FEMS Microbiology Letters, 368, 2021, fnab042

doi: 10.1093/femsle/fnab042
Advance Access Publication Date: 17 April 2021
Research Letter

R E S E A R C H L E T T E R – Food Microbiology

Acidic pH enhances butyrate production from pectin

by faecal microbiota

Downloaded from https://academic.oup.com/femsle/article/368/7/fnab042/6232156 by guest on 05 July 2021

Grete Raba1,† , Signe Adamberg1 and Kaarel Adamberg1,2, *
1
Department of Chemistry and Biotechnology, Tallinn University of Technology, Akadeemia tee 15, Tallinn,
Estonia and 2 Center of Food and Fermentation Technologies, Akadeemia tee 15A, Tallinn, Estonia
∗
Corresponding author: Department of Chemistry and Biotechnology, Tallinn University of Technology, Akadeemia tee 15, 12618, Tallinn, Estonia.
Tel: +372 620 2955; E-mail: [email protected]
One sentence summary: pH is a crucial trigger in the development and metabolic activity of faecal microbial consortia.
Editor: Gisele LaPointe
†
Grete Raba, http://orcid.org/0000-0002-7764-3878

ABSTRACT
Environmental pH and gut transit rate are the key factors determining the dynamics of colonic microbiota. In this study, the
effect of changing pH on the composition and metabolism of pooled faecal microbiota was elucidated at physiologically
relevant dilution rates Dhigh = 0.2 and Dlow = 0.05 1/h. The results showed the best adaptability of Bacteroides ovatus within
the pH range 6.0–8.0 at both dilution rates. The butyrate producing Faecalibacterium and Coprococcus comes were extremely
sensitive to pH > 7.5, while the abundance of Akkermansia muciniphila increased significantly at pH >7 at Dhigh , causing a
pH-dependant shift in the dynamics of mucin degrading species. Increased gas formation was observed at
pH < 6.5. Substantially more CO2 was produced at Dlow than at Dhigh (18-29 vs 12–23 mmol per L medium, respectively).
Methane was produced only at Dlow and pH > 7, consistent with the simultaneous increased abundance of
Methanobrevibacter smithii. Our study confirmed the importance of pH in the development of faecal microbiota in
pectin-supplemented medium. Fermentation of other dietary fibres can be studied using the same approach. The
significance of pH should be more emphasized in gut research and diagnostics.

Keywords: continuous culture; changestat; faecal microbiota; pH; growth rate; apple pectin

INTRODUCTION in cultivation and analytical methods provide tools to study

these interactions and to elucidate the relationships between
Human gastrointestinal tract contains trillions of bacteria which
gut microbiome and human health.
are in a dynamic relationship with the host. These bacteria
Nutrient and energy metabolism in human colon depends on
can break down complex polysaccharides such as plant-derived
colonic transit rate as it influences the specific growth rates of
dietary fibres, which are indigestible by host digestive enzymes
gut bacteria. The density of bacteria increases along the colon
(Martens et al. 2011; Kaoutari et al. 2013). Consumption of dietary
while the transit rate decreases, causing a non-linear relation-
fibre as a part of normal diet ensures the diversity of the colon
ship. Parameters such as environmental pH, peristalsis, redox
microbial consortium and helps to supress overgrowth of poten-
potential and others affect the composition and physiology of
tial pathogens (O’Keefe 2016; Chung et al. 2018; Schroeder et al.
gut microbiota and vice versa (Duncan et al. 2009; Vandeputte
2018). Bacterial metabolites provide energy and substrates for
et al. 2015; Chung et al. 2016; Roager et al. 2016). The effect of pH
mucin-producing epithelial cells and protect from gastrointesti-
on the growth of faecal microbiota has been studied by Chung
nal diseases (Martens, Neumann and Desai 2018; Schroeder et al.
et al. (2016) and on the growth of individual strains by Duncan
2018; Birchenough et al. 2019; Schroeder 2019). Recent advances

Received: 31 July 2020; Accepted: 15 April 2021


C The Author(s) 2021. Published by Oxford University Press on behalf of FEMS. All rights reserved. For permissions, please e-mail:

[email protected]

1
 2 FEMS Microbiology Letters, 2021, Vol. 368, No. 7

et al. (2009). Moreover, the gastrointestinal acidity depends on Defined base medium and substrates
an individual’s dietary habits. Certain polysaccharides bind sub-
stantial amounts of water and increase the volume and moving The defined growth medium (pH 7.2 ± 0.1) was prepared in
rate of the chyme (Cummings 1984; Elia and Cummings 2007; de 0.05 M potassium phosphate buffer made from 1 M stock solu-
Vries, Miller and Verbeke 2015). Degradation of polysaccharides, tions (ml/L): K2 HPO4 (28.9) and KH2 PO4 (21.1). The medium
release of organic acids, absorption of bile salts and water create contained mineral salts (mg/L): MgSO4 ∗ 7H2 O (36), FeSO4 ∗ 7H2 O
a pH gradient inside the large intestine (Cummings and Macfar- (0.1), CaCl2 (9), MnSO4 ∗ H2 O (3), ZnSO4 ∗ 7H2 O (1), CoSO4 ∗ 7H2 O
lane 1991). (1), CuSO4 ∗ 5H2 O (1), (NH4 )6 Mo7 O24 ∗ 4H2 O (1), NaCl (527); hemin
Mucins, the main components of intestinal mucus, pro- (5 mg/L); vitamin K1 (0.5 mg/L); l-amino acids (g/L): Ala (0.044),
vide substrate for mucin-degrading bacteria such as Akkerman- Arg (0.023), Asn (0.038), Asp (0.038), Glu (0.036), Gln (0.018),
sia muciniphila and Bacteroides thetaiotaomicron (Van Bueren et al. Gly (0.032), His (0.027), Ile (0.060), Leu (0.120), Lys-HCl (0.080),
2017; Van Herreweghen et al. 2017). Pectins are a group of car- Met (0.023), Phe (0.050), Pro (0.041), Ser (0.095), Thr (0.041),
bohydrates found in the cell walls and middle lamellae of Trp (0.009), Val (0.060), Tyr (0.015); vitamins (mg/L): biotin
fruits and vegetables and are common in human diet. Pectins (0.25), Ca-pantothenate (0.25), folic acid (0.25), nicotinamide

