Content uploaded by Françoise Wilhelmi de Toledo
Author content
All content in this area was uploaded by Françoise Wilhelmi de Toledo on Nov 13, 2019
Content may be subject to copyright.
Available via license: CC BY 4.0
Content may be subject to copyright.
RESEARCH ARTICLE
Changes in human gut microbiota composition are linked to the energy
metabolic switch during 10 d of Buchinger fasting
Robin Mesnage
1
†, Franziska Grundler
2,3
†, Andreas Schwiertz
4
, Yvon Le Maho
5,6
and
Françoise Wilhelmi de Toledo
2
*
1
Gene Expression and Therapy Group, King’s College London, Faculty of Life Sciences & Medicine, Department of Medical and Molecular Genetics,
8th Floor, Tower Wing, Guy’s Hospital, Great Maze Pond, London SE1 9RT, UK
2
Buchinger Wilhelmi Clinic, Wilhelm-Beck-Straße 27, 88662 Überlingen, Germany
3
Charité-Universitätsmedizin Berlin, Corporate Member of Freie Universität Berlin, Humboldt-Universität zu Berlin, and Berlin Institut of Health,
Berlin, Germany
4
Institute of Microecology, Auf den Lüppen 8, 35745 Herborn, Germany
5
Université de Strasbourg, CNRS UMR 7178, Institut Pluridisciplinaire Hubert Curien, 23 rue du Loess, 67200 Strasbourg, France
6
Département de Biologie Polaire, Centre Scientifique de Monaco, 8 Quai Antoine 1
er
, 98000 Monaco, Monaco
(Received 30 July 2019 –Final revision received 2 October 2019 –Accepted 8 October 2019)
Journal of Nutritional Science (2019), vol. 8, e36, page 1 of 14 doi:10.1017/jns.2019.33
Abstract
Fasting is increasingly popular to manage metabolic and inflammatory diseases. Despite the role that the human gut microbiota plays in health and diseases,
little is known about its composition and functional capacity during prolonged fasting when the external nutrient supply is reduced or suppressed. We
analysed the effects of a 10-d periodic fasting on the faecal microbiota of fifteen healthy men. Participants fasted according to the peer-reviewed
Buchinger fasting guidelines, which involve a daily energy intake of about 1046 kJ (250 kcal) and an enema every 2 d. Serum biochemistry confirmed
the metabolic switch from carbohydrates to fatty acids and ketones. Emotional and physical well-being were enhanced. Faecal 16S rRNA gene amplicon
sequencing showed that fasting caused a decrease in the abundance of bacteria known to degrade dietary polysaccharides such as Lachnospiraceae and
Ruminococcaceae. There was a concomitant increase in Bacteroidetes and Proteobacteria (Escherichia coli and Bilophila wadsworthia), known to use host-
derived energy substrates. Changes in taxa abundance were associated with serum glucose and faecal branched-chain amino acids (BCAA), suggesting
that fasting-induced changes in the gut microbiota are associated with host energy metabolism. These effects were reversed after 3 months. SCFA levels
were unchanged at the end of the fasting. We also monitored intestinal permeability and inflammatory status. IL-6, IL-10, interferon γand TNFαlevels
increased when food was reintroduced, suggesting a reactivation of the postprandial immune response. We suggest that changes in the gut microbiota are
part of the physiological adaptations to a 10-d periodic fasting, potentially influencing its beneficial health effects.
Key words: Periodic fasting: Buchinger fasting: Intestinal permeability: Inflammation: Well-being
Alternation of food abundance and food scarcity (feast and
fast) is part of human and animal physiology. Fasting happens
on a daily basis, usually during night hours, but also during
short or longer periods of time, e.g. in humans without
access to technologies of food conservation
(1)
or for religious
reasons
(2)
. Humans also voluntarily fast because it has been
†These authors contributed equally to this work.
Abbreviations: BCAA, branched-chain amino acid; BWC, Buchinger Wilhelmi Clinic; EDN, eosinophil-derived neurotoxin; LBP, lipopolysaccharide-binding protein; LPS,
lipopolysaccharide; sIgA, secretory IgA.
*Corresponding author: Françoise Wilhelmi de Toledo, fax +49 7551807806, email francoise.wilhelmi@buchinger-wilhelmi.com
© The Author(s) 2019. This is an Open Access article, distributed under the terms of the Creative Commons Attribution licence (http://creative-
commons.org/licenses/by/4.0/), which permits unrestricted re-use, distribution, and reproduction in any medium, provided the original work is
properly cited.
JNS
JOURNAL OF NUTRITIONAL SCIENCE
1
Downloaded from https://www.cambridge.org/core. IP address: 80.148.22.98, on 13 Nov 2019 at 06:48:18, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/jns.2019.33
documented to safely enhance well-being and to have thera-
peutic benefits
(3,4)
. Most scientific investigations in the last
decade were, however, focused on intermittent fasting, a recur-
rence of 16 to 48 h energy restriction alternating with food
intake
(3)
. By contrast, very few studies have examined
human physiology during a periodic fast lasting from 2 to
21 d or more, such as in the empirically based Buchinger
Wilhelmi fasting programme, described in peer-reviewed
therapeutic fasting guidelines
(5)
.
When animals or humans switch from eating to periodic
fasting, the source of energy for their cells switches from
food molecules absorbed through the gastrointestinal tract to
the utilisation of energy substrates mobilised out of several
body tissues, primarily adipose tissue but also body protein
(6)
.
In other words, there is a metabolic switch in energy utilisation
from glucose to fatty acids and ketones. This increase in the
rate of lipolysis and ketogenesis is reflected by a decrease in
blood glucose, insulin, insulin-like growth factor-1, concomi-
tant to an increase in glucagon, growth hormone, NEFA,
ketone bodies and insulin-like growth factor binding protein-1
levels
(4,7–9)
.
When the digestive tract is put temporarily at rest during peri-
odic fasting, it remodels its structure leading to a reversible atro-
phy as shown for rat intestinal villi
(10)
. In addition, changes in
gut motility, total intestinal mass
(11)
and microbiota compos-
ition have been described in the gut of fasting animals
(12)
.
