The effect of calcium palmitate on bacteria associated with infant gut microbiota

Abstract Gut microbiota development in formula‐fed and breast‐fed infants is known to differ. This could relate to the usage of unmodified vegetable oil instead of mammalian fat in infant formula (IF), causing the enhanced formation of the poorly soluble soap calcium palmitate (CP) in the infant's gut. Here we investigate in vitro the possible influence of CP on the infant gut bacteria. The growth of several bacterial species dominant in the infant's gut was analyzed by culturing in media with CP. Faecalibacterium prausnitzii as a sensitive representative was analyzed in detail by scanning transmission electron microscopy, membrane staining, gas chromatography, and microbial fuel cell experiments. Of all bacteria tested, the growth of several bifidobacteria and F. prausnitzii was reduced at 0.01 mg/ml CP, Bifidobacterium infantis stopped growing completely. CP reduced the cell envelope thickness of F. prausnitzii, disturbed the cell membrane fatty acids and function of membrane proteins involved in electron transport. CP inhibited the growth of bifidobacteria and faecalibacteria. This suggests that modification of fat in IF may benefit the development of the gut microbiota in formula‐fed infants by supporting the colonization of important beneficial bacteria in early life. Future clinical studies are needed to confirm this.

strongly stimulate the bifidobacterial population in the infant's gut (Moore & Townsend, 2019). Nowadays, IF is often supplemented with various oligosaccharides to mimic the effects of breast milk on the gut microbiota.
Another difference is that during the production of IF, milk fat is often replaced by vegetable oils. Both fats are triglycerides, that is esters of fatty acids and glycerol, but there are important differences (Havlicekova et al., 2015). In breast milk triglycerides, ~60% of the major saturated fatty acid, palmitic acid, is esterified at the sn-2 (or β-) position (Innis et al., 1994). In contrast, triglycerides in standard IFs have only 13% of the palmitic acid esterified at the sn-2 position, while most palmitic acid is esterified at the sn-1 and sn-3 positions.
Yet, some IFs with modified vegetable oil contain triglycerides with ~50% palmitic acid at the sn-2 position (Yaron et al., 2013). Upon digestion, palmitic acid esterified at the sn-2 position will be absorbed in the form of mono-acylglycerol. However, palmitic acids at the sn-1 and sn-3 position are released into the gut and can then bind minerals like calcium to form insoluble soaps, especially calcium palmitate (CP). Such soaps can cause a hard stool, and lower the calcium and fatty acids absorption (Manios et al., 2020;Havlicekova et al., 2015). Changed stool consistency suggests that CP formation might influence the gut microbiota. Thus, it was proposed that sn-2 esterified palmitic acid in IFs may influence the composition of the infant gut microbiota, especially the levels of lactobacilli and bifidobacteria (Yaron et al., 2013). In turn, this might influence late-colonizing butyrate-producing bacteria, like Faecalibacterium prausnitzii, which is regarded as a potential probiotic because of its anti-inflammatory properties (Goffau et al., 2013;Sokol et al., 2008).
Little was thus far known about the effects of CP on bacteria colonizing the gut of infants. In the present study, we selected a panel of infant gut bacteria based on their described abundance and important role in gut microbiota development. We aimed at investigating the effect of CP on the growth characteristics of these selected bacteria in vitro, including early-colonizing bifidobacteria and Bacteroides spp., and the late-colonizing F. prausnitzii. Furthermore, we investigated how interactions between bifidobacteria, F. prausnitzii, and Bacteroides thetaiotaomicron are influenced by CP.

| Bacterial strains
Bacterial strains used in this study were obtained from culture collections (ATCC, DSMZ, NIZO) and our local strain collection (Dept. of Medical Microbiology, UMCG, Groningen, Netherlands (MMB)).

| Growth conditions
Strains were inoculated in YCFAG medium (Lopez-Siles et al., 2012) and incubated under anaerobic conditions (80% N 2 , 12% CO 2 , and 8% H 2 ) at 37°C. To measure the effects of CP on bacterial growth, the YCFAG medium was supplemented with different concentrations of CP. To this end, 1.0 g CP (Cayman Chemical) was dissolved into 20 ml of propionic acid (Sigma). Subsequently, the dissolved CP was filtered and added to a sterile YCFAG medium.

