Human Milk Oligosaccharides Exhibit Biofilm Eradication Activity Against Matured Biofilms Formed by Different Pathogen Species

Isolation of the Carbohydrate Fraction From Human Milk

The milk samples were polled after thawing, and skimmed milk was obtained by centrifugation at 3,000 g and 4°C for 20 min in two repetitions. The proteins were precipitated by the addition of 96% ethanol to the supernatant (1:1) and incubation at 4°C overnight followed by centrifugation at 3,000 g and 4°C for 30 min. The ethanol in the supernatant that contained an enriched carbohydrate fraction was evaporated by a rotary evaporator at approximately 80°C (Laborota 4001, Heidolph Instruments, Schwabach, Germany). The total saccharides hereinafter referred to as HMOs were finally freeze-dried, and the powder was stored at 4°C. Based on the preliminary works, it can be expected that after the extraction, the oligosaccharide fraction contains primary lactose and approximately 7–21% of HMOs (Bode, 2012).

The lactose was enzymatically digested by adding 100 U of β-D-galactoside galactohydrolase (lactase F) (Sigma-Aldrich, St. Louis, United States) to 100 mg of lyophilized HMOs resolved in 1 ml of deionized water and subsequently incubating overnight at 37°C. The digested HMOs were freeze-dried thereafter and stored at 4°C.

Bacterial Strains

Different clinical isolates (provided by Jena University Hospital (JUH)) and laboratory standard strains (American Type Culture Collection (ATCC) and institutional stocks) were used in this study (Table 1). Species identification and antimicrobial resistance testing of the clinical isolates were routinely carried out using VITEK 2 (bioMèrieux, Marcy-l’Étoile, France). The minimal inhibitory concentrations (MICs) of the antibiotics for the laboratory strains were determined by the microdilution technique in accordance with European Antimicrobial Susceptibility Testing Committee (EUCAST) standards [ISO 20776-1:2019] by using a cation-adjusted Mueller Hinton (MH) broth. All isolates were stored in 10% (V/V) glycerol stocks at −80°C and freshly struck on blood agar plates (BD Diagnostics, Heidelberg, Germany) and grown overnight at 37°C before use and adjusted to 0.5 McFarland in an MH medium (Roth GmbH, Karlsruhe, Germany).

Human Milk Oligosaccharide Activity Against Planktonic Bacteria

From fresh agar plates, a cell suspension was adjusted to 0.5 McFarland in fresh MH broth and further diluted to 1:100 (correlates to approximately 10⁵–10⁶ CFU/ml). One hundred microliters of the bacterial suspensions was mixed with 100 μl of HMOs or lactose per well in 96-well plates (Greiner Bio-One GmbH, Frickenhausen, Germany) to obtain final concentrations of 0, 5, 10, 20, 50, and 100 mg/ml of HMOs in MH broth or 0, 4, 8, 16, 40, and 80 mg/ml of lactose and incubated at 37°C for 24 h without shaking. As control, the biofilms were treated in the same manner by MH broth only (0 mg/ml). The supernatant was serially diluted in a logarithmic manner up to 10^–8 in sterile saline (9 g/L NaCl). Subsequently, 10 μl of the dilutions was dropped on non-selective MH-agar plates and incubated overnight at 37°C. The colony-forming units (CFU) per ml were assessed from two to three countable droplets. Each experiment was independently performed in triplicate.

Determination of the Antibiofilm Activity of the Human Milk Oligosaccharides

For biofilm formation, overnight cultures were diluted to 0.5 McFarland (correlates to 10⁷–10⁸ CFU/ml) in fresh MH broth, and 200 μl of the bacterial suspensions was applied per well into 96-well plates and incubated for 48 h without shaking at 37°C. The supernatants were removed, and the biofilms were carefully washed with 200 μl sterile saline per well. The biofilms were treated for 24 h with 200 μl per well of serially diluted HMOs (5, 10, 20, 50, and 100 mg/ml) or lactose (4, 8, 16, 40, and 80 mg/ml) or 1:2 diluted HMOs in MH broth. As control, the biofilms were treated in the same manner by MH broth only (0 mg/ml). The concentration was not assessed after size-exclusion chromatography (SEC), see below). After treatment, the supernatants were removed, and the biofilms were carefully washed twice with 200 μl sterile saline. The biofilms were resuspended in 200 μl sterile saline by scraping them from the cavities and vortexing. After a serial dilution in a logarithmic manner up to 10^–8, 10 μl of the dilutions were dropped on MH-agar plates, and the CFU/ml was determined after overnight incubation at 37°C from the countable droplets. Each experiment was independently performed in triplicate.

