpA, we also used Foldseek to perform structure-based searches for Ca2+-binding sites in structurally homologous proteins. Using each of the reported structures of domains I-III in SlpA, we found Foldseek search hits for domain II that included Ca2+-binding cadherin proteins from several species, although these hits were all based on AlphaFold models or had relatively low scores and low sequence identities. Ca2+ binds at sites between extracellular domains in cadherin to facilitate linearization and rigidification117,118. Foldseek also yielded some hits for domain I that included Ca2+-binding proteins, but these were all low-scoring hits that did not align well with domain I of SlpA. The Foldseek search for domain III yielded no hits that were calcium associated, which is consistent with the AlphaFold 3 results suggesting low likelihood of a Ca2+ site in this domain.

We also investigated the structures of N-terminal domains from the auxiliary S-layer protein, SlpX (PDB 7QFI for domain I, residues 31–182, PDB 7QFJ and 7QFK for domain II, residues 194–362, and PDB 8AOL for domain III, residues 363–499)114. Although SlpX is not annotated as a Ca2+-binding protein, the crystal structures of domains I and III reveal a Ca2+ binding site in a loop region of each of the protein domains (PDB 7QFI, 8AOL). In the structure of domain I (PDB 7QFI, Fig. 8), the distances between Ca2+ and the nearest potential binding atoms, which are part of the side chains or main chains of the surrounding amino acids and one water molecule, are shown and represent a distorted octahedral geometry (verified by CheckMyMetal)119. Some of the longer Ca2+-ligand distances (3.6–3.9 Å) could represent weaker interactions. Foldseek searches on this domain reveal various AlphaFold models or low scoring hits for Ca2+-binding proteins, as well as one hit with a reasonably good score for a bacterial Ig-like domain from Bacillus cereus (PDB 2KPN)120. The Ca2+ site in this structure aligns well with the structure of SlpX domain I, although the sequence identity is only 8.8% and the coordination sphere is different (Supplementary Fig. S14). All coordinated atoms originate from carboxylate side chains, with distances ranging from 2.2 to 2.4 Å. In the crystal structure of domain III (PDB 8AOL), the Ca2+ binding site includes an asparagine, three water molecules, and two phosphate ions, but given that this site is made up of mostly solvent ligands it may not be physiologically relevant. A Foldseek search for domain III did not yield any good-scoring structurally homologous structures with likely Ca2+ binding sites. An AlphaFold 3 model for domain II (PDB 7QFJ sequence) has a low confidence score for a Ca2+ site, suggesting there is no Ca2+ binding site in this domain. Overall, this analysis suggests that it is possible that Ca2+ ions could play a role in S-layer protein assembly and/or stabilization in L. acidophilus ATCC 4356, particularly through weak binding interactions with SlpX domain I and/or SlpA domain I, which could promote biofilm formation, but further experimental studies would be needed to test this hypothesis.

Fig. 8
figure 8

Structure of SlpX and the Ca2+-binding site. Crystal structure (a) of domain I (amino acids 31–182) of S-layer protein SlpX from L. acidophilus (PDB 7QFI). The Ca2+ ion is shown as a green sphere and the coordinating side chains and main chain atoms are shown as sticks. Detailed close-up view (b) of the Ca2+-binding site from (a). The bond distances from each side chain or main chain ligand are labelled in black. Ca2+ is shown as a small sphere for a clearer view of the coordination site. A nearby water molecule is shown as a small red sphere.