The reduction of the external nutrient supply to the gut
microbiota, and the associated remodelling of the gastrointes-
tinal tissues of the host, have been qualified as a ‘microscopic
energy crisis’coupled to a ‘housing crisis’for the gut
microbiota
(13)
. The gut microbiota relies almost entirely on
host diet composition as well as on host food processing
capacity to obtain metabolic substrates, and to cover its energy
requirements of 150–450 kcal/d (628–1883 kJ/d)
(14)
. It seems
therefore inevitable that periods of fasting where no external
food enters the digestive tract have repercussions on the
microbiota composition.
Changes in gut microbiota caused by fasting have been stud-
ied in hibernating animals like hamsters
(15)
, squirrels
(16)
and
brown bears
(17)
. Another study described shared responses
of the microbiota across five vertebrates of different classes
(tilapia, toads, geckos, quail and mice) during prolonged fast-
ing
(12)
. The most common response was a decrease in the
abundance of microbial species using plant glycans as a source
of energy, while species using host glycans as a source of
energy had their abundance increased
(18)
. Despite the growing
interest in several patterns of fasting and particularly periodic
fasting, its effects on the human microbial ecosystems remain
essentially unknown
(19)
.
Since dietary changes have a large impact on the gut micro-
biota, it is likely that fasting can also trigger important gut
microbiota changes, which may in turn influence host health
and immunity
(20,21)
. This has been demonstrated in laboratory
animals, already. In mice, intermittent fasting prevents diabetic
retinopathy by restructuring the gut microbiota
(22)
. Other pub-
lications link therapeutic effects of fasting to changes of gut
microbiota in patients with rheumatoid arthritis
(23,24)
. Only
one study performed in human subjects showed that after 1
week of fasting, followed by 6 weeks of refeeding and probiotic
supplementation, an increase in abundance of lactobacilli,
Enterobacteriaceae and Akkermansia could be observed within
the gut microbiota
(19)
. However, the gut microbiota compos-
ition was not measured at the end of the fasting period.
To our knowledge no study so far has been performed to
investigate the effects of periodic fasting on the gut microbiota
in humans using high-throughput sequencing
(25)
. In order to
fill this important gap, we studied the composition of the
gut microbiota and a large range of health biomarkers in a
cohort of fifteen healthy men before, on the last day of fasting,
during the refeeding-period and 3 months after a 10-d periodic
Buchinger fasting.
Materials and methods
Study design
This study was conducted according to the guidelines laid
down in the Declaration of Helsinki and all procedures involv-
ing human subjects were approved by the Baden-Württemberg
medical council (application no. F-2016-090; 27 September
2016). Written informed consent was obtained from all sub-
jects. The study was registered at the German Clinical Trials
Register (DRKS-ID: DRKS00011165, trial registry name:
Effects of the Buchinger Wilhelmi fasting programme on
energy metabolism and muscle function in humans (https://
www.drks.de/drks_web/setLocale_DE.do) 24 October
2016). It was conducted at the Buchinger Wilhelmi Clinic
(BWC) in Überlingen, Germany, between 20 November
2016 and 10 December 2016. All participants were in general
good health. Four time points were specified for each individ-
ual. The baseline examination was conducted 1 d before fast-
ing (time point 1). The second examination was done at the
end of the 10-d fasting period (time point 2). The third exam-
ination was conducted on the fourth day of the progressive
refeeding period after fasting (time point 3). The last examin-
ation took place 3 months after the fasting (time point 4). For
this follow-up, the subjects returned for 1 d to the BWC
between 1 March 2017 and 5 March 2017.
Participants
Participants were recruited in August 2016 among organisa-
tions involved in the practice of fasting, which is very popular
in Germany. Out of a total of fifty-eight men, we recruited six-
teen men according to age, physical and psychological health
criteria. Included were men aged between 18 and 70 years
with a BMI between 20 and 32 kg/m
2
(26·5(SD 3·0) kg/
m
2
). One participant was excluded retrospectively due to
incomplete collection of stool samples. Thus, the data analysis
included fifteen men (Fig. 1). The age of the participants was
44·6(
SD 13·5) years. BMI was 26·5(SD 3·0) kg/m
2
. Exclusion
criteria were predefined according to a list and included cach-
exia, anorexia nervosa, advanced kidney, liver or cerebrovascu-
lar insufficiency
(5)
. Smoking and the intake of antibiotics
within the last 8 weeks, as well as the intake of probiotics
within the last 4 weeks, led to exclusion.
2
journals.cambridge.org/jns
Downloaded from https://www.cambridge.org/core. IP address: 80.148.22.98, on 13 Nov 2019 at 06:48:18, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/jns.2019.33
Fasting intervention
All subjects fasted according to the fasting programme of the
BWC which is documented in the guidelines of the fasting
therapy
(5)
. They stayed under daily supervision of nurses and
physicians specialised in fasting therapy. On the day of admis-
sion in the BWC the participants received a standardised low-
carbohydrate vegetarian dinner. On the next day before the
beginning of the fast, the participants were given a 600 kcal
(2510 kJ) vegetarian diet consisting of rice and vegetables
divided in three meals. During fasting all subjects were
asked to drink 2–3 litres of water or non-energy herbal teas
on a daily basis. Additionally, all participants received a portion
of 20 g honey. Furthermore, an organic freshly squeezed fruit
juice (250 ml) was served at noon and a vegetable soup (250
ml) in the evening. On average, the total daily nutrient com-
position is described in Table 1. With the beginning of fasting
the subjects entered a standardised programme of physical
exercise alternating with rest. The exercise programme con-
sisted of outdoor walks and gymnastic groups. The whole pro-
gramme was supervised by certified trainers. To initiate the
10-d fasting period, the intestinal tract was emptied through
the intake of a laxative (20–40 g NaSO
4
in 500 ml water
according to body weight). During the fasting period an
enema (1 litre water at 37°C) was applied by a certified
nurse every second day. This procedure, whose effects remains
to be investigated in a clinical study, is assumed to remove
intestinal remnants of the last meals, desquamated mucosal
cells from the gastrointestinal walls and basal secretions both
occurring during fasting. Based on clinical and empirical
observations, it facilitates the transition to the fasting mode,
leads quicker to the fasting-specific absence of hunger, and
prevents common symptoms observed at the beginning of
the fasting like headaches and fatigue. Although the safety of
enemas is still debated, their use for more than 60 years at
the BWC has never been linked to clinical complications.