Live/Dead Staining of the Biofilms and Confocal Laser Scanning Microscopy Image Acquisition

These experiments were performed only with the selected strains K. pneumoniae ATCC 700603, A. baumannii IIMK450, P. aeruginosa PA 03, B. cenocepacia BC 1, S. aureus ATCC 29213, E. faecalis ATCC 29212, and E. faecium Bk 013000. Biofilm formation was performed as previously described but in lumox™ 96-well plates for microscopic analysis (Greiner Bio-One GmbH, Frickenhausen, Germany) in duplicate. The supernatants were removed, and the biofilms were carefully washed with 200 μl sterile saline per well. The biofilms were treated with 200 μl of the HMOs (0, 5, 10, and 20 mg/ml in MH broth) for 24 h. After treatment, the supernatants were removed, and the biofilms were washed twice with 200 μl sterile saline and stained with the LIVE/DEAD BacLight Bacterial Viability Kit for microscopy (Life Technologies GmbH, Darmstadt, Germany) according to the manufacturer’s protocol. Stained biofilms were analyzed under vital conditions using the inverse CLSM LSM780 and the × 40 air objective (both Carl Zeiss AG, Germany) at 488 nm excitation by the argon laser line. An area of approximately 200 μm (X) × 200 μm (Y) was screened in 1 μm Z-intervals (Z-stack) at green (emission 522 nm) and red (emission 635 nm) channels. The pinhole was adjusted to approximately 1 μm.

Size-Exclusion Chromatography Fractionation of the Human Milk Oligosaccharides

The ÄKTA chromatography system (GE Healthcare, Chicago, United States) was used for the separation of the saccharides by SEC. A P-2 Bio-Gel (Bio-Rad Laboratories, Hercules, United States) column (5 × 175 mm) was manually prepared according to the manufacturer’s instruction. The flow rate was set to 0.1 ml/min, and deionized and degassed water was used for elution. Commercially available D-glucose and D-lactose monohydrate (both Merck KGaA Darmstadt, Germany), as well as the pentose verbascose (O-VER) (Megazyme Inc., Bray, Ireland), were used as standards. The saccharides were prepared as 20 mg/ml solutions in deionized water, and 200 μl was used for SEC separation. The lactase F-digested HMOs were resuspended in 100 μl deionized water and separated. Fractions of 200 μl were collected.

Thin-Layer Chromatography

The chromatographic fractions were analyzed on TLC silica gel 60 F254 plates (Supelco, Bellefonte, United States). Briefly, 2.5 μl (in 0.5 steps) of fractions F6–F16 and 1 μl each of glucose and lactose (each 20 mg/ml) standards were applied to a silica gel plate. After drying at room temperature, the plate was set into the TLC chamber pre-equilibrated with the liquid phase, a mixture of butanol–isopropanol–water in the volume ratio 3:12:6. The chromatography was performed for 1.5 h, and the plates were dried under the chemical hood overnight. Thereafter, the plates were briefly rinsed in orcinol staining solution (80 mg orcinol monohydrate in 160 ml acetone and 8 ml concentrated sulfuric acid) and then baked at 100°C to stain the sugars.

Orbitrap Measurements (Ultra-High-Performance Liquid Chromatography/High-Resolution Mass Spectrometry)

Ultra-High-Performance Liquid Chromatography (UHPLC) coupled with High-Resolution Mass Spectrometry (HRMS) was carried out using the UltiMate UHPLC system (all Thermo Fischer Scientific, Bremen, Germany), equipped with an UltiMate HPG-3400 RS binary pump, WPS-3000 autosampler (set to 10°C) with a 25 μl injection syringe, and a 100 μl sample loop. The Hypercarb^® column (100 × 2.1 mm; 3 μm) (Thermo Fischer Scientific) was kept at 25°C within the column compartment TCC-3200. Eluent A was water with 5 mM NH₄OAc buffer (adjusted to pH 9.6) supplemented with 2% acetonitrile. Eluent B was 80% acetonitrile in 5 mM NH₄OAc buffer (adjusted to pH 9.6). Chromatographic separation was performed by applying a constant flow rate of 0.22 ml/min and a two-step gradient: min 0.2–5 from 0 to 10% eluent B, min 5–20–100% eluent B, min 20–22 constant at 100% eluent B. The HMO fractions were used in a native non-derivatized form. Column recovery was performed by quickly changing (30 s) to 100% eluent A that was run for an additional 3.5 min.