| Influence of CP on bacterial growth in mono-and co-cultures
For monocultures, 50 μl overnight culture of each strain was used to inoculate 5 ml of YCFAG medium with different concentrations of CP. In co-culture experiments, two combinations of bacterial strains (F. prausnitzii and B. longum spp., or F. prausnitzii and B. thetaiotaomicron) were used to inoculate 5 ml CP-supplemented YCFAG media with equal numbers of colony-forming units (CFU, 1 × 10 7 cells) of F. prausnitzii, B. longum spp. and/or B. thetaiotaomicron from overnight monocultures. Growth was monitored spectrophotometrically by measuring the optical density at 600 nm (OD 600 ). All growth experiments were performed in triplicate.

| Quantification of bacteria in co-culture experiments by fluorescent in situ hybridization (FISH)
Samples of co-cultures were collected at different time points after inoculation. FISH was performed according to the procedure described by Harmsen . Details of the experimental procedure are presented in the Appendix.

| Staining of F. prausnitzii with FM 4-64
Faecalibacterium prausnitzii A2-165 was cultured for 16 h at 37°C in YCFAG medium without or with either 0.003, 0.006, 0.01, or 0.02 mg/ml CP. Culture aliquots of 100 µl were collected and centrifuged at 8000 g for 4 min. Cell pellets were incubated with 200 μl FM 4-64 membrane dye solution (5 μg/ml) for 10 min at room temperature. Then 10 μl of cell suspension was mounted on a slide coated with a 2% agarose pad, and the slides were kept in the dark for 15 min before imaging with a Leica epifluorescence microscope equipped with a Nikon EOS500 camera. The fluorescence intensity was measured for 25 cells per micrograph using ImageJ software (Version 1.51n; National Institutes of Health, USA).

| Fatty acid composition of bacterial cell membranes
Faecalibacterium prausnitzii A2-165 was grown for 16 h in YCFAG medium without or with either 0.003 or 0.01 mg/ml CP, and YCFAG without propionic acid followed by a 2-h treatment with propionic acid or 0.03 mg/ml CP. The fatty acid composition was measured by gas chromatography as described by Muskiet (Muskiet et al., 1983).
For details, see the Appendix.

| Scanning transmission electron microscopy (STEM)
To investigate cell structure changes in F. prausnitzii A2-165 without or with either 0.003 or 0.01 mg/ml CP, and YCFAG without propionic acid followed by a 2-h treatment with 0.03 mg/ml CP, STEM images were taken using a Zeiss Supra55 SEM equipped with an external scan generator (ATLAS, Fibics, Canada). Large area scans enabled the analysis of many bacteria within one data set. Sample preparation was based on a protocol described by Silva (Silva et al., 2014) with some modifications. For details, refer to the Appendix.
The MFC experiments were conducted as previously described by Khan (Khan, Browne, et al., 2012). For details, refer to the Appendix.

| Statistical analyses
Statistical analyses were performed using GraphPad Prism version 5 (GraphPad Prism, San Diego, CA, USA). Unpaired t-tests (twotailed) were performed to assess significance. P-values <0.05 were regarded as significant.

| Effects of CP on the growth of infant gut bacteria
Breast-fed and formula-fed infants display a different gut microbiota composition, which may relate to CP formation .
To investigate how CP could influence the gut microbiota, in vitro growth experiments were performed with a representative panel of dominant bacteria from the infant gut (Table 1). Figure

| Influence of CP on cell morphology of F. prausnitzii A2-165
To investigate whether CP affects the cell morphology of F. prausnitzii, this bacterium was grown for 16 h in YCFAG medium with either 0, 0.003, or 0.01 mg/ml CP, and YCFAG without propionic acid followed by a 2-h treatment with 0.03 mg/ml CP. Subsequently, the cell morphology was inspected by STEM. The most prominent detected change was a reduced cell envelope thickness. Figure 2b,c show that the cell envelope thickness decreased from 36.2 nm to 16.1 nm upon increasing CP concentration from 0 to 0.03 mg/ml, respectively. When F. prausnitzii was treated with 0.03 mg/ml CP, a condition that stops the bacterial growth (Figure 1a), cell envelope thickness was significantly decreased (p < 0.0001; Figure 2c). This shows that CP affects the cell envelope structure of F. prausnitzii.