Downloaded from https://academic.oup.com/femsle/article/368/7/fnab042/6232156 by guest on 05 July 2021

are structurally complex and contain arabinan, arabinogalac- (0.25), pyridoxine-HCl (0.50), riboflavin (0.25), thiamine-HCl (0.25)
tan and galactan side chains connected to a galacturonan back- and (g/L): bile salts (0.5), NaHCO3 (2.0), Tween-80 (0.5), Na-
bone (Muzzarelli et al. 2012). Such dietary fibres are important thioglycolate (0.5), Cys-HCl (0.5). The medium was supple-
in modulating the composition of the gut microbial consor- mented with porcine gastric mucin (2.5 g/L; Type II, Sigma-
tium and in controlling the colonic transit rate (Parkar et al. Aldrich, USA). Apple pectin (2.5 g/L; Sigma-Aldrich, USA) was
2010; Larsen et al. 2019). The effect of pectins from citrus and used as a widely consumed dietary fibre in Estonian.
apple as substrates for faecal microbiota was recently stud-
ied by us (Adamberg and Adamberg 2018; Adamberg, Raba and
Cultivation system and culture conditions
Adamberg 2020).
Compositional and metabolic responses to various changes Four continuous cultivation experiments were carried out to
of environmental parameters can be studied by using a pooled study the combined effect of dilution rate and pH on faecal
faecal inoculum (Aguirre et al. 2014, 2015; de Souza et al. 2019; microbiota: (i) gradual pH decrease from 7.0 to 6.0 (D = 0.05 1/h),
Adamberg, Raba and Adamberg 2020). Microbiota standardisa- (ii) pH increase from 7.0 to 8.0 (D = 0.05 1/h), (iii) pH decrease
tion by pooling is useful for studying the growth trends in a from 7.0 to 6.0 (D = 0.2 1/h) and (iv) pH increase from 7.0 to 8.0
complex faecal consortium, unaffected by variations from indi- (D = 0.2 1/h), all at 36.6◦ C.
vidual donors. Although resulting in slightly higher biodiver- D-stat cultivation experiments were carried out in a 1 litre
sity, the majority of OTUs are shared in pooled and individ- fermenter using Biobundle cultivation system controlled by
ual samples (Aguirre et al. 2014). Advanced cultivation meth- ‘BioXpert’ software (Applikon, The Netherlands) as described in
ods enable to mimic specific conditions of human gut with Adamberg and Adamberg (2018). Culture volume was kept con-
precise computer-controlled algorithms and frequent sampling stant 300 mL by monitoring the weight of the fermenter with
(Chung et al. 2016, 2018; Adamberg, Raba and Adamberg 2020). PC-linked balance and outflow pump. Feeding and outflow were
In a changestat (D-stat), a chosen environmental parameter is controlled with variable speed pumps controlled by D-stat algo-
gradually changed while all the other parameters are kept con- rithm: N = N0 + a∗ t, where ”N” is the parameter that is being
stant, allowing to scan a steady state growth space within a sin- changed (in this case pH), ”N 0 ” is the initial value of the param-
gle experiment (Kasemets et al. 2003; Adamberg, Valgepea and eter, ”a” is the rate of change of parameter N (unit per hour), and
Vilu 2015). An essential advantage of continuous cultures over ”t” is the time (h).
in vivo experiments is the precise control of the environmen- Cell growth in steady state is described by Monod equation:
tal and nutritional parameters to elucidate specific effects of, μ = μmax ∗ S/(K s + S), where μ is the specific growth rate of the
for example, a single substrate or dilution rate on the micro- cell, S is the limiting substrate and K s is a constant describing
bial diversity and metabolic potential (Adamberg, Raba and the cell’s affinity to the substrate (Monod 1950). In a change-
Adamberg 2020). stat culture, the dilution rate D is directly related to μ, as the
The aim of the present study was to elucidate the effect of dilution rate controls the inflow of the limiting substrate. Thus,
smooth change of pH on the composition and metabolism of bacteria with a low specific growth rate may be washed out at
pooled faecal microbiota. A range of pH between 6.0 and 8.0 was a high dilution rate and vice versa (Adamberg and Adamberg
scanned at dilution rates Dlow = 0.05 1/h and Dhigh = 0.2 1/h, cor- 2018; Adamberg, Raba and Adamberg 2020). Two different dilu-
responding to slow and fast colonic transit rates, respectively. tion rates, Dlow = 0.05 1/h and Dhigh = 0.2 1/h were tested to
represent slow and fast transit rates of the human colon (Cum-
mings, Jenkins and Wiggins 1976; Fallingborg et al. 1989; Kozi-
olek et al. 2015; Maurer et al. 2015; Roager et al. 2016). The dilu-
MATERIALS AND METHODS tion rates were chosen to promote the growth of physiologically
Faecal inoculum relevant consortium whose specific growth rate was calculated
based on the estimation of the colonic transit time on a Western
Faecal samples were donated by seven healthy volunteers (19– diet between 10–120 h and the increase of bacterial counts from
37 years old, Caucasian). Exclusion criteria comprised the use of 10 8 in proximal colon to 1011 cfu/g in the faeces (Sender et al.
prebiotics and probiotics, laxatives and antibiotics four weeks 2016).
prior to sample collection. Faecal samples were homogenized The healthy colonic pH is neutral or mildly acidic (pH = 6.1–
in 4 volumes of 5% DMSO-containing PBS buffer as described 7.5; Nugent et al. 2001) but may increase up to pH 8 in the case of
previously (Adamberg et al. 2015). Equal volumes of seven faecal colorectal cancer (Kashtan et al. 1990; Ohigashi et al. 2013). More-
slurries were pooled and 2 mL aliquots were kept at −80◦ C for over, protein-rich diets can increase the colonic pH as a result of
use in repeated cultivation experiments. ammonia release from protein fermentation (Russell et al. 2011;
 Raba et al. 3

