Method for assessing inter-individual variability in the carbohydrate utilization potential of gut microbiota ex vivo

Abstract

The microbial utilization of carbohydrates produces metabolic byproducts that are associated with health benefits, such as short-chain fatty acids (SCFAs). The rapid absorption of SCFAs by intestinal epithelial cells makes them challenging to study in vivo. Culturing methods address this limitation by providing controlled conditions that allow for monitoring substrate utilization and the production of fermentation byproducts. In this paper, we describe an easily implemented, cost-effective method for screening inter-individual differences in carbohydrate degradation potential among gut microbiota from different individuals using batch cultures in 96-deep well plates. In addition to culturing methods, we describe procedures for obtaining stool samples from human volunteers, as well as for processing and storing them. We demonstrate that complete or near-complete substrate utilization is typically reached within 24 hours of culturing in buffered media in 96-deep well plates, and inter-individual differences in substrate utilization rate and SCFA production are observable by 12 hours.

Graphical abstract

Specifications table

Subject area	Immunology and Microbiology.
More specific subject area	Human gut microbiome fermentations.
Name of your method	Method for assessing inter-individual variability in the metabolic potential of gut microbiota ex-vivo.
Name and reference of original method	This research builds on previously published work to culture gut microbial communities ex vivo, to study carbohydrate utilization [1,2].
Resource availability	If applicable, include links to the resources necessary to reproduce your method (e.g., equipment, data, software, hardware, reagents).

Background

The fermentation of carbohydrates by gut microbiota into short-chain fatty acids (SCFAs) has important implications for health, and gut microbial communities between different individuals vary widely in their carbohydrate utilization abilities [[3], [4], [5], [6], [7], [8]]. While the ability to degrade polysaccharides that are inaccessible to human degradative enzymes exists widely across gut microbial communities, the abundance of carbohydrate-degrading microbes and the diversity of utilization pathways results in heterogeneous responses to dietary interventions with non-digestible carbohydrates [3,9,10].

As the primary byproducts of carbohydrate fermentation, SCFAs are an important metric for evaluating inter-individual variability in carbohydrate utilization by gut microbial communities [3,5,11,12]; however, their rapid absorption by intestinal epithelial cells makes them challenging to study in vivo [13,14]. Therefore, culturing systems where synthetic microbial communities (in vitro) or communities directly from stool samples (ex vivo) are grown in media supplemented with carbohydrates are superior for assessing metabolic potential, allowing substrate utilization and metabolite production to be assessed in real time [2].

One challenge in gut microbial culturing systems is managing the decrease in pH that coincides with the production of SCFAs and other organic acids during fermentation processes. In the proximal colon, where most carbohydrates are fermented, the absorption of acidic byproducts and bicarbonate secretion maintains pH conditions between 5.5–6.5 [4,15], but within lab conditions, the decrease in pH can lead to culture death and incomplete substrate utilization. Bioreactors, such as batch or chemostat systems, offer dynamic pH control [2,16]; however, these systems can be costly and challenging to scale for the high-throughput studies needed to assess inter-individual variability.

In developing this method, we found that pH can be maintained over 48 hours at a biologically optimal range (5.5–7.5). This was achieved using buffered media supplemented with carbohydrates in a simple culturing setup, which allowed for complete substrate utilization. We validate this method with cultures supplemented with manno-oligosaccharides (MOS) and show that it is feasible to assess inter-individual variation in substrate utilization rates, SCFA production profiles, and community composition through 16S rRNA bacterial gene sequencing. We demonstrate that our culturing method can recover greater bacterial diversity than is detected in baseline stool samples alone. We also use this method to compare SCFA production profiles between individual microbial communities supplemented with MOS and other commercial prebiotics. In addition to culturing methods, we present a full workflow for obtaining stool samples from human participants, along with procedures for sample processing and storage. We also describe how this protocol can be used alongside other measures known to be associated with gut microbial composition, such as participant characteristics including anthropometric data and dietary intake, to examine how these factors contribute to carbohydrate utilization and SCFA production.