| Microbial fuel cell
Faecalibacterium prausnitzii A2-165 can employ an extracellular electron shuttle of flavins and thiols to transfer metabolically generated electrons to oxygen, allowing this obligate anaerobe to grow in a niche with oxygen influx from epithelial cells (Khan, Browne, et al., 2012;Khan, Duncan, et al., 2012b). This extracellular electron when F. prausnitzii A2-165 was exposed to 0.03 mg/ml CP ( Figure 2d and Table 2). Also, the time necessary to reach the maximum current was increased at increasing CP concentrations. In particular, it took non-treated bacteria ~11 min to generate a maximum current of ~26.5 mA, while it took bacteria treated with 0.03 mg/ml CP ~37 min to generate a maximum current of ~19.5 mA (Table 2). This shows that exposure of F. prausnitzii to CP interferes with EET and suggests an impaired function of membrane proteins needed for EET.

| DISCUSS ION
Vegetable oil blends, rich in sn-1 and sn-3 palmitate, are commonly used as the main fat source in IF. This may lead to increased CP concentrations in the infant's gut. Since CP is an almost insoluble soap with a water solubility of 0.03 mg/ml, it can disturb nutrient absorption in the gut of formula-fed infants (Forsyth et al., 1999).
Accordingly, it was previously reported that CP is associated with hard stools and decreased absorption of calcium and fatty acids . Our present study is the first to show that

CP inhibits in vitro growth of various prominent infant gut bacteria.
We report also that the growth of several bifidobacteria, such as B.
infantis, B. breve, and B. bifidum, as well as three different strains of F. prausnitzii, is inhibited by CP even at low concentrations of 0.003 or 0.01 mg/ml. In contrast, the growth of other gut microbes, like E. coli and B. thetaiotaomicron, is not affected by CP.
The exact CP concentration in the infant's gut is unknown (Jandacek, 1991). Digestion of dietary fat occurs on emulsified triglyceride droplets with lipase splitting interfacial triglycerides into monoglyceride and free fatty acid. Local high concentrations of free medium-chain and branched-chain fatty acids may, however, prevent precipitation of calcium soaps. Other factors, like bile salts and soap solubilization by low-melting fatty acids, may enhance the absorption of calcium and fat (Jandacek, 1991 palmitic acid varied from 16 to 48 mg/g dry stool weight, and the calcium concentration was ~20 mg/g. In contrast, for formula-fed infants, 72-187 mg/g palmitic acid and 32 mg/g calcium were reported (Bar-Yoseph et al., 2016;Nowacki et al., 2014;Yao et al., 2014). Based on these numbers and stool water contents of ~73%, we estimate the maximum CP amount in the gut of breast-fed infants at ~9 mg/g wet stool weight and in formula-fed infants ~23 mg/g ( Bar-Yoseph et al., 2016). However, breast milk fat includes higher mediumchain and branched-chain fatty acids, which increase CP solubility.
Therefore, the CP concentration may be lower in breastfed infants' stools. Nevertheless, the estimated CP amounts are so high that most CP will precipitate, so that the actual dissolved concentration in the infants' guts may approach the highest dissolved CP concentration in our study (0.03 mg/ml). Importantly, even a low CP concentration (0.01 mg/ml) inhibited the growth of some beneficial gut bacteria. Here, it is noteworthy that the common dietary emulsifier sodium stearoyl lactylate (SSL) has similar effects on human gut microbiota as IF. Elmén (Elmén et al., 2020) reported that a low concentration of 0.025% (w/v) of SSL already altered the human gut microbiota. Thus, our findings could help to explain the observation that B. longum and B. infantis are better colonizers of breastfed infant guts, while guts of formula-fed infants are more often colonized by E. coli, Clostridium difficile, and B. fragilis group members (Penders et al., 2005(Penders et al., , 2006Yasmin et al., 2017). Considering the different CP amounts in the guts of breast-fed and formula-fed infants, we hypothesize that excess formation of calcium soaps contributes to differences in gut microbiota composition in breast-fed and formulafed infants. This is an innovative concept because current attempts to establish a regular gut microbiota in bottle-fed infants rely on IF supplementation with probiotics and fructo-, galacto-or human milk oligosaccharides. In contrast, our results suggest that modified fats rich in sn-2 palmitic acid, or long-chain (poly-)unsaturated fatty acid F I G U R E 4 Numbers of F. prausnitzii A2-165 cultured with or without other bacteria characteristic for the infant gut microbiota. The bacterial numbers were counted by FISH upon growth for 24 h in YCFAG medium without or with different concentrations of CP. F. prausnitzii was grown either in monoculture or co-culture with ( may prevent CP formation, thereby also protecting the microbiota (Jandacek, 1991;Forsyth et al., 1999).
Possible mechanisms underlying the detrimental effects of CP on gut bacterial growth were investigated using F. prausnitzii A2-165.
Cell membrane staining, measurements of the fatty acid composition of the membrane, and imaging of bacterial morphology by STEM revealed that CP affects the integrity of F. prausnitzii's cell envelope.
This idea was supported by MFC experiments, showing that increasing CP concentrations decreased the bacterial capacity for EET. This demonstrates that CP affects the EET machinery, which is sufficient to explain the growth impairment of F. prausnitzii. However, CP could also affect other physiological processes in the cell envelope of F. prausnitzii, as suggested by the observed growth inhibition of bacteria that cannot perform EET, like B. fragilis.
Altogether, our study shows that CP inhibits the in vitro growth Alfred Haandrikman for the critical review of the manuscript.