Aguirre et al. 2016). pH of the in vitro culture was controlled by The DNA sequence data was analysed using BION-meta (ww
the addition of 1M NaOH. The D-stat algorithm was started dur- w.box.com/bion). The sequences were first cleaned at both ends
ing the exponential growth phase of the bacteria, 15–17 h after using a 99.5% minimum quality threshold for at least 18 of
inoculation. Culture stability was evaluated by a stabilization of 20 bases for 5 -end and 28 of 30 bases for 3 -end, then joined,
the titration (indicating the acid production rate) and gas pro- followed by the removal of contigs shorter than 250 bp. The
duction rates. Average fluctuations of these parameters below sequences were then cleaned of chimeras and clustered by 95%
5% within the last three residential times were considered as oligonucleotide similarity (k-mer length of 8 bp, step size 2 bp).
indicator of a stable culture to start the changestat algorithm. In Lastly, consensus reads were aligned to the SILVA reference 16S
average, 6–7 residential volumes were needed to achieve stable rDNA database (v123) using a word length of 8 and similarity
cultures. When stabilized at pH 7, the pH was either gradually cut-off of 90%.
changed from 7.0 to 6.0 or from 7.0 to 8.0 with acceleration rates
a = 0.005 and 0.02 U/h, corresponding to Dlow and Dhigh , respec-
Statistical analysis
tively (Fig. S1, Supporting Information). This strategy allowed to
start all experiments from a comparable steady state. Differences in bacterial abundances and metabolite productions

Downloaded from https://academic.oup.com/femsle/article/368/7/fnab042/6232156 by guest on 05 July 2021

The feeding medium and the culture were flushed with during D-stat experiments were calculated by dividing samples
sterile-filtered nitrogen gas (99.9%, AGA) overnight before inoc- into two groups: (i) samples taken at pH < 6.5 (low pH) and (ii)
ulation and throughout the experiments to maintain anaerobio- samples taken at pH > 7.5 (high pH). Mean values and stan-
sis in the fermenter. The redox potential of the outflow culture dard deviations of bacterial abundances and metabolite produc-
was monitored regularly using InLabRedoxR electrode (Mettler tions were calculated in both groups and single parametric t-test
Toledo, USA). Samples were taken from the fermenter after every Benjamini-Hochberg correction of P-values was used to estimate
0.05pH-units. statistical significance.