This article describes a single, scalable workflow for measuring carbohydrate utilization, taxonomic composition and metabolite production of gut microbial communities at high-throughput using a simple culturing setup within 96-deep well plates, alone or alongside participant characteristics such as anthropometric data and dietary intake.

Method details

• 1.
  Pre-study: Identify study endpoints and obtain ethics approval. This is a standard process for working with human biological samples.

  • 1.1.
    Define study inclusion/exclusion criteria. Set study inclusion and exclusion criteria to select research participants. Eligibility criteria may include not having taken antibiotics in the past 6 months and not currently on any medications or microbiome supplements such as probiotics and prebiotics. We validated this method in healthy participants using eligibility criteria provided in SI 1.
  • 1.2.
    Define microbiome study outcomes/endpoints. Microbiome endpoints may include taxonomic analyses (16S rRNA bacterial gene sequencing, shotgun metagenomic sequencing) or metabolite quantification. This method was validated using outcomes of 16S rRNA bacterial gene composition as well as concentration of carbohydrates (monomers and manno-oligosaccharide (MOS) oligomers) and SCFAs (acetic acid, propionic acid, and butyric acid), and lactic acid.
  • 1.3.
    Define additional study outcomes. Define additional study outcomes or endpoints aligning with research objectives and available study personnel. Additional participant characteristics that may be relevant for microbiome analyses include dietary, anthropometric, or demographic data. Methods of dietary assessment include food frequency questionnaires, 3- or 7-day food records, or 24-hour recalls. Facilitating proper data collection may require additional skills or training from the research team. Anthropometric data such as height and weight may be collected via participant self-reporting or collected during in-person screening.
  • 1.4.
    Develop appropriate documentation. Develop relevant study documents including a description of the research project for study participants, screening form, research consent form, and instructions for research participants on study expectations (including sample collection and drop-off). Some institutions allow for screening and consent to be obtained electronically through secure web applications.
  • 1.5.
    Obtain ethics approval. Obtain ethics approval from relevant institutional research ethics boards.
• 2.
  Recruit participants and collect samples.

  • 2.1.
    Recruit participants. Participants may be recruited through word of mouth, recruitment posters, and online advertising, among other approaches. Recruitment posters and advertisements may include a link to an electronic screening form to ensure eligibility. Additional study data, such as demographic and anthropometric data, can be collected during screening and destroyed if a participant is not eligible to participate. To protect participant privacy and ensure blinding of the research team involved in data analysis, eligible and enrolled participants should receive a unique, anonymous study identifier. At least one member of the research staff who is not involved in the data analysis should handle communication with participants to maintain blinding.
  • 2.2.
    Collect stool samples. Provide sample collection kits and education on sample collection to eligible research participants. Coordinate sample drop-off and allow adequate time (>1 day) for the research team to prepare for sample processing.

    • 2.2.1.
      Stool collection kits

      • •
        20 mL screw-top collection containers with spoon in lid (e.g., Sarstedt #80.734.001)
      • •
        Flushable specimen collection sheet
      • •
        Opaque bag for sample transport
      • •
        Instructions for participants for sample collection
• 3.
  Stool sample processing. Preparing media for use in stool homogenization and processing, process samples, and store at −80 °C. Samples should be transferred to anaerobic conditions and processed immediately, as viability of cells decrease rapidly with oxygen exposure [17]. Storing samples within glycerol reduces ice crystals and assists in cryopreservation [18].