Mass spectra were recorded with a Q Exactive Plus Orbitrap (Thermo Fischer Scientific) MS coupled to a heated electrospray source (HESI). Column flow was switched at 0.5 min from waste to the MS and at 23.5 min again back to the waste, to prevent source contamination. For monitoring, a full-scan mode was selected with the following parameters: polarity: positive; scan range: 200–2,000 m/z; resolution: 280,000; AGC target: 3 × 10⁶; and maximum IT: 100 ms. Additionally, a data-dependent MS² mode was executed: scan range: 200–2,000 m/z; resolution: 35,000; AGC target: 1 × 10⁵; maximum IT: 500 ms; loop count: 2; TopN: 2; isolation window: 0.5 m/z; no fixed first mass; stepped NCE 15, 30, and 45; exclude isotopes: on; and dynamic exclusion: 10 s. General settings were as follows: sheath gas flow rate: 60; auxiliary gas flow rate: 20; sweep gas flow rate: 5; spray voltage: 2.5 kV; capillary temperature: 360°C; S-lens RF level: 50; vaporizer temperature: auxiliary gas heater temperature: 400°C; and acquisition time frame: 0–24 min.

Fucosyllactose and Other Compounds

Two isomeric forms of Fucosyllactose (FL) (2′-FL and 3′-FL) were purchased in lyophilized form from Biosynth Carbosynth (Thal, Switzerland). The vials were completely resuspended in 1 ml deionized water to obtain a 10 mg/ml stock solution and stored at 4°C.

Other compounds (if not specified) were purchased from Carl Roth GmbH or Sigma Aldrich.

Effects Against Mature Biofilm

The effect of the pooled HMOs on mature biofilms was investigated by treating 48 h-grown biofilms for a further 24 h with different HMO or lactose concentrations. Here, too, it was assumed that lactose makes up 80% of the total HMOs.

In general, the HMOs showed almost no activity against biofilms of Gram-negative species (Figure 3). A slightly reduced number of viable cells was observed at concentrations below 20 mg/ml in Gram-negative species that, however, did not reach 1-log-level reduction. The effects on B. cenocepacia showed more strain-specific variance that might result from the higher genetic variability within this species. On A. baumannii, the lactose showed some bactericidal and concentration-dependent effects on the biofilms, reducing the viable cells by 2-log magnitudes (Figure 3B), which might be the result of the unfavorable carbon source for this non-fermenting species. On the other hand, lactose showed no specific effects on P. aeruginosa or B. cenocepacia (Figures 3C,D), both representing also non-fermenting species. Thus, the observed effect of lactose on A. baumannii was elusive.

Open in a separate window

FIGURE 3

Determination of the viable cell fraction (as CFUs/ml) in biofilms of laboratory and clinical strains of different Gram-negative species after treatment with the total HMOs and lactose in corresponding concentrations (assuming that lactose makes up to 80% of the HMOs) for 24 h. (A) K. pneumoniae, (B) A. baumannii, (C) P. aeruginosa, (D) B. cenocepacia.

Like the planktonic-grown Gram-positive bacteria, lactose did not considerably influence the viable cell numbers, while the HMOs showed a visible reduction of the viable cells of the biofilms by up to a few log levels (Figures 4A–C), particularly in S. aureus and E. faecalis (Figures 4A,B, respectively). Again, isolate-specific variations in sensitivity were observed, although sensitivity could not be derived directly from the planktonic experiments. Thus, in addition to Bk 848, the biofilms of E. faecalis ATCC 29212 and Bk 1653 were also reduced by nearly 3-log levels.