From the tenth day of fasting on, food was stepwise reintro-
duced during the following 4 d. The food consisted of an
ovo-lacto-vegetarian organic diet with a progressive increase
of energy from 3347 to 6694 kJ/d (800 to 1600 kcal/d)
(Supplementary Table S3). The third faecal sample was
obtained from the first defecation after refeeding, varying
between the first and fourth refeeding day.
Clinical parameters
All participants underwent a thorough physical examination,
their medical history was documented and their height was
assessed in the admission consultation by the physician (Seca
285; Seca). Every morning trained nurses recorded body
weight while the subjects wore only underclothing (Seca 704;
Seca). Blood pressure and heart frequency were measured at
the non-dominant arm in a sitting position (upper-arm
blood pressure monitor, boso Carat professional; BOSCH +
SOHN GmbH u. Co. KG). Waist circumference was deter-
mined with a measuring tape mid-way between the lowest
rib and the iliac crest (openmindz® GmbH). The participants
self-reported their physical and emotional well-being on
numeric rating scales from 0 (very bad) to 10 (excellent).
Also, 3-d food protocols were documented prospectively by
each subject before fasting and 3 months afterwards. The
safety of the Buchinger Wilhelmi periodic fasting programme
was continuously monitored by a staff of doctors and nurses.
Biochemical parameters
Blood samples were collected by trained physicians in the
morning and drawn into EDTA (S-Monovette
®
2·7ml K3
Fig. 1. Flow chart of the recruitment procedure. BWC, Buchinger Wilhelmi Clinic.
Table 1. Nutrient composition of the diet during fasting
Parameter Amount
Fat intake (g/d) 0·2
Protein intake (g/d) 1·8
Carbohydrate intake (g/d) 56·2
Fibre intake (g/d) 1·1
Energy intake 234·4
kJ/d 980·7
kcal/d 234·4
3
journals.cambridge.org/jns
Downloaded from https://www.cambridge.org/core. IP address: 80.148.22.98, on 13 Nov 2019 at 06:48:18, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/jns.2019.33
EDTA), citrate (S-Monovette
®
3 ml 9NC, citrate 3·2%
(1:10)), and blood sedimentation tubes (S-Sedivette
®
3·5ml
4NC, ESR/citrate buffer (1:5)), and were gently shaken after
filling. Additionally, serum tubes including serum gel with clot-
ting activator (S-Monovette
®
9 ml Z-Gel) were used and
stored upright for 30 min until coagulation with a subsequent
centrifugation step at 3920 gfor 10 min at room temperature.
All tubes were manufactured by Sarstedt AG & Co. The rou-
tine parameters were sent to MVZ Labor Ravensburg and ana-
lysed according to the manufacturer’s instructions in a fully
automated laboratory. Blood cell count (leucocytes, erythro-
cytes, Hb, mean cell volume, thrombocytes) was measured
using the blood analyser Sysmex XN-9000 (Sysmex Europe
GmbH). Glucose, total cholesterol and TAG were analysed
with ADVIA 2400 (Siemens Healthcare GmbH). Insulin
was measured with Centaur XP (Siemens Healthcare
GmbH). Remaining blood samples were immediately frozen
at −70°C.
Inflammatory blood parameters were measured using the
Human High Sensitivity T Cell Panel kit (HSTCMAG-
28SK). Results were obtained by the Bio-Plex 200 Analyser
System and data were analysed with Bio-Plex Manager
software (Bio-Rad Laboratories). Bacterial lipopolysaccharide
(LPS) serum levels were measured by a commercial ELISA
kit (Cusabio). Antibodies specific for LPS were pre-coated
onto a microplate and 100 µl of standards or sample were
incubated for 2 h at room temperature. After incubation, sam-
ples were analysed at 450 nm. Values were expressed as pg/ml;
intra-assay and inter-assay CV were 8 and 10 %, respectively.
Plasma LPS-binding protein (LBP) was measured by a LBP
soluble ELISA kit (Hycult Biotechnology) according to the
manufacturer’s protocols.
The semi-quantitative concentration of ketone bodies was
self-measured in the first morning urine using Ketostix
(Bayer AG) that react according to the concentration of acet-
oacetic acid.
Faecal parameters
Faecal samples were collected four times using sterile contain-
ers (MED AUXIL stool collector set; Süsse). The first sample
was collected before the start of the fasting and the second fol-
lowing intake of a laxative (Laxoberal Abführ-Tropfen,
sodium picosulphate; Sanofi-Aventis Deutschland GmbH) in
the evening of the ninth fasting day. The third sample was
obtained from the first defecation after refeeding and the
last sample in the follow-up phase. Faecal samples were imme-
diately frozen and stored at −70°C. The samples were sent to
the Institute of Microecology for analysis.
Faecal calprotectin and zonulin concentrations were mea-
sured by an ELISA as described elsewhere
(26)
. Faecal lactofer-
rin concentrations were determined using the IBD-SCAN
®
test (TechLabR, Inc.), faecal α−1-antitrypsin concentrations
were analysed using the AAT test (Maier Analytic) following
the instructions. Lysozyme and β-defensin were measured by
an ELISA (Immunodiagnostik). Branched-chain amino acids
(BCAA), EDN (eosinophil-derived neurotoxin, eosinophil
protein x), sIgA and bile acid concentration were determined
with the BCAA, EDN, IDK
®
sIgA, and IDK
®
bile acids
test kits, respectively (Immunodiagnostik). SCFA were deter-
mined using GC as previously described
(27)
.