  • 3.1.
    Materials and Reagents

    • •
      Media components

      • ○
        Brain Heart Infusion (BHI) broth powder
      • ○
        L-cysteine hydrochloride
    • •
      Glycerol
    • •
      MilliQ water
  • 3.2.
    Consumables

    • •
      Sterile syringe and syringe filters (0.22 µm)
    • •
      Sterile 1 mL pipette tips
    • •
      Sterile 2 mL microtubes
    • •
      Large (>40 cm) resealable plastic-bags (e.g. Ziploc)
    • •
      Disposable spatulas
    • •
      Homogenizer/masticator bags for paddle homogenizer
    • •
      50 mL Falcon tubes
    • •
      50 mL microbiological pipette and pipette filler
    • •
      Ice and ice bucket
    • •
      0.5–2 mL screw-top cryogenic tubes
    • •
      70% Ethanol
  • 3.3.
    Equipment

    • •
      Analytical balance
    • •
      Autoclave
    • •
      Paddle homogenizer
    • •
      Anaerobic chamber (80% N2–10% CO2 and 10% H2)
    • •
      Class II biosafety cabinet (BSC) or Bunsen burner
    • •
      Stirring stick or magnetic plate and stir rod
    • •
      Glass bottles (1 L)
    • •
      1 L beaker
    • •
      Pipette (P-1000)
    • •
      Serum bottles (200 mL)
    • •
      Scissors
  • 3.4.
    Media preparation (day before collection).

    • 3.4.1.
      Prepare 500 mL of BHI media according to instructions using ultrapure water and mix in a 1 L beaker using a stir rod or magnetic stir plate. Transfer media to a 1 L glass bottle.
    • 3.4.2.
      Prepare 50% glycerol by diluting glycerol with equal parts ultrapure water in a 1 L beaker. Volume can be adjusted depending on how many samples there are to process (∼100 mL per sample).
    • 3.4.3.
      Mix and transfer to individual serum bottles (∼100 mL per bottle). Use a separate glycerol stock for each sample to avoid contamination between samples.
    • 3.4.4.
      Heat sterilize BHI media and glycerol stock by autoclaving at 121C for 20 min. Allow to cool.
    • 3.4.5.
      Add l-cysteine: prepare 500 mg/mL stock of l-cysteine and filter-sterilize using a 0.22 µm syringe filter. Under sterile conditions (working in BSC or near Bunsen burner), pipette 0.5 mL into 500 mL cooled media (0.5 mL/L final concentration – 0.05% v/v).
    • 3.4.6.
      Reduce media by flushing with nitrogen gas or placing under anaerobic conditions for at least 8 h.
  • 3.5.
    Procedure for stool sample processing (day of collection).

    • 3.5.1.
      Weigh masticator bag and record results.
    • 3.5.2.
      Weigh resealable plastic bag (e.g., Ziploc) and record results.
    • 3.5.3.
      Thoroughly sterilize all surfaces within the anaerobic chamber before beginning lab work. Sterilize both the inside and the outside of the resealable plastic bag and masticator bag with ethanol and allow to dry completely. Sterilize all consumables before transferring into the anaerobic chamber. Prepare a waste collection bag within the chamber.
    • 3.5.4.
      Transfer the stool sample into the anaerobic chamber. Using a disposable spatula, transfer sample from the collection container into the masticator bag. Carefully fold the masticator bag, removing as much air as possible, then insert the folded bag containing the sample into the plastic zip-loc bag (minimize sample exposure to oxygen). Transfer the sample out of the anaerobic chamber through the airlock.
    • 3.5.5.
      Weigh the sample (which is within the masticator bag and the resealable plastic bag) on a calibrated analytical balance and record the results. Return the sample to the anaerobic chamber.
    • 3.5.6.
      Calculate the volume of media required to dilute the sample to 20% w/v. This will be the total weight (Section 3.5.5) minus the masticator bag weight (3.5.1) and the resealable plastic bag weight (3.5.2).