Open in a separate window

FIGURE 4

Determination of the viable cell fraction (as CFUs/ml) in biofilms of laboratory standard and clinical strains of different Gram-positive species. (A–C) After treatment with the total HMOs and lactose in corresponding concentrations (assuming that lactose makes up 80% of the HMOs) for 24 h. (A) S. aureus, (B) E. faecalis, (C) E. faecium. (D) Activities of the FL isomers 2′-FL and 3′-FL against biofilms of S. aureus and E. faecalis expressed as CFU/ml after treatment. Data of three independent experiments performed in triplicate are given as mean and standard deviation (SD).

At HMO concentrations higher than 20 mg/ml, the bactericidal effect was abolished in almost all Gram-positive species and isolates. It is likely that at higher saccharide concentrations, lactose and some of the oligosaccharides might enhance growth as nutrients.

Biofilm Activity in Microscopic Evaluation

The effect of the HMOs (at 5, 10, and 20 mg/ml) on biofilms was also analyzed by CLSM after live (green)/dead (red) staining and compared to that on MH-only-treated (0 mg/ml) biofilms. Accordingly, in the CFU assessment, the Gram-negative bacteria (K. pneumoniae ATCC 700603, A. baumannii IIMK450, P. aeruginosa PA 03, and B. cenocepacia BC 1) showed less sensitivity to the HMOs (Figure 5A). In general, the biofilms of the Gram-negative species showed a higher number of red-stained cells also in the controls (treatment only with saline); thus, it was more difficult to interpret the number of dead cells by purely visual evaluation. The number of red-strained cells seemed to increase while the biofilm size (biomass) decreased in an HMO-concentration-dependent manner in all tested Gram-negative species. This reduced biofilm size agrees with the observed slightly reduced CFU/ml at HMO concentrations below 20–50 mg/ml, but because no strong reduction of the CFU/ml was determined, the red staining seemed not to correspond to cell death in the Gram-negative species. There might be two reasons for this observation: (i) the HMOs might increase the susceptibility to the red dye by interaction with membranes, increasing their permeability; (ii) or the external DNA content of the Gram-negative species led to overlapping staining of the outer layers of the bacteria.

Open in a separate window

FIGURE 5

Exemplary 3D images of the CLSM of untreated (0 mg/ml) and HMO-treated (5, 20, and 50 mg/ml) biofilms. (A) Gram-negative biofilms of K. pneumoniae ATCC 700603, A. baumannii IIMK450, P. aeruginosa PA 03, and B. cenocepacia BC 1. (B) Gram-positive biofilms of S. aureus ATCC 29213, E. faecalis ATCC 29212, and E. faecium Bk 013000. Sections of 212 μm × 212 μm (X/Y axes) were screened, and the Z-size varied depending on the biofilm thickness. Ticks on the Z-axis for K. pneumoniae indicate 4 μm, and for the other strains 5 μm. The biofilms were fluorescently stained by SYTO 9 (live, green) and propidium iodide (dead, red) before imaging.

In the Gram-positive species (S. aureus ATCC 29213, E. faecalis ATCC 29212, and E. faecium Bk 013000), the HMO effect on the biofilms was more pronounced (Figure 5B). The S. aureus controls were stained intensively green, and the cells turned red in an HMO-concentration-dependent manner while the biofilm size was not visibly reduced. Considering the strong reduction in the CFU/ml, the red staining indicated cell death in S. aureus biofilms. Similarly, the red cell fraction increased in E. faecium and E. faecalis biofilms up to an HMO concentration of 20 mg/ml, but at 50 mg/ml, the biofilms became thin and holey. This indicates that the HMOs not only killed the cells but also led to cell lysis of the enterococci. However, this was not proved by additional experiments and might result from increased release of the cells into the planktonic phase due to matrix modulation by HMOs.

Evaluation of the Active Human Milk Oligosaccharide Fractions

To decide, after SEC, which fraction should be investigated on the biofilms, the standard oligosaccharides were fractioned by SEC, and their presence in specific fractions was investigated by TLC and staining. The glucose (the monosaccharide) was identified in fractions 12–14 (maximum in 13) and lactose (disaccharide) in fractions 11 –13 (maximum in 12). The pentose O-VER was identified in fractions 10 and 11 (maximum in 10).