We also analyzed the reported proteomes of each species to identify additional predicted Ca2+-binding extracellular proteins. To identify putative Ca2+-binding proteins in L. acidophilus ATCC 4356, we used the published proteome (UniProt Knowledgebase, UniProtKB) of L. acidophilus NCFM, which is 99.96% identical to strain ATCC 4356121. Likewise, we used the published proteome (UniProtKB) of L. plantarum ATCC 14917 to identify known and predicted Ca2+ binding proteins in strain ATCC 14917. We also examined the proteome of another related L. plantarum strain, WCFS1, which has more reviewed proteins. By searching the proteomes for proteins annotated with the keyword “calcium”, we identified several predicted calcium-binding proteins. After excluding Ca2+ transporters and likely intracellular proteins, we identified a predicted Ca2+ binding surface protein in L. acidophilus (gene name LBA1633, UniProt ID Q5FIM8). This protein is larger than SlpA or SlpX (1650 amino acids compared to 444 and 499, respectively) and has two relatively small domains which have been structurally characterized (PDB 6S5W, residues 1165–1268 and PDB 6SX1, residues 1269–1351)122. The characterized domains are Rib domains, of which there are several in the protein. These types of domains have been linked to infection and immune invasion in pathogens122. This surface protein is only 22% similar to SlpA (sequence alignment performed on UniProtKB). AlphaFold 3 models predict Ca2+ binding sites with very high pLDDT confidence of > 90 for the sequences of PDB 6S5W and 6SX1. The Ca2+ ion is predicted to coordinate with an Asp (bidentate) and interact with the main chain carbonyl oxygens of Ala and His residues for both structures (Fig. 9). Foldseek hits reveal that the “Rib long” domain of this surface protein (PDB 6S5W) overlays with several Ca2+-binding proteins that have similar folds, including a fragment of the binding region of SraP, a surface protein from Staphylococcus aureus involved in adhesion to host cells (PDB 4M02)123, a structural Ca2+-binding domain in a hydrolase (PDB 6DMS)124, and a protocadherin (PDB 4ZPM)125. The former two structures each have a single Ca2+ ion bound, and the protocadherin domain has a three Ca2+ ion site. Structural alignments reveal that each Ca2+ site overlays with a loop region between two two-layer β sheet folds, similar to extracellular cadherin (EC) domains. We also performed a Foldseek search for the “Rib standard” domain (PDB 6SX1) of this L. acidophilus surface protein. Here, fewer Ca2+-binding proteins were found as matches compared to the Rib long domain and these matches had poorer scores. To further explore the predicted calcium binding sites of surface protein Q5FIM8, we performed a protein sequence-based (BLASTP) search using Q5FIM8 as the query and limiting the taxonomy of results to representative intestinal Lactobacillaceae species and the model organisms E. coli and B. subtilis (see Methods). BLASTP hits from this search include mostly Rib/alpha-like domain-containing proteins from L. acidophilus, Lactobacillus gasseri, Limosilactobacillus reuteri, and Lactobacillus salivarius. Multiple sequence alignment between the Rib long domain of Q5FIM8 (PDB 6S5W sequence) and these BLASTP hits highlights their similarity (Supplementary Fig. S15). The residues making up the predicted Ca2+ binding site (as main chain or side chain ligands) in the AlphaFold 3 model of the Q5FIM8 Rib long domain (Ala4, His7, and Asp90 in Supplementary Fig. S15) are moderately conserved between the species. Next, we conducted a structural motif search using RCSB PDB (see Methods) with Q5FIM8 Rib long domain (PDB 6S5W sequence) and its VTYPD motif as the query and refining search results to PDB structures with calcium ions. We first chose VTYPD as the structural motif query because the sequence alignment revealed that this motif was well conserved between the species and one of the residues predicted to be involved in Ca2+ binding in the Q5FIM8 Rib long domain is nearby. This search yielded only one structure with a Ca2+ site (PDB 8PXC) so we reduced the search sequence to TYPD, which yielded additional results. Multiple sequence alignment between the Q5FIM8 Rib long domain and Ca2+-binding structure results from the TYPD motif search (mostly proteins from Streptococcus species) revealed conservation of TYPD-STD motifs between several Lactobacillaceae and Streptococcus species (residues 417–424, Supplementary Fig. S16). Many of these structures have calcium binding sites conserved at Ala/Thr/Asn (residue 291), Tyr/His (residue 294), and Asp (bidentate, residue 424), along with a weak interaction (4–5 Å) at Tyr (residue 418) (Supplementary Fig. S16). Sequence and structure analyses agree with the AlphaFold 3 prediction for a Ca2+ site in the Q5FIM8 Rib long domain (Supplementary Figure S16).