Analytical methods Ethics statement

Samples from the fermenter outflow were collected on ice, cen- The study was approved by the Tallinn Medical Research Ethics
trifuged (14 000 g, 5 min, +4◦ C) and stored separately as super- Committee, Estonia (protocol No. 554).
natants and cell pellets at −20◦ C until analysis. The super-
natants were filtered using AmiconR Ultra-10K Centrifugal Filter RESULTS
Devices, cut-off 3 kDa (Millipore, USA). Concentrations of succi-
nate, lactate, formate, acetate, propionate, isobutyrate, butyrate, The effect of pH on the metabolism of apple pectin by
isovalerate and valerate were determined by HPLC (Alliance 2795 faecal microbiota at low dilution rate (Dlow = 0.05 1/h)
system; Waters, USA) equipped with BioRad HPX 87H column
The most abundant taxa at Dlow within the whole pH range
(Hercules, USA) with isocratic elution of 0.005 M H2 SO4 , flow
tested were Bacteroides ovatus (that combines B. ovatus, B. thetaio-
rate 0.5 mL/min, 35◦ C. The RI (model 2414; Waters, USA) and
taomicron and B. xylanisolvens which could not be differenti-
UV (210 nm; model 2487; Waters, USA) detectors were used
ated by SILVA 16S rDNA reference database v123), Ruminococ-
for quantification. Amino acid concentrations were determined
caceae UCG-013 and Akkermansia muciniphila (Fig. 1). B. ovatus is
with UPLC (Acquity; Waters, USA). The chromatographic data
a common species in the human colon and its active growth
were processed in Empower software (Waters, USA).
has been shown at pH 6.0–6.9 (Chung et al. 2016). As opposed
Gas volume was recorded using MilliGascounter (Ritter, Ger-
to Ruminococcaceae UCG-013 and A. muciniphila, B. ovatus was
many). Composition of the gas outflow was analysed with gas
highly abundant also at Dhigh . The other taxa with distinctly
analyser (Agilent 490 MicroGC Biogas Analyzer; Agilent 269
higher abundance at Dlow were Bacteroides dorei/vulgatus, Alistipes
Technologies Ltd., USA). Soluble gas concentration (c) in the cul-
sp. and Ruminococcaceae UCG-002 (Fig. 1).
ture liquid was calculated using Henry law: c = Hcp∗∗ p, where p
The growth of B. cellulosilyticus, Clostridium subterminale,
is the partial pressure of the given gas in the gas phase and Hcp∗
Clostridium tertium and Sutterella wadsworthensis was significantly
(M/atm) is the effective Henry constant of the given gas depen-
favored at pH >7 (Fig. 2). In contrast, growth of Dorea longicatena,
dent on pH (Sander 2015).
Faecalibacterium sp. and Escherichia coli, were supported by pH < 7.
The metabolite patterns followed the microbiota along with
the pH change. The molar ratio of acetate:propionate:butyrate
DNA extraction, sequencing and taxonomic profiling
was 1:0.47:0.34 at pH 6 and 1:0.31:0.16 at pH 8. Acetate, the main
DNA was extracted from the cell pellets using PureLink Micro- metabolite of many gut bacteria, was produced within the whole
biome DNA extraction kit (Thermo Fisher Scientific, UK). Univer- pH range and at both dilution rates (Fig. 3). Slightly less acetate
sal primers: F515 5’-GTGCCAGCMGCCGCGGTAA-3’ and R806 5’- was detected at low pH (19 vs 25 mM at pH 6.0 vs pH 8.0, respec-
GGACTACHVGGGTWTCTAAT-3’ were used for PCR amplification tively). Significant correlations between the consortial changes
of the V4 hypervariable regions of the 16S rRNA genes. Sequenc- and acetate production could not be elucidated from these data.
ing libraries were prepared with Nextera XT Index Kit (Illu- In accordance with the prevalence of B. ovatus, Ruminococ-
mina). Prepared libraries were quantified using QubitTM dsDNA caceae and A. muciniphila, production of propionate and butyrate
HS Assay Kit (quantitation range 0.2-100 ng; Thermo Fisher Sci- was seen throughout the whole tested pH range with slightly
entific) or QubitTM dsDNA BR Assay Kit (quantitation range 2– lower values at higher pH-s (Fig. 3). The family Ruminococ-
1000 ng; Thermo Fisher Scientific). All reagent kits were handled caceae includes several butyrate producing genera such as
in accordance with manufacturer’s instructions. Pooled libraries Faecalibacterium. In agreement with the decreasing abundance
were sequenced using Illumina iSeq 100 platform and i1 reagent of Faecalibacterium, significantly less butyrate was formed at
kit. The amplified region was 250–280 bp long and in average pH > 7.5, compared to that at pH < 6.5. However, similar trend
75 474 reads per sample were obtained. was not observed for other Ruminococcaceae.
 4 FEMS Microbiology Letters, 2021, Vol. 368, No. 7

Downloaded from https://academic.oup.com/femsle/article/368/7/fnab042/6232156 by guest on 05 July 2021

Figure 1. Dynamic changes of the dominant species of faecal microbiota in pectin supplemented medium during changing the pH from 6.0 to 8.0, at dilution rates
Dlow = 0.05 1/h (left) and Dhigh = 0.2 1/h (right).

Figure 2. Dynamic changes of the abundances (log scale) of the species of significant difference (P-value < 0.05) between pH 6.0 and 8.0 at dilution rates Dlow = 0.05
1/h (dark blue dots) and Dhigh = 0.2 1/h (light blue dots). Dark and light pink dots indicate the steady state conditions at pH = 7.0, before increase or decrease of the pH
at Dlow and Dhigh , respectively. The species detected at least 0.5% in at least one sample are shown. Data on abundances of all bacteria and P-values can be found in
the Table S1, Supporting Information.
 Raba et al. 5

Downloaded from https://academic.oup.com/femsle/article/368/7/fnab042/6232156 by guest on 05 July 2021

Figure 3. Dynamic changes of the metabolites (acids in mM, gases in mmol per L medium) produced by faecal microbiota in pectin supplemented medium at pH
between 6.0 and 8.0 at two dilution rates: Dlow = 0.05 1/h (dark blue dots) and Dhigh = 0.2 1/h (light blue dots). Dark and light pink dots indicate the steady state
conditions at pH = 7.0, before increase or decrease of the pH at Dlow and Dhigh , respectively. Data on metabolite concentrations and P-values can be found in the
Supplementary Table S1, Supporting Information.