From lactase F-digested HMOs, 100 mg was separated by SEC per run. Fractions 6–14 (each 200 μl) were diluted 1:2 in deionized water and tested for their antibiofilm activity against matured biofilms of S. aureus ATCC 29213 for 24 h. MH broth was used as control; additionally, 100 mg/ml glucose and lactose was applied to compare the effects of possible residues. The effect of galactose was not tested. Before washing of the biofilms and CFU/ml assessment after treatment, the cells released from the biofilms into the planktonic phase were harvested and serially diluted, and the CFUs/ml were assessed as well.

Compared to glucose and lactose, fraction 12 showed the highest antibiofilm activity, reducing the viable bacteria of the biofilms by up to 2-log magnitudes in a significant (P = 0.002) manner (Figure 6A). Compared to lactose, fractions 11 and 13 also showed significant reduction (P = 0.027 and 0.044, respectively). Since monosaccharides and disaccharides were eluted in this fraction in the preliminary tests, this indicates that the active ingredients must have a low molecular weight with few saccharide moieties. No significant effects of the fractionated HMOs were observed in the cells of the planktonic phase (Figure 6B).

Open in a separate window

FIGURE 6

Analysis of UHPLC/HRMS fractionated HMOs. Activities of the HMO fractions against biofilms (A) and planktonic bacteria (B) of S. aureus expressed as CFU/ml after treatment. The X-axis indicates the different treatments. C, control, no treatment (red); G, glucose; L, lactose (both yellow). The SEC fractions used for the treatment are indicated with blue bars according to their number. NF, non-fractionated HMOs (gray). = = = The boxes represent the range of the data (max to min), the bars show the standard deviation (SD), the lines in the boxes indicate the median, and the crosses indicate the averaged values. (C) UHPLC sensorgrams of the HMO fractions F8–F14. The sensorgrams were normalized to the control (phosphate-buffered saline). (D) Relative abundance of the oligosaccharides detected in the SEC fractions F8–F14 by HRMS. Only signals above a relative abundance of 103 were shown. The saccharides or saccharide moieties were calculated based on the m/z signals (mass/charge) of the H⁺, Na⁺, or K⁺ adducts; the stereochemistry was not assessed. The compounds are given as moieties as follows: F, fucose; G, glucose or galactose; NAcG, N-acetyl glucosamine; NAcN, N-acetylneuraminic acid (sialic acid); and L, lactose; * indicates possible cleavage products.

Fractions 8–14 were further analyzed by UHPLC/HRMS to assess the most abundant saccharides in these fractions. All identified compounds are listed in the Supplementary Material. In the UHPLC, a peak that was unique in fractions 11–13, but highest in fraction 12 (Figure 6C), was visible at 8.02 min (retention time), which might correspond to the active compound as identified in the biofilm assays. However, in fraction 12, only galactosyl lactose (GL) as a possible active compound was identified but at low relative abundance (3.4 × 10⁵) compared to other compounds (Figure 6D). Lactose was generally the second-highest abundant compound and was found in all fractions. In fractions 13–14, lactose might result from still incomplete digestion but with decreasing fraction number (higher molecular weights); the source must be rather a defragmentation of the compounds by the ionization energy of the lactase F digestion. Lactase F hydrolyzes the terminal non-reductive galactose residues of β-D-galactosides, and thus, some of the HMOs might be possible substrates. It could be previously shown that lactase F partially digested lacto-N-tetraoses (LNT) but not the fucosylated or sialylated HMOs resulting in lacto-N-triose as a new digestion product (Gnoth et al., 2000). Thus, the observed trioses [LG* and L(NAcN)*] in the present study might correspond to lactase digestion of LNT [L(NAcG)G in Figure 6D].

FL isomers showed the highest abundance and were identified in fractions 10–12 (with > 10⁷ being highest in fraction 11). FL might be fucosylated either at 2′-C of galactose or at 3′-C of glucose. Difucosyllactose (LFF) was also present in lightly elevated concentrations in fraction 10 (8.6 × 10⁵) to 12 (1.9 × 10⁵).

No sialylated HMOs were identified in the tested fractions; however, an N-acetylneuraminic acid residue was identified in all fractions except fractions 12 and 14 at very low abundance; thus, it cannot be excluded that such compounds were present in very low concentrations but were defragmented during the analysis.

Capacious thanks to the student, Yuliya Skril, who performed some of the biofilm assays with the HMOs and to Lisa Jasef and Margot Voigt, who assisted in the experiments.