      $$\text{Volume}\text{required}\text{to}\text{dilute}20\%\mathrm{w}/\mathrm{v}\left(\text{mL}\right)=\text{total}\text{stool}\text{weight}\left(\mathrm{g}\right)÷0.2\mathrm{g}/\text{mL}.$$
    • 3.5.7.
      Open the plastic bag and masticator bag and use a microbiological pipette to pipette media into the masticator bag.
    • 3.5.8.
      Fold the masticator bag by first removing as much air as possible, then folding the top over a few times to prevent leakage. Fold the masticator bag and place back into the plastic bag. Completely seal the plastic bag and transfer back out through the airlock.
    • 3.5.9.
      Homogenize the sample in the paddle blender (3x for 10 s each). Remove the bag from the blender and complete the homogenization by hand, especially in the corners, for ∼1 min.
    • 3.5.10.
      Return the bagged stool back into the chamber through the airlock. Carefully open the resealable plastic bag and masticator bag. The sample may have leaked from the masticator bag into the plastic bag. Pour the sample from the masticator bag into the plastic bag and discard the masticator bag.
    • 3.5.11.
      Cut approximately 0.5–1 cm from one of the bag corners and carefully transfer the whole sample to 50 mL Falcon tubes.
    • 3.5.12.
      From the 50 mL tube, aliquot the sample into freezer-safe cryotubes along with equal volume of the 50% glycerol stock (0.2–0.5 mL each).

      • •
        Cutting the end of the pipette tip with sterilized scissors may be helpful as the sample contains solids.
      • •
        Periodically mix the sample in the 50 mL Falcon tube to prevent sample separation and ensure that the sample is as homogenous as possible across microtubes.
      • •
        Keep sample on ice while aliquoting.
    • 3.5.13.
      Transfer cryotubes out of anaerobic chamber and freeze at −80 °C until used in culturing experiments.
• 4.
  Culturing with prebiotics. Culture stool-derived microbial communities within complex media in 96-deep well plates. The media recipe for BHI and supplements was referenced from Surette et al. [19]. Adding reducing agents such as l-cysteine, as well as hemin and vitamin K, helps to support the growth of anaerobes [20,21].

  • 4.1.
    Materials and Reagents

    • •
      BHI broth powder
    • •
      L-cysteine hydrochloride (500 mg/mL)
    • •
      5mg/mL stock of hemin made up in 25 mM NaOH
    • •
      1:1 ratio of vitamin K:ethanol
    • •
      Sodium bicarbonate
    • •
      Stool-derived mixed bacterial cultures, frozen
    • •
      Phosphate-Buffered Saline (PBS) (pH 7.4), autoclaved or filter-sterilized
  • 4.2.
    Consumables

    • •
      Ice and ice bucket
    • •
      1 mL and 200 µL pipette tips, sterile
    • •
      1 mL microtubes
    • •
      Filter syringes and 0.2 µm membrane filters
    • •
      Breathable plate seals
    • •
      70% alcohol for decontamination
  • 4.3.
    Equipment

    • •
      Analytical balance
    • •
      Magnetic stir plate and stirrer
    • •
      Pipettes (P-1000, P-200), P-1000 multi-channel pipette
    • •
      Anaerobic chamber
    • •
      Thermomixer
    • •
      Vortexer
    • •
      Class II BSC or Bunsen burner
  • 4.4.
    Prepare media (day before)

    • 4.4.1.
      Prepare fresh BHI media (as described in 3.4.1, 3.4.4). After autoclaving, add supplements, working within a BSC:

      • •
        500 µL of 500 mg/mL stock of l-cysteine dissolved in water, filter sterilized, to 500 mL of media (final concentration of 0.5 g/L)
      • •
        1 mL of 5 mg/mL stock of hemin in 25 mM NaOH, filter sterilized, to 500 mL of media (final concentration of 10 mg/L)
      • •
        1 µL of 1:1 ratio of vitamin K:ethanol to 500 mL of media (final concentration of 1 mg/L)
      • •
        10.5 g sodium bicarbonate to a small amount of media, filter-sterilize, then add to stock.
    • 4.4.2.
      Prepare prebiotic stocks by dissolving dry prebiotics in ultrapure water to reach a final concentration of 100 mg/mL and filter-sterilize under sterile conditions (working within BSC or near Bunsen burner).
    • 4.4.3.
      Reduce oxygen in media and prebiotics by either flushing with nitrogen gas or leaving under anaerobic conditions for at least 8 h.
  • 4.5.
    Begin culturing experiment

    • 4.5.1.
      Thaw bacteria stocks on ice and transfer to the anaerobic chamber. Important: keep bacteria on ice as much as possible. Mix by pipetting up and down 10x or gently vortexing.
    • 4.5.2.
      Pipette bacteria culture, prebiotic stock (final concentration 25 mg/mL), and media into wells.