Notable changes in gas formation profiles were seen depend-

ing on the dilution rate and changing pH. The amount of car-
bon dioxide, the main gaseous product, was 27 mmol per L
medium at pH 6.0 but only 18 mmol per L medium at pH 8.0 at
D low . Methanogenesis is a process carried out by slow-growing
archeae who convert acetate, carbon dioxide and hydrogen gas
into methane. Production of methane at D low was observed
at pH >7, in accordance with the increase of Methanobre-
vibacter smithii from < 0.05 (pH 7.5) to 0.5% (pH 7.9) (Fig. 2).
Methanogenic species in general are sensitive to the accumu-
lation of organic acids, especially propionate, and typically have
neutral or slightly alkaline pH optima (Angelidaki et al. 2011).
However, occurrence of M. smithii in our experiments at Dlow ,
with elevated concentrations of propionate, suggests its toler-
ance to propionate. Similarly, Barredo and Evison ( 1991) have
reported increased methane production by M. smithii at pH 8.0
despite the presence of propionate. Conversion of CO2 (and H2 )
into CH4 could partially explain the drop of CO2 (and a small drop
in hydrogen concentrations) at increasing pH. In addition, CO2
is fixed during production of succinate, amount of which was
rising at pH > 7.
Most of the amino acids from the growth medium were
Figure 4. Dynamic changes in concentrations of valerate, isovalerate, isobutyrate
depleted at Dlow (Table S1, Supporting Information). Distinctly
(mM) and hydrogen sulphide (mmol per L medium) during the growth of fae-
higher concentrations of isobutyrate, isovalerate and valerate —
cal microbiota on apple pectin at pH between 6.0 AND 8.0 at two dilution rates:
the metabolites derived from the breakdown of the branched Dlow = 0.05 1/h (dark blue dots), and Dhigh = 0.2 1/h (light blue dots). Dark and
chain amino acids (BCAA) were found, compared to those at light pink dots indicate the steady state conditions at pH = 7.0 before increase
Dhigh (Fig. 4). Production of hydrogen sulphide was more preva- or decrease of the pH at Dlow and Dhigh , respectively. Data on metabolite concen-
trations and P-values can be found in the Table S1, Supporting Information.
 6 FEMS Microbiology Letters, 2021, Vol. 368, No. 7

lent at Dlow . This agrees with our previous findings about pectin-degrading taxa changed from butyrate-producing Faecal-
conversion of the reducing agent cysteine into H2 S by faecal ibacterium and Lachnospira to propionate-producing B. cellulosilyti-
microbiota at low dilution rate (Adamberg and Adamberg 2018; cus. This indicates the dynamics of microbiota and metabolite
Adamberg, Raba and Adamberg 2020). patterns in the colonic environment in regard to the acid for-
mation and pH. Pure cultures of B. caccae were shown to con-
vert apple pectin into acetate and propionate (Sirotek et al. 2004).
The effect of pH on the metabolism of apple pectin by Depending on the species, Bacteroides may produce either succi-
faecal microbiota at high dilution rate (Dhigh = 0.2 1/h) nate, propionate, or both in parallel with acetate. The production
of propionate via succinate requires CO2 which is used for regen-
The most abundant taxa at Dhigh throughout the tested pH range
eration of NAD+ . On the other hand, when propionate is pro-
were B. ovatus and Lachnospiraceae UCG-008 (Fig. 1). The growth
duced via catabolism of puryvate, CO2 is released. Our data sug-
of some other species, namely Bifidobacterium bifidum, Clostrid-
gest that Dlow favours the development of consortia that produce
ium subterminale, Dorea longicatena, Lachnoclostridium torques, Sut-
propionate through succinate and CO2 , while Dhigh supports the
terella wadsworthensis and Escherichia coli, was also favored by
fast growing microbes that produce succinate without its further

Downloaded from https://academic.oup.com/femsle/article/368/7/fnab042/6232156 by guest on 05 July 2021