CO N FLI C T S O F I NTE R E S T
None declared. Writing-review and editing-Lead.

DATA AVA I L A B I L I T Y S TAT E M E N T
All the data are provided in full in the result section of this paper except for the STEM data which is available at http://nanot omy.org/

Quantification of bacteria in co-culture experiments by fluorescent in situ hybridization (FISH)
Samples of co-cultures were collected at different time points after inoculation. FISH was performed according to the procedure described by Harmsen

Fatty acids composition of bacterial cell membranes
Faecalibacterium prausnitzii A2-165 was grown for 16 h in YCFAG medium with or without CP. The fatty acid composition was measured by gas chromatography as described by Muskiet (Muskiet et al., 1983). 10 ml aliquots of the cultures were centrifuged at 10,000 g for 15 min at 4°C twice to remove the growth medium. Cell pellets were resuspended in buffer containing 50 mM Tris. HCl (pH 7.5) 150 mM NaCl, 5 mM MgCl 2 and 10% glycerol and disrupted with 0.2-0.3 mm glass beads in a FastPrep tissue homogenizer in 5 cycles of 30 s at 6.5 m/s. Membranes were collected by centrifugation at 13,000 g. Briefly, membranes were trans-esterified and trimethylsilylated with methanol-hydrochloric acid solution and hexane.
Aliquots of 2 μl were automatically injected into a Hewlett-Packard Model 5880 gas chromatograph equipped with a Model 7672 A automatic injection system. Nonadecanoic acid (C19:0) was used as an internal standard for this experiment. Subsequently, the relative abundance of each fatty acid was calculated for cells grown in different conditions.

Scanning transmission electron microscopy (STEM)
To investigate cell structure changes in F. prausnitzii A2-165 treated with CP, STEM images were taken using Zeiss Supra55 SEM equipped with an external scan generator (ATLAS, Fibics, Canada).
Large area scans enabled the analysis of many bacteria within 1 data set. Sample preparation was based on a protocol described by Silva (Silva et al., 2014) with some modifications. A 5 ml overnight culture was treated for 1 h with different concentrations of CP. 2 ml of each sample were centrifuged at 2000 g for 5 min. Pellets were fixed using 2% glutaraldehyde/2% paraformaldehyde in 0.1 M sodiumcacodylate pH 7.3 and post-fixed in 1% osmium tetroxide/1.5% potassium ferrocyanide. They were then sequentially dehydrated in increasing concentrations of ethanol. After embedding in EPON, ultra-thin sections (80 nm) were cut and contrasted with 2% uranyl acetate in water followed by Reynolds' lead citrate. Large area scans were made using STEM and the thickness of the cell envelope of F. prausnitzii was measured for 25 cells per image using ImageJ (Version 1.51n; National Institutes of Health, USA).

Microbial fuel cell (MFC) experiments
Faecalibacterium prausnitzii A2-165 was grown for 16 h in YCFAG medium with or without CP or followed by 2-h incubation with 0.03 mg/ml CP. The MFC experiments were conducted as previously described by Khan (Khan, Browne, et al., 2012). Briefly, 38 ml culture with OD 600 of 0.8 broth were harvested at 2000 g for 15 min, washed and resuspended in potassium phosphate glucose buffer pH 7, and added to a 37°C anode chamber continuously flushed with N 2 .
40 ml 100 mM potassium phosphate buffer with 50 mM potassium ferricyanide was added to the cathode chamber. After running the device for 5 min, 0.003 g vitamin B 2 was added to the anode chamber. Data was collected after closing the circuit for 45 min using a CHI electrochemical instrument.