      • •
        Volumes for 1 mL total volume: 50 µL bacteria culture, 250 µL prebiotic stock, 700 µL media
      • •
        Volumes for 750 µL total volume: 37.5 µL bacteria culture, 187.5 µL prebiotic stock, 525 µL media
      • •
        Note: previous research from our lab suggests that 25 mg/mL best supports the growth of gut microbes in monocultures compared to other doses [22], but different concentrations for the selected prebiotic may be screened.
    • 4.5.3.
      Cover the well plate with a gas-permeable plate seal, ensuring that the seal is properly adhered with no gaps.
    • 4.5.4.
      Incubate the deep well plate in the thermomixer at 37 °C and 300 rpm for the duration of the culturing experiment (12–48 h).
  • 4.6.
    End culturing experiment

    • 4.6.1.
      Collect culturing aliquots and separate bacteria pellet from supernatant:

      • 4.6.1.1.
        Remove the entire contents from each well in the 96-deep well plate into labeled sterile tubes
      • 4.6.1.2.
        Pipette 250 µL autoclaved PBS (pH 7.4) into each well. Using a new pipette tip for each well, scrape the sides of the well with the pipette tip and wash the sides of the well with the PBS to remove any bacteria that might have adhered to the well (Note: this dilution must be accounted for in downstream analyses, e.g. quantification of metabolites in the supernatant).
      • 4.6.1.3.
        Remove tubes from the chamber, then handle on ice.
      • 4.6.1.4.
        Centrifuge tubes at 4 °C and 13,000 g for 10 min to pellet bacteria.
      • 4.6.1.5.
        Working within a BSC, remove the supernatant. If doing metabolite analysis, transfer supernatant to a separate sterile microtube and store at −80 °C.
      • 4.6.1.6.
        Store pellet at −80 °C.
    • 4.6.2.
      Sample preparation for DNA extraction:

      • 4.6.2.1.
        Perform DNA extraction with a Zymobiomics DNA Miniprep Kit (D4300) or equivalent kit (protocol described by manufacturer).
      • 4.6.2.2.
        Note: sequencing methods for determining taxonomic composition are not described here, and may differ depending on study objectives. For the validation of this method, sequencing and bioinformatic processing were outsourced to the Centre for the Analysis of Genome Evolution and Function (CAGEF), University of Toronto. Extensive resources for sequencing methods for microbiomes are available on the CAGEF website [21].
    • 4.6.3.
      Sample preparation for metabolite analysis

      • 4.6.3.1.
        Thaw the sample from the freezer.
      • 4.6.3.2.
        For HPLC or LCMS: dilute the sample 1:1 in mobile phase and filter-sterilize with 0.2 µm filter.
      • 4.6.3.3.
        For the method validation, carbohydrate quantification was performed using the methods described in Asbury et al. [23] and SCFA and lactate quantification were performed using methods described elsewhere [24].
• 5.
  Data Analysis

  • 5.1.
    Quantify inter-individual variability in microbiome endpoints. Inter-individual variability in the metabolic potential of gut microbial communities can be determined by evaluating metabolite production (functional ability) or through differences in the composition of the microbial community (taxonomically).
  • 5.2.
    Statistical and computational analyses. Once microbiome endpoints have been obtained, various analytical techniques can be used. Principal Component Analysis (PCA) is a dimensionality reduction method commonly used in microbiome studies due to the high-dimensional nature of taxonomic data. Clustering methods can be used to group samples according to metrics such as metabolite production or substrate utilization. Predictive modeling and other statistical analyses can further help identify microbiome features that discriminate between individuals [25] (Fig. 1).

Fig. 1.

Graphical abstract for method. Produced with BioRender.