16S rRNA gene amplicon sequencing
In order to determine the composition of the gut microbiota
during the fasting intervention, we sequenced PCR-amplified
marker 16S ribosomal RNA genes fragments which contain
a bacterial taxa-specific region
(28)
. Microbial DNA was
extracted from 200 mg of faecal sample using the
QIAsymphony
®
DSP Virus/Pathogen Mini-Kit (Qiagen)
according to the manufacturer’s instructions on the
QIAsymphony
®
SP (Qiagen). DNA purity and concentration
were measured with an Implen NanoPhotometer P-Class 360
(Implen GmbH).
The partial sequences of the hypervariable region of the 16S
rRNA gene (V4 and V5) were PCR amplified using the primer
520 forward (5′-AYTGGGYDTAAAGNG-3′) and 907
reverse (5′-CCGTCAATTCMTTTRAGTTT-3′)
(29)
. PCR
amplification was performed at least twice for each sequencing
set up. The PCR mixture with a final volume of 25 µl con-
sisted of 0·5 µl of each primer (10 µM), 0·6 µl of dNTP-mix
(10 mMeach), 5 µl 5× KAPA HifiPuffer including 20
mM-MgCl
2
(Roche), 0·1 µl KAPA HifiPolymerase (Roche),
1 µl DNA isolate and filled up with nuclease-free water.
PCR reactions were performed in a T100 Thermal Cycler
(Bio-Rad Laboratories) using the following programme: 3
min at 95°C for initial denaturation, twenty-five cycles of 30
s at 95°C for denaturation, 30 s at 55°C for annealing, and
45 s at 72°C for elongation, followed by a final elongation
step for 5 min at 72°C. Water-template control and
Escherichia coli DNA as positive control were included for
each set of PCR reactions. Success of PCR were verified by
agarose gel electrophoresis using Midori Green as DNA-dye
(Biozym). Both PCR were pooled and purified with
Agencourt AMPure beads (Beckman Coulter) into 50 µl of
10 mM-Tris (tris(hydroxymethyl)aminomethane; pH 8·5).
A second PCR step was then performed to add unique
index barcodes with sequencing adaptors to the amplicon tar-
gets. The Ion Torrent set up custom-made index sequences
were chosen (added in the supplement; Integrated DNA
Technologies) as forward primer. The index PCR reaction
had a total volume of 50 µl and included 1 µl
Ion-index-primer forward and 1 µl 926Rcomb reverse primer
(5′-CCTCTCTATGGGCAGTCGGTGAT CCGTCAATTC
MTTTRAGTTT-3′) for Ion-Torrent set ups with 1·2µl of
dNTP-Mix (10 mMeach), 10 µl 5× KAPA HifiPuffer includ-
ing 20 mM-MgCl
2
(Roche), 0·2 µl KAPA HifiPolymerase
(Roche), 5 µl amplicon DNA and filled up with nuclease-free
water. PCR reactions were performed in a T100 Thermal
Cycler (Bio-Rad Laboratories) using the same programme as
above with eight cycles. Primers that were used were designed
for the V4 and V5 variable region of the bacterial 16S-rRNA
gene which led to a length of around 364 bp for bacterial iden-
tification. With indices and linker sequences, libraries have a
mean sequence length of 528 bp. After purification with
AMPure beads, quality checks for library sizes and DNA
4
journals.cambridge.org/jns
Downloaded from https://www.cambridge.org/core. IP address: 80.148.22.98, on 13 Nov 2019 at 06:48:18, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/jns.2019.33
concentration were performed with the Aglient Bioanalyzer
using Aglient DNA 1000 chips (Agilent Technologies). To
determine the DNA concentration, the Qubit dsDNA HS
Assay Kit (Thermo Fisher Scientific) was used.
Libraries were finally pooled in equivalent 100 pMfor Ion
Torrent. Libraries prepared for Ion Torrent sequencing
template-positive Ion PGM template Hi-Q ion sphere particles
were produced using the emulsion PCR technique in the Ion
One Touch 2 and the Ion PGM Hi-Q OT2 Kit (Life
Technologies) following the manufacturer’s instructions for
PCR, recovery and quality control. The libraries were divided
to be run on four ION 318 chip-kit v2 BC chips (Life
Technologies) executing 850 sequencing flows on an Ion
PGM sequencer (Life Technologies) following the manufac-
turer’s instructions.
A total of six sequencing runs were performed. Each
sequencing run was processed separately in order to account
for the differences in sequencing quality. The 16S sequencing
data included a total of 31 556 258 reads (average of 498 493
reads per samples, range 70 875–2 884 392) available as
demultiplexed FASTQ files on the SRA archive in the project
PRJNA531091. Data analysis was done with Rosalind, the
BRC/King’s College London high-performance computing
cluster. The DADA2 algorithm was used
(25)
to correct for
sequencing errors and identify amplicon sequence variants
(ASV) using R version 3.5.0. We trimmed the first fifteen
poor-quality bases at the 5′side of reads as recommended
by the DADA2 manual
(25)
. A total of 20 728 sequence var-
iants was identified. The taxonomy was assigned using the
SiLVA ribosomal RNA gene database v132 up to the species
level. ASV with fewer than ten counts in less than 20 % of the
individuals were discarded. This resulted in a dataset com-
posed of 1673 ASV in sixty-three individuals that was brought
forward to a statistical analysis.
Statistical analysis
Justification of sample size. The sample size was calculated
based on the data obtained with Ramadan intermittent
fasting in a similar population
(30)
. Assuming a test–retest
correlation coefficient for the measure of 0·9, a group of
fourteen subjects was necessary to detect a post–pre
difference of about 7 % with a power of 0·95 and an αfor
unilateral test of 0·05.
The αdiversity, representing the diversity of the total num-
ber of species within the samples
(31)
, was measured using
Shannon’s diversity index, and the difference between groups
evaluated with the Kruskal–Wallis rank sum test. We also mea-
sured the βdiversity, representing the diversity of the total
number of species between the samples
(31)
. We used the
Bray–Curtis dissimilarity index which was calculated using
the R package Vegan and recommended for proportion data.