Fig. 9
figure 9

AlphaFold 3 models of L. acidophilus surface protein (UniProt ID Q5FIM8) with predicted Ca2+-binding sites. Ca2+ ions are shown as spheres and all coordinating side chain and main chain atoms are shown as sticks. AlphaFold 3 model (a) of the surface protein Q5FIM8 Rib long domain (generated from PDB 6S5W sequence, residues 1165–1268). The Ca2+ ion is shown as a green sphere. Detailed close-up view (b) of the Ca2+-binding site from (a). The bond distances from each side chain or main chain ligand are labelled in black. Ca2+ is shown as a smaller sphere for a clearer view of the coordination site. Structural alignment (c) of the Rib long and Rib standard domains of the surface protein (Rib long domain: PDB 6S5W sequence, residues 1165–1268; Rib standard domain: PDB 6SX1 sequence, residues 1269–1351). The Ca2+ ion is shown as a green sphere for the Rib long domain and a purple sphere for the Rib standard domain. Detailed close-up view (d) of the Ca2+-binding sites from (c). The bond distances from each side chain or main chain ligand are labelled in black for the Rib long domain and purple for the Rib standard domain. All amino acids except those involved in coordinating Ca2+ ions are omitted for a clearer view of the coordination site.

Collectively, based on this bioinformatic analysis of extracellular and known or predicted calcium-binding proteins in L. acidophilus and L. plantarum, we find that L. acidophilus has several potential Ca2+-binding surface proteins compared to L. plantarum. These extracellular proteins are likely to have structural roles in the observed Ca2+-induced biofilm formation of L. acidophilus ATCC 4356.

Conclusions

This study reveals that Ca2+ ions play roles in growth and biofilm formation of Lactobacillaceae species. Calcium is beneficial to the growth of both L. acidophilus ATCC 4356 and L. plantarum ATCC 14917 although the impacts on the growth kinetics vary somewhat between the species and in different growth media. Ca2+ ions strongly promote biofilm formation of L. acidophilus ATCC 4356. This Ca2+-induced L. acidophilus ATCC 4356 biofilm mass accumulation can be prevented in the presence of an extracellular chelator, suggesting that extracellular Ca2+ ions are important for biofilm formation of this species. On the other hand, Ca2+ induces only minor effects on biofilm formation of L. plantarum ATCC 14917, with some additional biofilm mass detected at lower Ca2+ concentrations and less at higher Ca2+ concentrations, and an extracellular chelator does not reverse these effects. Microscopy analysis also shows that L. acidophilus ATCC 4356 undergoes a change in morphology with the addition of Ca2+, which may affect the cell-cell interactions thought to be important for biofilm formation. Given that L. acidophilus harbors S-layer proteins not found on L. plantarum, and that our bioinformatic analysis reveals known and potential Ca2+ binding sites in these proteins and another predicted L. acidophilus surface protein, we hypothesize that the mechanism of Ca2+-induced biofilm promotion involves Ca2+ binding to the surface of L. acidophilus ATCC 4356. Ca2+ ions may structurally stabilize surface proteins to enhance cell-cell interactions and/or form binding sites between cell surface proteins. Overall, this work highlights two separate roles for Ca2+ ions in Lactobacillaceae species, as an intracellular nutrient and extracellular enhancer of cell-cell interactions, both of which may impact their ability to survive in a multi-species community such as the gut microbiota. This work also provides insight to understanding why Ca2+ supplementation can promote the viability of Lactobacillaceae in the gastrointestinal tract. More broadly, similar roles for Ca2+ in other intestinal bacteria could be considered. Understanding the roles of Ca2+ ions and other metal nutrients in the survival and persistence of gut bacteria is important for unraveling how diet affects health and disease through the gut microbiota.