Method validation

We developed this method using stool culture samples from individual research participants that were selected from a larger group recruited using the methods described herein. First, we determined the optimal media composition by comparing the metabolite production and taxonomic composition of stool cultures using the methods described, using different media compositions supplemented with manno-oligosaccharides (MOS) (Fig. 2). Sequencing metrics for determining bacterial taxonomic composition with 16S rRNA gene sequencing indicated high quality sequencing across samples, and detailed metrics including sequencing depth and rarefaction curves are included in Supplementary S2. Analyses of taxa recovery across culturing conditions are included in Supplementary S3. Brain-Heart Infusion (BHI) media is well supported for its ability to grow a wide range of gut microorganisms, especially Bacteroides, which is an important genus for carbohydrate degradation [19,26]. Lau et al. showed that a base media of BHI with supplements effectively supported the growth of gut microbes during culturing, even those thought to be unculturable [1]. While existing literature demonstrated that supplementing media with fetal bovine serum or mucin supports microbial growth [1], we found that the SCFA production and taxonomic composition at phylum and genus levels were comparable between fetal bovine serum or mucin-supplemented and non-supplemented media. Further, cultures grown within BHI media were associated with greater richness and alpha-diversity than other media conditions (Supplementary S2 and S3). Therefore, the base media of BHI broth supplemented with 0.5 g/L-cysteine hydrochloride, 10 mg/L hemin and 1 mg/L vitamin K was selected for this culturing method.

Fig. 2.

Impact of media conditions on SCFA production (n = 3) and taxonomic composition (n = 2) of gut microbial cultures. Media tested include Brain Heart Infusion (BHI), BHI with 0.04% mucin (BHI-FBS), BHI with 3.5% fetal bovine serum (BHI-FBS), and Gifu Anaerobic Broth (GAM). All media conditions were supplemented with 250 mM NaHCO₃^- and BHI media conditions were supplemented with 0.5 g/L-cysteine hydrochloride, 10 mg/L hemin and 1 mg/L vitamin K. A) SCFA production in BHI with and without prebiotic (25 mg/mL MOS) compared to SCFA production in other media types. Data points represent an average of two technical replicates, with error bars showing standard deviation. B) Taxonomic composition at the phylum level across all media types supplemented with 25 mg/mL MOS, including only taxa with >0.5% relative abundance. C) Taxonomic composition at the genus level.

Next, a range of bicarbonate buffers was tested for their ability to maintain pH within an optimal range of 5.5–7.5. Stool cultures were grown anaerobically using BHI media supplemented with MOS. We demonstrate that a starting buffer concentration of 250 mM, and in some cases, 100 mM, is sufficient to maintain the pH within this range over 48 hours of carbohydrate fermentation (Fig. 3). For these reasons, the final media composition selected was BHI media with supplements and 250 mM bicarbonate.

Fig. 3.

pH time-dependence with different concentrations of NaHCO₃- buffer across gut microbial cultures (n = 3) supplemented with different carbon substrates (media alone, 25 mg/mL FOS, 25 mg/mL MOS). All buffer conditions completed in BHI media with 0.5 g/L-cysteine hydrochloride, 10 mg/L hemin and 1 mg/L vitamin K. Data points represent an average of two technical replicates, with error bars showing standard deviation.

We validated that this method can be used to assess inter-individual variability in substrate utilization and metabolite production. Independent stool cultures were supplemented with fermentable MOS, and the concentrations of metabolites in the supernatant were quantified at various timepoints using HPLC. We demonstrated that our culturing conditions achieves complete or near-complete substrate utilization over 24 hours, as evidenced by the concentration of MOS oligomers and monomers in the culturing supernatant (Fig. 4A). Inter-individual variability in metabolite production and substrate utilization can be observed (Fig. 4B-C). In addition to the production of SCFAs, we also assessed the production of lactic acid, which has been shown to vary meaningfully between individuals due to the amount and types of lactate-utilizing bacteria [27]. In our study, we found a high amount of variability in the production of all acids, with the highest inter-individual variability in the production of lactic acid. Notably, this work demonstrates the importance of sampling at multiple time points: the greatest variability in substrate utilization is observed at the 8- and 12-h time points. While trends in acetic acid and propionic acid are seen from the first time-point, variability is only observed after 8 hours for lactic acid, and after 24 hours for butyric acid. For some cases such as for high-throughput studies where only one time point is feasible, we recommend a time point of 8 or 12 hours if variability in carbohydrate utilization is the primary endpoint, and 24 hours where variability in SCFA production is the primary endpoint.