Dhigh (Fig. 1). The most distinct difference in microbial compo-
conversion into propionate.
sition between the two dilution rates was the prevalence of var-
A surprising dilution rate specific effect was seen for mucin
ious Ruminococcaceae at Dlow , and that of Lachnospiraceae at
degrading bacteria Akkermansia and Lachnoclostridium (L. torques
Dhigh . Two species from the family Lachnospiraceae, Lachnospira
and L. gnavus). These bacteria can degrade mucins but have dif-
pectinoschiza and Lachnoclostridium lactaris, were highly abundant
ferent pH optima. At Dhigh , the abundance of lachnoclostridia
(1–5%) at Dhigh , whilst not detected at Dlow .
decreased while that of A. muciniphila increased along with the
Growth of B. dorei/vulgatus, Blautia unclassified and Fae-
rise of pH from 7 to 8. This trend was not seen at Dlow . Although
calibacterium was significantly stimulated at pH <7 (Fig. 2). In
A. muciniphila is generally regarded as a beneficial gut bacterium,
contrast, the abundances of Bacteroides uniformis, C. subterminale
its abundance may increase in the case of colorectal cancer
and A. muciniphila, increased significantly at pH > 7. Competi-
(CRC) progression along with elevated pH (pH > 7.5) (Weir et al.
tiveness of A. muciniphila at alkaline pH at Dhigh was especially
2013; Zackular et al. 2013; Borges-Canha et al. 2015). Moreover,
interesting, considering that this species is known to have a
we saw the decline of butyrate producing taxa as well as lower
low specific growth rate. Although being below the detection
concentrations of butyrate at pH > 7.5, that often occur together
limit in a steady state at pH 7, its population achieved levels
with CRC. Moreover, the alkaline pH usually accompanies slow
comparable to those at Dlow at pH 8. Bifidobacterium bifidum and
gastrointestinal transit (Lewis and Heaton 1997) and constipa-
Bacteroides caccae had the highest abundances around neutral
tion, that is typical for CRC (Kojima et al. 2004). This suggests that
pH (6.5–7.5), a feature that was not seen for other taxa, neither
alkaline environments enable this bacterium to acquire bene-
at Dhigh or Dlow (Fig. 2). Significant reorganisation within the
fits over lachnoclostridia or other mucin degrading bacteria such
genus Bacteroides was observed regarding to the dilution rate
as Bifidobacterium bifidum, B. caccae and B. fragilis which abun-
(Fig. 2). While Bacteroides caccae was more prevalent at Dhigh , the
dances decreased at higher pH. A. muciniphila was a dominant
opposite was observed for B. dorei/vulgatus.
bacterium also at Dlow , independent of the pH.
Acetate was the main metabolite produced at all pH values
High pH in colonic lumen is one of the key parameters related
(23-27 mM, Fig. 3). Molar ratio of acetate:propionate:butyrate was
to CRC (Kashtan et al. 1990; Walker and Walker 1992; Ohigashi
1:0.35:0.20 at pH 6 and 1:0.24:0.10 at pH 8. At Dhigh , trace amounts
et al. 2013), however, the data on the influence of alkaline envi-
of lactate were seen, which were not detected at Dlow . This is
ronment on the development of colon microbiota are missing. It
most likely linked to the higher abundance of Bifidobacterium
might be explained by the relevance of alkaline pH to protein fer-
bifidum at Dhigh .
mentation resulting in formation of ammonia from amino acids.
Substantially less CO2 was produced at Dhigh than at Dlow (12-
Branched short chain fatty acids (BCFA) isobutyrate and isovaler-
23 vs 18–29 mmol per L medium). Similar to Dlow , decrease in
ate, which are produced during protein degradation have been
production of CO2 and H2 was observed with the change of pH
considered to be biomarkers for gut health as protein fermenta-
towards alkaline (Fig. 3). No methane was detected at Dhigh .
tion by genera such as Bacteroides and Clostridium can yield phe-
The majority of amino acids were depleted from the cul-
nols, p-cresol and biogenic amines—potentially harmful com-
ture medium within the whole pH range at Dhigh (Table S1,
pounds for the intestinal epithelium (Aguirre et al. 2016; Rios-
Supporting Information). Notable amounts of alanine, leucine,
Covian et al. 2020). Our results revealed the increase of prote-
isoleucine and valine were detected at Dhigh . Between pH 6.3–
olytic bacteria (Clostridium, Escherichia) along with the decrease
7.9, only 15%–30% of BCAA were metabolized. Below pH 6.3, the
of butyrate producers (Faecalibacterium, Blautia, Lachnospira) at
consumption of BCAA decreased even more and was minuscule
pH > 7.5, even in the presence of fermentable carbohydrate
at pH 6 (Table S1, Supporting Information). This was supported
pectin. A decrease in abundances of Faecalibacterium and Blautia
by the negligible amounts of metabolites of BCAA fermentation.
as a result of changing pH from 5.5 to 6.9 in chemostat experi-
ments with apple pectin as substrate (D = 0.04 1/h) was reported
by Chung et al. (2016). Similarly, Reichardt et al. (2018) showed an
DISCUSSION
enhanced growth of butyrate producers in faecal microbiota at
The species Bacteroides caccae, B. cellulosilyticus, B. ovatus and pH 5.5, compared to that at pH 6.5. In parallel with the decreas-
B. thetaiotaomicron, Faecalibacterium and Lachnospira pectinoschiza ing abundances of butyrate producers, we saw an increase in
comprised 30%–50% of the total population in th faecal cul- production of BCFA at Dlow . In addition, at pH close to 8, the
tures grown in apple pectin supplemented medium. The preva- increased solubility of H2 S would lead to formation of HS− ions,
lence of these pectin-degrading bacteria (Sirotek et al. 2004; Bid- that might inhibit the butyrate producing bacteria.
dle et al. 2013; Magnúsdóttir et al. 2017) emphasizes the impor- Methane production was detected only at Dlow . Increased
tance of primary fibre degraders in faecal microbiota. Inter- methane formation is also related to higher proportion of
estingly, along with the increasing pH, the composition of the colonic methanogenic microbes in CRC patients. We observed
 Raba et al. 7

gradual increase of Methanobrevibacter together with that of Adamberg K, Valgepea K, Vilu R. Advanced continuous cul-
methane production while changing the pH > 7.5 at Dlow . tivation methods for systems microbiology. Microbiology
Methanobrevibacter is commonly associated with slow transit 2015;161:1707–19.
microbiomes (Vandeputte et al. 2015), however, the effect of pH Aguirre M, Eck A, Koenen ME et al. Diet drives quick changes
on its growth has not been described earlier. The increased in the metabolic activity and composition of human gut
abundance of Methanobrevibacter may be related to its lower sen- microbiota in a validated in vitro gut model. Res Microbiol
sitivity to H2 S, while the abundances of butyrate-producing H2 S- 2016;167:114–25.
sensitive bacteria such as Faecalibacterium decreased along with Aguirre M, Eck A, Koenen ME et al. Evaluation of an optimal
the increase of H2 S concentrations. The main source of H2 S preparation of human standardized fecal inocula for in vitro
was cysteine, added to the culture medium. According to Meta- fermentation studies. J Microbiol Methods 2015;117:78–84.
Cyc.org database, Alistipes, B. cellulosilyticus, B. ovatus, Escherichia Aguirre M, Ramiro-Garcia J, Koenen ME et al. To pool or not
coli, and Klebsiella (all detected in this study) can convert cys- to pool? Impact of the use of individual and pooled fecal
teine to H2 S. Reaction of H2 S with metal (for example iron) ions samples for in vitro fermentation studies. J Microbiol Methods
results in formation of insoluble sulphides that are not available 2014;107:1–7.