Methods

General considerations

All reagents were purchased from commercial sources and used as received. Trace metals and amino acids are BioXtra (99.9%) and BioUltra (99.5%) grade from Sigma-Aldrich. De Man, Rogosa, and Sharpe (MRS medium) and mucin from porcine stomach (type II) were purchased from Sigma-Aldrich. Aqueous solutions were prepared using Milli-Q water. Solutions were treated with Chelex-100 resin (Bio-Rad) according to the manufacturer’s batch protocol to remove metal ions. Chelex-treated solutions were stored in acid-washed plastic containers and transferred using acid-washed pipet tips.

Bacterial strains and culture conditions

The bacterial strains used in this study are Lactiplantibacillus plantarum ATCC 14917 and Lactobacillus acidophilus ATCC 4356. Bacteria were incubated aerobically at 37 °C, 5% CO2 with shaking in MRS medium126. Bacteria stocks were stored in MRS medium containing 25% (vol/vol) glycerol at − 80 °C. Metal-limited and chemically defined medium (CDM) were used to study calcium effects (previously reported and described below)71.

Media and supplements

MRS broth was prepared using MRS broth powder (51 g/L, Sigma-Aldrich) and 1 mL polysorbate 80 (Tween 80, Sigma-Aldrich). MRS/agar plates were prepared with MRS agar powder (61 g/L, Sigma-Aldrich) and 1 mL Tween 80. CDM medium was adapted from McFeeters et al.127 using the amino acid concentrations from Wegkamp et al.128 and was prepared from several Chelex-treated or trace metal reagent-containing stock solutions71. Wolfe solution was used as a trace mineral supplement for cultures grown in CDM medium, and contains 1.5 g/L NTA, 1 g/L NaCl, 0.1 g/L FeSO4, 0.1 g/L Co(NO3)2, 0.018 g/L ZnSO4, 0.01 g/L CuSO4, 0.01 g/L H3BO3, 0.01 g/L Na2MoO4, 0.001 g/L Na2SeO3, 0.01 g/L Na2WO4, 0.02 g/L NiCl2129. Trace metal grade salts were used for components that could not be treated with Chelex. CDM medium and Wolfe solution were prepared and stored in sterile and metal free plastic centrifuge tubes. mCDM medium was prepared by adding 0.1% mucin to CDM medium.

Measurement of planktonic growth kinetics

Starter cultures were grown overnight (18 h to 20 h) in MRS from a fresh (1 week or less) MRS/agar streak plate (Scheme S1, day 1). On day 2, subcultures in MRS, CDM, and mCDM were prepared. MRS subcultures were prepared by inoculating 1:100 overnight starter culture in MRS and incubating for 22–24 h. Overnight starter cultures were washed with CDM before inoculation into fresh CDM (1:25 ratio) or fresh CDM supplemented with 0.1% mucin (mCDM, 1:25 ratio) and grown for 22–25 h for all CDM and mCDM subcultures, respectively. On day 3, each of the above subcultures was then washed and inoculated in CDM (1:100 ratio) supplemented with varied concentrations of CaCl2 and 1% Wolfe trace mineral solution. To improve the solubility of CaCl2 in CDM medium, medium supplemented with 0.1 and 1 mM CaCl2 contains 2 mM EGTA and 5 mM NTA; medium supplemented with 5, 10, and 25 mM CaCl2 contains 5 mM EGTA and 5 mM NTA. A similar approach was taken to monitor the effects of Mg2+ on planktonic growth kinetics using the same concentration range for MgCl2 as CaCl2, as well as the same amounts of EGTA and NTA in each sample. CDM medium used for Mg2+ study was prepared without MgSO4. Growth kinetics were monitored every 30 min with shaking using a multimode microplate reader (Tecan Spark 10 M) with 96-well polypropylene clear round-bottom plates (Greiner Bio-One). Measured data are shown with error bars as SD from three biological replicates with ≥2 technical replicates each. Lag time, growth rate, and max OD represent the mean ± SEM of three biological replicates, each with ≥2 technical replicates. Growth parameters were calculated using nonlinear regression curve fitting to a 4-parameter Logistic equation with GraphPad Prism 9 software. ns (not significant), *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001, ****p ≤ 0.0001 as determined by one-way ANOVA with Tukey multiple comparison test.