A stress function was used to measure the goodness of fit
between the ordination and the original data. The 16S rRNA
gene amplicon sequencing dataset was further processed
using the R package Phyloseq in order to collapse the DNA
sequences to different taxonomic levels
(32)
. Relative abundance
values were transformed using the centred log-ratio (clr)
transformation with the R package compositions. We then per-
formed a statistical analysis of the differences in microbiota
composition between time points for each taxa and of the
associations between taxa abundance and health parameters,
by fitting linear mixed models with lmer, considering repeated
sampling of individuals as random effects. Pvalues where cal-
culated using the difflsmeans function with the R package lmert-
est. The Benjamini & Hochberg correction procedure was
applied to control the false discovery rate
(33)
.
Results
The aim of this investigation was to describe the changes
within the gut microbiota caused by a 10-d fast. This was
done by measuring the microbiota composition and clinical
parameters 1 d before fasting (time point 1), at the end of
the 10-d fasting period (time point 2), on the fourth day of
the following progressive refeeding period (time point 3),
and 3 months after the fasting period (time point 4). Fasting
resulted in a significant weight reduction of 5·9(
SD 0·8) kg
(P=0·0002). Abdominal circumference, systolic blood pres-
sure and diastolic blood pressure were also significantly
reduced after fasting and at the end of the refeeding period
(Table 2). Furthermore, fasting was accompanied by an
enhancement of well-being (Table 2). Emotional well-being
increased at the end of fasting and was significantly enhanced
after the refeeding period (P=0·00079). These parameters
returned to a baseline level 3 months after the fasting period.
Physical well-being also increased and reached significant
enhancement after refeeding (P=0·000094), and was main-
tained 3 months after fasting (P=0·049).
Clinical data measurements
Clinical data measurements confirmed the energy metabolism
switch from carbohydrates to fatty acids and ketones (Fig. 2).
Glycaemic control improved during fasting: glucose and insulin
decreased significantly at the end of fasting and the refeeding per-
iod (Fig. 2). Parameters of glucoregulation returned to baseline
levels 3 months after the fasting. TAG and total cholesterol
were significantly reduced by the fasting period (Fig. 2).
Acetoacetic acid in urine increased significantly during fasting
(P=1·5×10
−6
) and declined during refeeding (P=3·8×10
−5
).
Microbiome analysis
We identified 213 unique species in the gut microbiota of fif-
teen individuals. The results from the 16S rRNA gene ampli-
con sequencing were validated by a taxa-specific quantitative
PCR: the Spearman’s rank correlations between these two
methods for eight taxa were all statistically significant (P
0·0016–4·9×10
−12
) (Supplementary Fig. S1). The compos-
ition of the microbiota was highly individualised (Fig. 3(A)).
There were no differences in α-diversity (Fig. 3(B)).
However, the comparison of Bray–Curtis distances revealed
that the composition of the microbiota changed during the
course of the intervention (Fig. 3(C)). Even if fasting had
large effects on the gut microbiome composition, Bray–
5
journals.cambridge.org/jns
Downloaded from https://www.cambridge.org/core. IP address: 80.148.22.98, on 13 Nov 2019 at 06:48:18, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/jns.2019.33
Curtis distances of the samples for a given individual remained
lower than distances across individuals, showing that gut
microbiomes remained individualised even after fasting. We
then used linear mixed models to identify the bacterial species
which were the most affected (Fig. 3(D)). A total of thirty-one
sequence variants had their abundances significantly changed
by the intervention (Supplementary Table S1). There was an
inversion of the Firmicutes:Bacteroidetes ratio (Fig. 3(D)).
Bacteroidetes (40·7 %) became the dominant taxa after
the fasting period due to a large decrease in the relative abun-
dance of Firmicutes (39·9 %). This included bacteria known to
degradedietary plant polysaccharides such as the Lachnospiraceae
family (Fusicatenibacter saccharivorans,Lachnospira pectinoschiza,
Coprococcus_2 eutactus,Pseudobutyrivibrio spp., Roseburia faecis)and
from the Ruminococcaceae family (Faecalibacterium prausnitzii,
Ruminococcus_1 bicirculans,Ruminococcus_2 bromii). There was a con-
comitant increase in Bacteroides abundance (Bacteroides nordii,
Bacteroides fragilis) and in Proteobacteria abundances (E. coli,
Bilophila wadsworthia). Following food reintroduction (days 10 to
14, from 3347 to 6694 kJ/d (800 to 1600 kcal/d)) the gut
microbiota composition reflected a partial recovery (Fig. 3).
After 3 months, subjects’gut microbiotas had returned to a
basal level in comparison with the baseline established on the
day of arrival at the clinic.
Faecal metabolites
We measured the metabolism of the gut microbiota by measur-
ing the levels of SCFA and BCAA. The concentrations of the
main SCFA (acetate, propionate, butyrate) were not changed
by the fast. However, a statistically significant decrease in
i-butyrate (P=0·005) and valerate (P=0·005) levels was
observed during the refeeding period. In contrast, the levels of
SCFA significantly increased 3 months after the fasting in com-
parison with pre-intervention levels. This could be linked to the
increased abundance of the known SCFA producer Coprococcus
eutactus. BCAA increased significantly (P=0·00005) during fast-
ing, returned to baseline after refeeding, and declined signifi-
cantly (P=0·0003) after 3 months (Table 3).
Faecal biochemical markers
The levels in faecal markers of gut permeability (zonulin,
α−1-antitrypsin, bile acids) were stable across the different
phases of the study (Table 3). Plasma LBP, which reflects
the exposure to bacterial LPS, was significantly decreased dur-
ing fasting and remained reduced during the refeeding period.
Faecal markers of intestinal inflammation suggested an inflam-
matory response after fasting that normalised after 3 months
(Table 3).