Downloaded from https://academic.oup.com/femsle/article/368/7/fnab042/6232156 by guest on 05 July 2021

for bacterial cells. For anaerobic bacteria, however, iron in the Angelidaki I, Karakashev D, Batstone DJ et al. Biomethanation and
form of ferredoxins is extremely important for regeneration of Its Potential. 1st ed. Elsevier Inc., 2011.
the whole redox system. Barredo MS, Evison LM. Effect of propionate toxicity on
In conclusion, we showed that pH is a crucial trigger in methanogen-enriched sludge, Methanobrevibacter smithii,
the development of continuously growing faecal consortia. The and Methanospirillum hungatii at different pH values. Appl
changestat experiments demonstrated the pH-specific changes Environ Microbiol 1991;57:1764–9.
in the faecal microbiota, for example in relation to alkaline pH Biddle A, Stewart L, Blanchard J et al. Untangling the genetic
environment, often associated with pathologies. The species basis of fibrolytic specialization by lachnospiraceae and
such as B. ovatus were prevalent within the whole range of pH, ruminococcaceae in diverse gut communities. Diversity
while butyrate producing Faecalibacterium or Coprococcus comes 2013;5:627–40.
were severely inhibited at pH > 7.5. Further variations in the Birchenough G, Schroeder BO, Bäckhed F et al. Dietary destabili-
composition and metabolism were caused by the dilution rate, sation of the balance between the microbiota and the colonic
representing either fast or slow colonic transit rate. Conse- mucus barrier. Gut Microbes 2019;10:246–50.
quently, the concomitant bacterial metabolites have specific Borges-Canha M, Portela-Cidade JP, Dinis-Ribeiro M et al. Role of
health effects. The effect of pH on the fermentation of other colonic microbiota in colorectal carcinogenesis: a systematic
dietary fibres, and their combinations, by faecal microbiota can review. Rev Española Enfermedades Dig 2015;107:659–71.
be studied by applying a similar cultivation approach. As the pH Chung WSF, Walker AW, Louis P et al. Modulation of the human
of the colonic environment is directly related to diet, science- gut microbiota by dietary fibres occurs at the species level.
driven diet modifications could be an easy solution to support BMC Biol 2016;14:1–13.
gut health. Chung WSF, Walker AW, Vermeiren J et al. Impact of carbohy-
drate substrate complexity on the diversity of the human
colonic microbiota. FEMS Microbiol Ecol 2018;95:1–13.
SUPPLEMENTARY DATA Cummings JH, Jenkins DJA, Wiggins HS. Measurement of the
Supplementary data are available at FEMSLE online. mean transit time of dietary residue through human gut. Gut
1976;17:210–8.
Cummings JH, Macfarlane GT. The control and consequences of
ACKNOWLEDGEMENTS bacterial fermentation in the human colon. J Appl Bacteriol
The authors would like to thank Madis Jaagura for carrying out 1991;70:443–59.
the bioinformatic analyses. Cummings JH. Constipation, dietary fibre and the control of large
bowel function. Postgrad Med J 1984;60:811–9.
de Souza CB, Jonathan M, Saad SMI et al. Degradation of fibres
FUNDING from fruit by-products allows selective modulation of the gut
The project has received funding from the Institutional Research bacteria in an in vitro model of the proximal colon. J Funct
Funding (IUT 1927) of the Estonian Ministry of Education and Foods 2019;57:275–85.
Research and the European Regional Development Fund project de Vries J, Miller PE, Verbeke K. Effects of cereal fiber on bowel
EU48667. function: a systematic review of intervention trials. World J
Gastroenterol 2015;21:8952–63.
Conflicts of Interest. None declared. Duncan SH, Louis P, Thomson JM et al. The role of pH in deter-
mining the species composition of the human colonic micro-
biota. Environ Microbiol 2009;11:2112–22.
REFERENCES
Elia M, Cummings JH. Physiological aspects of energy
Adamberg K, Adamberg S. Selection of fast and slow growing metabolism and gastrointestinal effects of carbohydrates.
bacteria from fecal microbiota using continuous culture with Eur J Clin Nutr 2007;61:S40–74.
changing dilution rate. Microb Ecol Health Dis 2018;29:1549922. Fallingborg J, Christensen LA, Ingeman-Nielsen M et al. pH-
Adamberg K, Raba G, Adamberg S. Use of Changestat for Growth profile and regional transit times of the normal gut mea-
Rate Studies of Gut Microbiota. Front Bioeng Biotech 2020;8: sured by a radiotelemetry device. Aliment Pharmacol Ther
1–12. 1989;3:605–14.
Adamberg K, Tomson K, Talve T et al. Levan enhances associ- Kaoutari AEl, Armougom F, Gordon JI et al. The abundance and
ated growth of Bacteroides, Escherichia, Streptococcus and variety of carbohydrate-active enzymes in the human gut
Faecalibacterium in fecal microbiota. PLoS One 2015;10:1–18. microbiota. Nat Rev Microbiol 2013;11:497–504.
 8 FEMS Microbiology Letters, 2021, Vol. 368, No. 7