Biofilm quantification using crystal violet (CV) assay

Inoculated cultures under varied conditions were prepared as previously described in the growth kinetics section and were grown in polystyrene microplates (nontreated surface, flat bottom) without shaking at 37˚C and 5% CO2 for 24–48 h to form biofilms. To quantify total biofilms, we first discarded the unbound bacteria by briskly shaking out the liquid in the microplate over a waste tray. After that, biofilms were washed two times by filling the microplate with DI water and shaking the liquid out. Next, the bacteria were stained with 0.1% (w/v) crystal violet (CV) for 30 min at room temperature, then unbound crystal violet was washed off by filling the microplate with DI water and shaking the liquid out three times, and the stain was released from the biofilms by treatment with 95% (v/v) ethanol for 30 min. The amount of biofilm was quantified based on the absorbance of released stain at 595 nm (A595). The higher this value is, the more biofilm was formed. Measured data are shown with error bars as SD from three biological replicates with ≥2 technical replicates. Significance was determined by one-way ANOVA with Tukey multiple comparison test: ns (not significant), *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001, ****p ≤ 0.0001​.

Viability assay for the planktonic cells and biofilms

Inoculated cultures under varied conditions were prepared as previously described in the growth kinetics section and were grown in polystyrene microplates (nontreated surface, flat bottom) without shaking at 37˚C and 5% CO2 for 24–48 h to form biofilms. After 24 and 48 h, planktonic cells from each microplate well were collected and serial dilution was performed in 0.85% saline solution. Sets of dilutions were streaked on MRS agar plates and incubated for 24 h. Colonies were counted manually or with ImageJ. ​

To determine viabilities of biofilm cells, we discarded unbound bacteria by aseptically briskly shaking out the liquid in the microplate over a waste tray. Biofilms were then aseptically washed two times by filling the microplate with sterile PBS and shaking the liquid out. ​Washed biofilms were resuspended in sterile PBS (200 µL/well) by scratching the well surface and wall with a 2 mm flat chisel blade. To avoid cross contamination between wells, the blade was rinsed after each use with sterile water, flame sterilized, then allowed to cool. Before each use, the blade was rinsed one more time with sterile water. ​Serial dilution of each resuspended biofilm sample was performed in 0.85% saline solution. Sets of dilutions were streaked on MRS agar plates and incubated for 24 h. Colonies were counted manually or with ImageJ. ​

Viable counts are expressed as CFU/well. Measured data are shown with error bars as SD of three biological replicates, each with ≥ 2 technical replicates. Significance was determined by one-way ANOVA with Tukey multiple comparison test: ns (not significant), *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001, ****p ≤ 0.0001.​

Effect of chelator on the growth kinetics and biofilm formation of L. acidophilus ATCC 4356 and L. plantarum ATCC 14917