The levels of IL-6, IL-10, interferon γand TNFα, which
are markers of inflammation, showed a trend towards increase
from the beginning to the end of fasting but this increase was
not significant. Yet, 4 d after food was reintroduced, there was
a significant increase in comparison with baseline levels of all
four cytokines (IL-6, IL-10, interferon γ, TNFα)(Table 3).
This suggest that an immune reaction associated with
inflammation results mainly from food reintroduction.
Table 2. Summary of the changes in clinical biomarkers and well-being during fasting
(Mean values, ranges and standard deviations)
Parameter Pre-fasting End of fasting 3 d after After 3 months
Mean Range SD Mean Range SD PMean Range SD PMean Range SD P
Weight (kg) 85·770·5–106·91·079·866·1–99·59·6 *** 79·766·5–100·19·5 *** 81·970·4–99·58·3NS
BMI (kg/m
2
)26·520·4–32·33·024·719·2–30·02·7 *** 24·719·4–30·12·7 *** 25·521·1–31·82·6*
SBP (mmHg) 128·396–158 14·6 121·1 104–152 11·3 * 116·1 103–140 9·6 *** 127·6 104–142 10·2NS
DBP (mmHg) 81·657–90 9·377·764–87 6·5* 77·666–100 9·0* 81·272–95 7·2NS
Waist (cm) 92·776–108 9·189·074–106 9·0** 89·780–104 7·5** 92·180–108 7·1NS
Emotional well-being, score 0–10 7·34–10 1·98·32–10 2·1NS 9·16–10 1·2 *** 7·94–10 1·6NS
Physical well-being, score 0–10 6·83–10 2·47·81–10 2·4NS 9·05–10 1·5 *** 8 5–10 1·4*
SBP, systolic blood pressure; DBP, diastolic blood pressure.
Significantly different at end of fasting in comparison with pre-fasting baseline levels: *P<0·05, **P<0·01, ***P<0·001 (ANOVA).
6
journals.cambridge.org/jns
Downloaded from https://www.cambridge.org/core. IP address: 80.148.22.98, on 13 Nov 2019 at 06:48:18, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/jns.2019.33
Associations between microbiome composition and health
markers
We ultimately evaluated if gut microbiota composition could be
associated with variations in markers of health. A total of fifty
associations were statistically significant (Benjamini–Hochberg
adjusted P<0·05) out of the 4136 associations tested.
Although our relatively small sample size only allowed us to
draw limited conclusions, the bacterial species which were sig-
nificantly associated with biochemical parameters were also
the bacterial species mainly affected by fasting (Fig. 4). The
abundance in Lachnospiraceae (Coprococcus_2 eutactus,
Fusicatenibacter saccharivorans,andLachnospira pectinoschiza)was
positively associated with plasma glucose levels and negatively
associated with BCAA levels. By contrast, Bacteroidetes
(Bacteroides dorei/fragilis and Bacteroides thetaiotaomicron), as well as
aProteobacterium (Bilophila wadsworthia), presented the opposite
trend and were negatively associated with plasma glucose levels
and positively associated with BCAA levels.
Discussion
There is a growing interest in biomedical research on fasting
and its therapeutic aspects
(7)
, as well as in the gut microbiota
and its influence on human health
(34)
. Here we document
that periodic fasting in humans has major effects on
biochemical markers such as blood lipids, glucoregulation,
enhancement of emotional and physical well-being, and
the faecal microbiota. We observed a major decrease in the
relative abundance of the Firmicutes, Lachnospiraceae
and Ruminococcaceae, concomitant to an increase in
Bacteroidetes and Proteobacteria. This profile correlates well
with known effects of fasting in hibernating animals
(17,18)
and with daily cyclical fluctuations in the composition of the
mouse gut microbiome according to feeding/fasting
rhythms
(35)
. When mice are eating, usually during the night,
Firmicutes are proliferating and become the dominant phylum
while Bacteroidetes rise during fasting (daytime).
The changes in gut microbiome composition and energy
metabolism were reversed after 3 months. This reversibility
in healthy subjects does not rule out the potential of this fast-
ing protocol to change over the long term the gut microbiota
composition of ill subjects. A large number of studies have
indicated that unhealthy individuals tend to have a lack of
microbial diversity in their gut
(34)
. Since more diverse micro-
bial environments are known to be more resilient
(36)
,we
hypothesise that the gut microbiome of unhealthy individuals
with a loss of diversity may have a different resilience to the
fasting intervention. Prolonged fasting is known to be an
Fig. 2. Metabolic switch from carbohydrates to fatty acids and ketones induced by a 10-d fasting. A regression spline was fitted on individual acetoacetic values to
show the variations in ketosis during the course of the study. The distribution is summarised by box plots, with the upper and lower hinges extending to the first and
third quartiles. Statistical significance was assessed with an ANOVA in comparison with pre-fasting baseline levels (* P<0·05, ** P<0·01, *** P<0·001). Biomarker
levels are presented for each individual across the four phases of the intervention (1, baseline examination; 2, at the end of the 10-d fasting period; 3, on the fourth
day of the following progressive refeeding (RF); 4, 3 months after the fasting period).