Fig. 4.

Metabolites in culturing supernatant at baseline and over 24 h of culturing in stool cultures (n = 5) participants. Media for all conditions was BHI media with 0.5 g/L-cysteine hydrochloride, 10 mg/L hemin, 1 mg/L vitamin K and 250 mM NaHCO₃-. Data points represent an average of two technical replicates, with error bars showing standard deviation. (A) Acetic acid, propionic acid, butyric acid and lactic acid production within 5 distinct gut bacterial cultures. (B) Utilization of oligomers (mannobiose, mannotriose, mannotetraose, and mannohexaose) and mannose within 5 distinct gut bacterial cultures. (C) Principal Component Analysis (PCA) biplot showing PC1 and PC2 of acid production and carbohydrate utilization variables, with loadings demonstrating variable contributions. Oligomer variables represent % utilization and acid variables represent production. PC1 was primarily associated with carbohydrate utilization variables (mannose, cellobiose, mannotetraose, glucose, and mannobiose), whereas PC2 was primarily associated with fermentation metabolites (butyric, acetic, propionic, and lactic acids) as well as mannotriose.

Finally, we validated that this method can be used to screen the inter-individual variability in SCFA production between different prebiotic substrates. Independent stool cultures were supplemented with 25 mg/mL of MOS, fructo-oligosaccharides (FOS) or xylo-oligosaccharides (XOS), as well as glucose (control). Differences in SCFA production were observed between different stool cultures as well as between different carbohydrate substrates within the same individual (Fig. 5). Notably, the highest concentrations of lactic acid were observed for cultures supplemented with glucose and FOS. During the pH measurement experiments, cultures supplemented with FOS exhibited the greatest pH drop across all buffer concentrations and were only controlled within the target pH range with the highest buffer concentration (Fig. 3). The acidic pH observed in the FOS-supplemented culture supernatant may be related to the increase in lactic acid, a stronger acid than the short-chain fatty acids. This demonstrates that cultures grown with certain substrates may achieve adequate pH control with lower concentrations of NaHCO₃^-, while others may require higher concentrations due to the accumulation of acidic fermentation byproducts.

Fig. 5.

(A) Acetic acid, propionic acid, butyric acid and lactic acid production within 5 distinct gut bacterial cultures using culturing method with different prebiotic substrates supplemented in the media (25 mg/mL), including glucose (positive control), fructo-oligosaccharides (FOS), xylo-oligosaccharides (XOS) and manno-oligosaccharides (MOS). Media for all conditions was BHI media with 0.5 g/L-cysteine hydrochloride, 10 mg/L hemin, 1 mg/L vitamin K and 250 mM NaHCO3-. Data points represent an average of two technical replicates, with error bars showing standard deviation. Significance measured with one-way ANOVA, with *** indicating p ≤ 0.001. (B) Principal Component Analysis (PCA) biplot showing PC1 and PC2 of acid production variables by substrate/prebiotic, with loadings demonstrating variable contributions (data points from A, as well as t=0 time points, circled). (C) Taxonomic composition at the phylum level of a subset of cultures (S002 and S003) after culturing with different prebiotics.

A detailed breakdown of taxa enriched and depleted across culturing conditions is provided in Supplementary data S3. Interestingly, our culturing method yielded a greater genus-level richness than what was seen in the baseline sample of one culture: 117 distinct genera were detected after culturing S003 in BHI medium without prebiotics, compared to 102 distinct genera found within the uncultured S003 stool sample. Among the genera detected in the cultured sample was Bifidobacterium and many members of Bacillota (formerly Firmicutes) including Bacillus, Clostridium and Eubacterium. This finding provides further rationale for culturing, as culturing in the proper media conditions can enrich bacteria that are present within samples at a low or undetectable abundance.