Inoculated cultures under varied conditions were prepared by washing the overnight MRS subculture and inoculating the bacteria in CDM (1:100 ratio) supplemented with 1% Wolfe trace mineral solution and varied concentrations of CaCl2 and chelator at 37˚C and 5% CO2. For media supplemented with 0, 10, and 25 mM CaCl2, EGTA was added to a final concentration of 50, 80, and 100 mM respectively. Growth kinetics were monitored every 30 min with shaking using a multimode microplate reader (Tecan Spark 10 M) using 96-well polypropylene clear round-bottom plates (Greiner Bio-One). For biofilm studies, inoculated cultures were grown in polystyrene microplates (nontreated surface, flat bottom) without shaking at 37˚C and 5% CO2 for 48 h to form biofilm. Measured data are shown with error bars as SD from three biological replicates with ≥2 technical replicates. Control experiments to estimate the chelator impact on these species without added Ca2+ were performed by washing overnight MRS starter cultures and inoculating the bacteria in CDM (1:100 ratio) supplemented with 1% Wolfe trace mineral solution with no added CaCl2 and varied chelator final concentrations of 50, 80, or 100 mM EGTA at 37˚C and 5% CO2. Growth kinetics analyses and total biofilm quantitation were performed as previously described. Measured data are shown with error bars as SD from three biological replicates with ≥2 technical replicates. Significance was determined by one-way ANOVA with Tukey multiple comparison test: ns (not significant), *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001, ****p ≤ 0.0001.​

Bright-field microscopy analysis

CultureWell™ Reusable Gaskets (Grace Bio-labs) with silicon padding were adhered to glass slides and autoclaved. Overnight bacterial cultures were washed and inoculated in CDM (1:100 ratio) supplemented with 1% Wolfe trace mineral solution and 0, 10, or 25 mM CaCl2 (medium supplemented with 10 or 25 mM CaCl2 contains 5 mM EGTA and 5 mM NTA). To observe a chelator effect in medium supplemented with 0, 10, and 25 mM CaCl2, EGTA was added to a final concentration of 50, 80, and 100 mM, respectively. 100 µL of each culture was pipetted into a well, incubated at 37 °C and 5% CO2 without shaking and in a humidity-controlled container. After 48 h, the glass slide with live biofilm was examined under the microscope without further washing or treatment. Microscope images were acquired using an ECHO Revolve microscope equipped with a 12 MP Color Camera.

Estimating binding affinity of calcium to L. acidophilus ATCC 4356 cell surface

L. acidophilus ATCC 4356 starter cultures were grown overnight (20 h) in MRS from a fresh (1 week or less) MRS/agar streak plate. On the next day, the MRS subculture was washed and inoculated in CDM (1:100 ratio) supplemented with 1% Wolfe trace mineral solution and varied concentrations of CaCl2 and 50 mM EGTA. Inoculated cultures were grown in polystyrene microplates (nontreated surface, flat bottom) without shaking at 37˚C and 5% CO2 for 48 h to form biofilms. Total biofilm was quantified using the crystal violet staining (CV assay) and measuring absorbance at 595 nm. Measured data are shown with error bars as SD of three biological replicates, each with 4 technical replicates. Absorbance at 595 nm (A595) was fitted by DynaFit with a 1:1 binding model to determine an apparent binding affinity (Kd)89,90.

Bioinformatic analysis of extracellular calcium binding proteins in Lactobacillaceae

To identify potential calcium binding sites in known extracellular surface proteins, including S-layer proteins of L. acidophilus, we searched for homologous structures with Ca2+ binding sites. Here, Foldseek was employed using PDB structures of surface protein domains. For each PDB structure, the Foldseek search was carried out on all structural databases available as of March 2024113. To perform a Foldseek search, the PDB file of each query was uploaded on the Foldseek webserver (https://search.foldseek.com/search). The search was performed on databases including AlphaFoldDB (version 4: Proteomes and Swiss-Prot), AlphaFoldDB (version 4) and CATH clustered at 50% sequence identity, Global Microbial Gene Catalog, ESM Atlas High-Quality and Protein Data Bank (PDB)130.

To search for additional putative calcium-binding proteins, we downloaded the complete proteome of L. acidophilus NCFM from UniProtKB database. L. acidophilus NCFM is 99.96% similar to L. acidophilus ATCC 4356 by comparative genomic analysis121. Likewise, we also downloaded the complete proteomes of L. plantarum ATCC 14917 and WCFS1 from UniProtKB. We searched all proteomes for annotated calcium proteins by using the keyword “calcium”. After filtering out all calcium transporters, we evaluated the remaining surface proteins with predicted Ca2+ binding sites using Foldseek.