7
journals.cambridge.org/jns
Downloaded from https://www.cambridge.org/core. IP address: 80.148.22.98, on 13 Nov 2019 at 06:48:18, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/jns.2019.33
Fig. 3. Fasting caused a decrease in the abundance of bacteria from the Lachnospiraceae and Ruminococcaceae families concomitant with an increase in
Bacteroidetes. (A) The gut microbiome of fifteen subjects undergoing a 10-d fasting was analysed by 16S rRNA gene amplicon sequencing. The taxonomic profiles
are presented for each individual (ID from 1 to 15) across the four phases of the intervention (1, baseline examination; 2, at the end of the 10-d fasting period; 3, on the
fourth day of the following progressive refeeding; 4, 3 months after the fasting period). (B) The αdiversity was not changed by fasting. (C) The measure of dissimi-
larities between samples (Bray–Curtis distance) revealed that the samples separate by time point along the yaxis. (D) The evaluation of changes in species relative
abundances across the fasting period revealed that fasting caused a statistically significant decrease in the abundance of bacteria from the Lachnospiraceae family
(in green), and from the Ruminococcaceae family (in blue), concomitant with an increase in Bacteroidetes (in pink). The composition of the microbiome returned to a
basal level during refeeding. NMDS, non-metric multidimensional scaling. 8
journals.cambridge.org/jns
Downloaded from https://www.cambridge.org/core. IP address: 80.148.22.98, on 13 Nov 2019 at 06:48:18, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/jns.2019.33
efficient therapy in humans to manage metabolic disorders like
obesity or high blood pressure
(37)
and rheumatoid arth-
ritis
(38,39)
. Yet, the present study was performed on healthy
subjects and further studies are therefore needed to determine
the effects of fasting, and their persistence, on the gut micro-
biome of patients with specific disorders. Furthermore, fasting
can be considered as a metabolic training switching from glu-
cose to fat and ketose and back
(7)
. Healthy subjects gain con-
fidence that fasting is safe and may be used in the case of
weight gain, metabolic disorders and other diseases which
can be improved by periodic fasting as well as intermittent
fasting. When used with a therapeutic goal a special focus is
made on the maintenance phase fasting to consolidate the
results. Our study lays the foundation for future studies to
investigate the impact of fasting on the microbiota of ill per-
sons and their relationship with metabolic and inflammatory
mediators.
Furthermore, it is the first to explore the links between the
changes in gut microbiome caused by periodic fasting and the
changes in inflammation and gut permeability. During periodic
Buchinger fasting, almost no nutrients are ingested and conse-
quently absorbed by the small intestine. Therefore, a signifi-
cant postprandial increase in gut permeability does not
occur. This is reflected by the lack of changes in gut perme-
ability markers in our study population in accordance with
earlier data
(40)
. However, effects of a periodic fasting on gut
permeability may differ between healthy subjects and patients
with metabolic disorders. In a previous study, a 4-week energy
restriction (800 kcal/d (3347 kJ/d)) in twenty obese women
reduced gut permeability
(41)
. Additionally, the gut barrier pro-
tects the host from influx of bacterial components such as
LPS, components of the outer membrane of Gram-negative
bacteria. A biomarker for the translocation of LPS is the
LBP
(41)
. In our cohort LBP decreased significantly during fast-
ing, probably reflecting the decreased food intake.
Interestingly, although we observed no changes of further
inflammation markers during the fasting period, all inflamma-
tion markers, except for LPS increases significantly after food
reintroduction. This suggests that food intake reactivated the
postprandial immune response.
In our cohort, we observed a decrease in blood leucocyte
count. This was associated with an initial autophagy triggered
by fasting, followed by the activation of bone marrow stem
cells
(42)
. The modulation of the immune response by fasting
is also shown by the increase in faecal lysozyme during fasting
and refeeding, indicating migration of leucocytes into the gut.
This is corroborated by the increase in secretory IgA (sIgA),
known for their protective action on the mucosal surface
(43)
.
An increase in sIgA levels was also demonstrated in a previous
study after 17-d Buchinger fasting, and correlated with
enhanced immune status
(44)
. Changes in bacterial antigens
are described in rheumatoid arthritis
(45)
concomitant to
improvement of the symptoms associated with fasting
(24)
.
The association between E. coli and sIgA levels detected in
our study suggests that the increase in sIgA levels during fast-
ing is due to the increased abundance in bacteria causing an
adaptive humoral local response. This is corroborated by
recent studies which have shown that commensal
Proteobacteria promote a T cell-dependent increase in serum
IgA in mice, conferring protection against sepsis
(46)
. In mice,
intermittent fasting increases resistance to Salmonella infec-
tion
(47)
. Since sIgA are known to be the principal eosinophil
1234
Time point
1234
Time point
12 34
Time point
1234
Time point
1
0
2
4
6
2·5
2·0
1·5
1·0
0·5
0·0
1·5
1·0
0·5
0·0
1·5
(D)
1·0
0·5
0·0
0
1
2
Ruminococcus_2 bromii (%)
Faecalibacterium prausnitzii (%)
Coprococcus_2 eutactus (%)
Lachnospira pectinoschiza (%)
Fusicatenibacter saccharivorans (%)Bacteroides fragilis SV_378 (%)
Roseburia faecis (%)
Bacteroides fragilis SV_96 (%)
Bacteroides nordii (%)
Ruminicoccus_1 bicirculans (%)
4
3
0
1
2
3
0
1
2
3
8
6
4
2
0
5
234
Time point
12 34
Time point
1
0
1
2
0
1
2
3
234
Time point
12 34
Time point
1234
Time point
12 34
Time point
Fig. 3. Continued.
9
journals.cambridge.org/jns
Downloaded from https://www.cambridge.org/core. IP address: 80.148.22.98, on 13 Nov 2019 at 06:48:18, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/jns.2019.33
mediator at mucosal surfaces, causing eosinophil degranula-
tion
(48)
, we could hypothesise that the migration of leucocytes
into the gut can also explain the increase in EDN levels during
fasting. Altogether, general aspects of the communication
between micro-organisms of the digestive tract and the
immune system are well described
(49)
.
Although Proteobacteria are generally regarded to have a
negative influence on physiological function
(50)
, with Bilophila
wadsworthia linked to the development of inflammatory bowel
disease
(51)
, it is important to note that effects are strain-
dependent. For instance, the probiotic E. coli Nissle 1917
strain can protect against the invasion by adherent-invasive
E. coli B2 strains
(52)
. It is thus not possible to definitely attri-
bute beneficial or detrimental health effects to the increase
in Proteobacteria abundance observed in our study since the
method we used does not allow the determination of gut
microbiome composition at the strain level. We recommend
new investigations using shotgun metagenomics
(53)
Moreover, the absence of food intake and thus of digestive
processes and absorption during fasting should be taken into
consideration when evaluating possible health effects of a bac-
terial strain. The assessment of physical well-being and
gastrointestinal symptomatic outcomes in another study did
not indicate a pathological situation
(4)
.