Some taxa detected at baseline were not recovered under any culturing condition. These lost taxa were generally present in low abundance (mean <1% relative abundance across baseline samples), except for Alloprevotella and Prevotella, which had a mean relative abundance in the baseline samples of 3.1% and 2.5%. Nevertheless, culturing recovered taxa representing 92.6% of the cumulative relative abundance at baseline. Across all culturing conditions, 189 distinct genera were identified with 16S rRNA bacterial gene sequencing. Of these, 158 were detected at baseline. Culturing maintained 134 of the 158 genus-level taxa found at baseline (85%), and a further 31 genera were enriched during culturing.

Overall, this work demonstrates an easily implemented, high-throughput method to screen for inter-individual variability in gut microbial carbohydrate utilization, taxonomic composition, and SCFA production, while maintaining culture pH and obtaining complete, time-dependent substrate utilization.

Limitations

This work describes a method to assess carbohydrate/prebiotic utilization and metabolite production at a community level within 96-well plates, using bicarbonate buffers to maintain the pH of cultures within a biologically acceptable range. This method was developed using stool samples from five healthy individuals, a sample size that was selected to demonstrate technical feasibility rather than to capture the full extent of inter-individual variability in carbohydrate utilization in human gut microbial communities. Further, this method has not been evaluated using stool samples from individuals with potential gut microbiome perturbations or dysbiosis, including those taking antibiotics or probiotics, and additional validation may be required for these sample types. Nonetheless, we expect that this framework can be used under these conditions albeit possibly with a greater number of samples required to assess differences.

Research objectives assessing the utilization of different carbohydrates or cross-feeding by different metabolites may require different media compositions. Further, some taxa were not recovered by our culturing method. Research objectives for targeting specific taxa such as fastidious or “hard-to-culture” microorganisms may require additional adjustments to the media composition and culturing conditions. Finally, little information is available on the impact, if any, of high concentrations of bicarbonate buffers on the viability of microorganisms. For example, high buffer concentrations may lead to osmotic stress on bacterial cells. While this was not explored in our study, we observed complete or near-complete substrate utilization over 24 hours, indicating that the culture conditions supported active microbial metabolism. Finally, as with all ex vivo culture systems, this method is limited in its generalizability to in vivo conditions.

Nonetheless, some advantages of ex vivo systems include the ability to directly monitor fermentation products and capture time-dependent changes in microbial taxa that would otherwise be inaccessible in vivo. Further, as demonstrated in our method validation, culturing can promote the growth of taxa that are undetectable in stool samples at baseline, providing a more comprehensive picture of taxonomic composition and metabolic potential. This method offers a pathway to screen microbial communities for their carbohydrate utilization capacity and to assess differences in metabolic potential between individuals. We envision that this simple and scalable culturing system can be readily adapted to explore the utilization of any soluble carbohydrate substrate, facilitating greater mechanistic insight into inter-individual variability in carbohydrate utilization by human gut microbial communities.

Ethics statements

Informed consent was obtained from all research participants. Ethics approval for the research project was obtained from the University of Toronto Research Ethics Board (REB) (#00044584).

CRediT authorship contribution statement

Rachel E. Asbury: Conceptualization, Methodology, Validation, Writing – original draft. Taryn A. Hoffman: Validation, Data curation, Writing – review & editing. Cathy Chu: Resources, Data curation. Radhakrishnan Mahadevan: Supervision, Writing – review & editing. Bradley A. Saville: Supervision, Writing – review & editing, Funding acquisition.

Declaration of interests

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix. Supplementary materials

S1: Inclusion/exclusion criteria

S2: Sequencing methods

S3: Taxa enrichment analysis

Data availability

Data will be made available on request.