To predict calcium binding sites for each PDB structure, we modeled Protein-Ion structures with AlphaFold 3 available on AlphaFold Server (https://alphafoldserver.com) using the sequence of the corresponding PDB structure as input for protein and Ca2+ as input for ion. AlphaFold measures a per-residue estimate of its confidence, specifically a pairwise representation to predict a modified local difference test or pLDDT, on a scale from 0 to 100. Regions with pLDDT higher than 90 are models expected to have high accuracy. Regions with pLDDT between 70 and 90 are considered modelled well (general good backbone prediction). Models with pLDDT between 50 and 70 should be treated with caution, and models with pLDDT lower than 50 should not be interpreted.

To discover homologs of surface protein Q5FIM8, we performed protein sequence-based searches using BLASTP in Escherichia coli K12 (taxid 83333), Bacillus subtilis subsp. subtilis str. 168 (taxid 224308), and representative intestinal Lactobacillaceae species in the non-redundant protein sequences database. Representative intestinal Lactobacillaceae species include Lactobacillus acidophilus NCFM (taxid:272621), Lactobacillus crispatus ST1 (taxid:748671), Lactobacillus delbrueckii subsp. bulgaricus ATCC 11842 (taxid:390333), Lactobacillus gasseri ATCC 33323 (taxid:324831), Lactobacillus reuteri DSM 20016 (taxid:557436), Lactobacillus fermentum IFO 3956 (taxid:334390), Lactobacillus vaginalis ATCC 49540 (taxid:1423814), Lacticaseibacillus paracasei ATCC 334 (taxid:321967), Lactobacillus rhamnosus GG (taxid:568703), Lactobacillus salivarius UCC118 (taxid:362948), Lactobacillus plantarum WCFS1 (taxid:220668), Lactobacillus sakei str. 23 K (taxid:314315), Lactobacillus brevis ATCC 367 (taxid:387344). BLASTP searches were initiated with the default substitution matrix BLOSUM-62 and the cutoff criteria used to identify homologs were based on the query coverage, E-value, total score, and percent identity. We manually inspected BLASTP hits and adjusted cutoff values for these parameters to include cases that have conserved motifs with the target sequence. Therefore, hits with query coverage more than 10%, E-values less than 1.E-06, total scores above 50, and percent identity above 30% were chosen. To investigate conserved motifs between surface protein Q5FIM8 and its BLASTP hits, we performed multiple sequence alignment between the Q5FIM8 Rib long domain sequence (corresponding to PDB 6S5W) and the aforementioned BLASTP hits (Supplementary Fig. S15a) using COBALT: Multiple Alignment Tool webserver (https://www.ncbi.nlm.nih.gov/tools/cobalt/re_cobalt.cgi). Aligned sequences were analyzed with Jalview.

The TYPD structural motif search was performed on the RSCB PDB website using Advanced Search Query Builder. Input query is listed as followed, QUERY: Structure Motif = 6S5W (A_1–87 [THR] AND A_1–88 [TYR] AND A_1–89 [PRO] AND A_1–90 [ASP]) AND RMSD Cutoff = 2 AND Atom Pairing = “All Atoms” AND Chemical Name HAS EXACT PHRASE “calcium”. Conserved motifs between Q5FIM8 Rib long domain sequence (corresponding to PDB 6S5W) and the hits from this search were analyzed by PyMOL and multiple sequence alignment was performed with COBALT: Multiple Alignment Tool. Aligned sequences were analyzed with Jalview.

Statistical analysis

All statistical analyses were carried out using GraphPad Prism 9 software. Significance was determined by one-way ANOVA with Tukey multiple comparison test: ns (not significant), *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001, ****p ≤ 0.0001.