Kasemets K, Drews M, Nisamedtinov I et al. Modification of A- Reichardt N, Vollmer M, Holtrop G et al. Specific substrate-driven
stat for the characterization of microorganisms. J Microbiol changes in human faecal microbiota composition contrast
Methods 2003;55:187–200. with functional redundancy in short-chain fatty acid produc-
Kashtan H, Stern HS, Jenkins DJA et al. Manipulation of fecal pH tion. ISME J 2018;12:610–22.
by dietary means. Prev Med 1990;19:607–13. Rios-Covian D, González S, Nogacka AM et al. An overview on
Kojima M, Wakai K, Tokudome S et al. Bowel movement fre- fecal short-chain fatty acids along human life and as related
quency and risk of colorectal cancer in a large cohort study with body mass index: associated dietary and anthropomet-
of Japanese men and women. Br J Cancer 2004;90:1397–401. ric factors. Front Microbiol 2020;11:1–9.
Koziolek M, Grimm M, Becker D et al. Investigation of pH and Roager HM, Hansen LBS, Bahl MI et al. Colonic transit time is
temperature profiles in the GI tract of fasted human subjects related to bacterial metabolism and mucosal turnover in the
using the IntelliCap system. J Pharm Sci 2015;104:2855–63. gut. Nature Microbiol 2016;1:1–9.
Larsen N, De Souza CB, Krych L et al. Potential of pectins to bene- Russell WR, Gratz SW, Duncan SH et al. High-protein, reduced-
ficially modulate the gut microbiota depends on their struc- carbohydrate weight-loss diets promote metabolite profiles
tural properties. Frontiers in Microbiology 2019;10:1–13. likely to be detrimental to colonic health. Am J Clin Nutr

Downloaded from https://academic.oup.com/femsle/article/368/7/fnab042/6232156 by guest on 05 July 2021

Lewis SJ, Heaton KW. Increasing butyrate concentration in 2011;93:1062–72.
the distal colon by accelerating intestinal transit. Gut Sander R. Compilation of Henry’s law constants (version 4.0) for
1997;41:245–51. water as solvent. Atmos Chem Phys 2015;15:4399–981.
Magnúsdóttir S, Heinken A, Kutt L et al. Generation of genome- Schroeder BO, Birchenough GMH, Ståhlman M et al. Bifidobacte-
scale metabolic reconstructions for 773 members of the ria or fiber protect against diet-induced microbiota-mediated
human gut microbiota. Nat Biotechnol 2017;35:81–9. colonic mucus deterioration. Cell Host & Microbe 2018;23:
Martens EC, Lowe EC, Chiang H et al. Recognition and degra- 27–40.e7.
dation of plant cell wall polysaccharides by two human Schroeder BO. Fight them or feed them: how the intestinal
gut symbionts. PLoS Biol 2011;9:e1001221. DOI: 10.1371/jour- mucus layer manages the gut microbiota. Gastroenterology
nal.pbio.1001221. Report 2019;7:3–12.
Martens EC, Neumann M, Desai MS. Interactions of commensal Sender R, Fuchs S, Milo R. Are we really vastly outnumbered?
and pathogenic microorganisms with the intestinal mucosal Revisiting the ratio of bacterial to host cells in humans. Cell
barrier. Nat Rev Microbiol 2018;16:457–70. 2016;164:337–40.
Maurer JM, Schellekens RCA, van Rieke HM et al. Gastrointestinal Sirotek K, Slováková L, Kopečný J et al. Fermentation of pectin
pH and transit time profiling in healthy volunteers using the and glucose, and activity of pectin-degrading enzymes in the
IntelliCap system confirms ileo-colonic release of ColoPulse rabbit caecal bacterium Bacteroides caccae. Lett Appl Microbiol
tablets. PLoS One 2015;10:1–17. 2004;38:327–32.
Monod J. [The technique of continuous culture, theory and appli- Van Bueren AL, Mulder M, Van Leeuwen S et al. Prebiotic galac-
cations]. Ann Inst Pasteur 1950;79;390–410, ( in French). tooligosaccharides activate mucin and pectic galactan uti-
Muzzarelli RAA, Boudrant J, Meyer D et al. Current views on lization pathways in the human gut symbiont Bacteroides
fungal chitin /chitosan, human chitinases, food preser- thetaiotaomicron. Sci Rep 2017;7:1–13.
vation, glucans, pectins and inulin : a tribute to Henri Vandeputte D, Falony G, Vieira-Silva S et al. Stool consistency
Braconnot, precursor of the carbohydrate polymers sci- is strongly associated with gut microbiota richness and
ence, on the chitin bicentennial. Carbohydr Polym 2012;87: composition, enterotypes and bacterial growth rates. Gut
995–1012. 2016;65:57–62.
Nugent SG, Kumar D, Rampton DS et al. Intestinal luminal pH Van Herreweghen F, Van den Abbeele P, De Mulder T et al.
in inflammatory bowel disease: possible determinants and In vitro colonisation of the distal colon by Akkermansia
implications for therapy with aminosalicylates and other muciniphila is largely mucin and pH dependent. Beneficial
drugs. Gut 2001;48:571–7. Microbes 2017;8:81–96.
O’Keefe SJD. Diet, microorganisms and their metabolites, and Walker ARP, Walker BF. Faecal pH and colon cancer. Letters
colon cancer. Nat Rev Gastroenterol Hepatol 2016;13:691–706. 1992:572.
Ohigashi S, Sudo K, Kobayashi D et al. Changes of the intestinal Weir TL, Manter DK, Sheflin AM et al. Stool Microbiome
microbiota, short chain fatty acids, and fecal pH in patients and Metabolome Differences between Colorectal Cancer
with colorectal cancer. Dig Dis Sci 2013;58:1717–26. Patients and Healthy Adults. PLoS One 2013;8:e70803. DOI:
Parkar SG, Redgate EL, Wibisono R et al. Gut health benefits of 10.1371/journal.pone.0070803.
kiwifruit pectins: comparison with commercial functional Zackular JP, Baxter NT, Iverson KD et al. The gut microbiome
polysaccharides. J Funct Foods 2010;2:210–8. modulates colon tumorigenesis. MBio 2013;4:1–9.