The enhancement of well-being is one of the most contra-
intuitive effects of voluntary fasting, whereas involuntary
food deprivation is generally experienced as dreadful. The
mechanisms underlying this mood enhancement might be
similar to those by which physical exercise can induce benefi-
cial responses in the brain, leading to improvements in the
processes of learning and memory formation
(54)
. A major con-
tributor is the modulation of brain-derived neurotrophic factor
(BDNF) levels by the ketone 3-hydroxybutyrate, as shown on
cultured cerebral cortical neurons via an effect on mitochon-
drial respiration
(55)
. Increasing levels of BDNF levels are also
linked to 5-hydroxytryptamine serum levels which are known
to increase during fasting, and which thus provide an explan-
ation for positive effects on mood
(53)
. It could also be hypothe-
sised that fasting-induced mood enhancement can be linked to
changes in the gut microbiota composition. This hypothesis is
corroborated by studies showing that the gut microbiota sup-
presses Bdnf expression in the hypothalamus in mice
(56)
.
Moreover, the administration of probiotics is a successful strat-
egy to reduce anxiety and indicates that changes in microbiome
Table 3. Serum and faecal biochemistry
(Mean values and standard deviations)
Parameter
Pre-fasting End of fasting 3 d after After 3 months
Mean SD Mean SD PMean SD PMean SD P
Serum biochemistry
Erythrocytes (10
6
/μl) 5·22 0·55 5·16 0·40 NS 5·03 0·41 NS 5·23 0·48 NS
Hb (mmol/l) 9·48 0·76 9·36 0·45 NS 9·10 0·52 * 9·39 0·59 NS
Haematocrit (%) 44·90 3·27 43·64 2·33 NS 43·75 2·93 NS 44·79 3·00 NS
Leucocytes (10
3
/μl) 6·01 1·38 4·37 1·18 *** 3·98 0·93 *** 5·51·45 NS
MCV (fl) 86·34 4·19 84·75 3·35 NS 87·09 3·06 NS 85·89 4·25 NS
Thrombocytes (10
3
/μl) 234·93 42·28 247·53 41·85 NS 247·13 52·04 NS 258·43 42·36 *
Inflammation parameters in blood
IFNγ(pg/ml) 78·77 89·43 107·36 70·09 NS 173·95 43·82 *** 43·04 21·34 NS
IL-10 (pg/ml) 24·47 14·05 19·67 7·91 NS 34·85 12·06 *** 21·88 12·99 NS
IL-6 (pg/ml) 12·57 10·617·83 7·89 NS 28·42 7·01 *** 13·69 7·03 NS
TNFα(pg/ml) 71·68 79·26 111·03 71·82 NS 165·01 56·9 *** 40·41 21·93 NS
LBP (μg/l) 11·82 2·51 9·78 3·80 ** 9·58 2·62 ** 11·01 1·72 NS
LPS (pg/ml) 15·03 16·29 12·23 13·35 NS 8·93 11·11 NS 8·88 11·65 NS
Inflammation parameters in faeces
sIgA (ng/ml) 1860·35 1557·2 2812·33 2544·79 NS 3807·01 3013·58 ** 1642·06 1148·3NS
EDN (ng/ml) 404·69 305·63 1176·75 708·56 *** 676·6 701·13 NS 437·86 360·56 NS
β-Defensin (ng/ml) 39·17 47·39 61·23 50·28 NS 30·78 27·31 NS 56·46 89·87 NS
Lysozyme (ng/ml) 555·93 304·06 961·33 572·01 * 1183 658·06 ** 532·07 236·42 NS
Lactoferrin (μg/g) 1·71·74·312·1NS 4·812·6NS 1·81·8NS
Calprotectin (μg/g) 29·322·240·771·2NS 69·3 180·6NS 26·914·2NS
Bile acid (μmol/100 ml) 287·84 231·04 192·11 131·64 NS 212·13 117·87 NS 291·22 235·97 NS
Gut permeability parameters in faeces
Zonulin (ng/ml) 74·21 47·69 95·94 107·33 NS 94·13 114·01 NS 107·50 76·05 NS
α-1-Antitrypsin (mg/l) 394 232 520 343 NS 441 389 NS 401 218 NS
Microbial energy metabolism
Acetate (mM)63·848·354·333·6NS 65·958·3 NS 100·959·4*
Butyrate (mM)6·76·74·83·1NS 4·24·8NS 15·113·2**
i-Butyrate (mM)1·81·81·41·1NS 0·70·7** 1·71·1NS
Propionate (mM)9·57·48·54·7NS 9·610·7NS 17·312·8*
Valerate (mM)2·72·62·01·6NS 1·01·2** 2·61·9NS
i-Valerate (mM)1·10·81·00·5NS 0·60·8NS 1·81·2**
BCAA (μmol/l) 486·58 86·66 575·09 99·24 *** 455·51 67·79 NS 403·95 70·35 ***
MCV, mean corpuscular volume; IFNγ, interferon γ; LBP, lipopolysaccharide-binding protein; LPS, lipopolysaccharides; sIgA, secretory IgA, EDN, eosinophil-derived neurotoxin;
BCAA, branched-chain amino acids.
Significantly different at end of fasting in comparison with pre-fasting baseline levels: *P<0·05, **P<0·01, ***P<0·001 (ANOVA).
10
journals.cambridge.org/jns
Downloaded from https://www.cambridge.org/core. IP address: 80.148.22.98, on 13 Nov 2019 at 06:48:18, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/jns.2019.33
profiles have a direct impact on mood
(57)
. As far as well-being
during fasting is concerned, additional scientific investigations
are needed to understand the role of enemas in preventing
symptoms which are often observed at the beginning of the
fasting such as headaches and fatigue. Since enemas are
known to affect gut microbiome composition
(58)
,itwillbe
important to separate the effects of the enema from the effects
of fasting in future studies. Furthermore, an interesting question
is whether the observed well-being-enhancing effect of fasting is
mainly due to the assumed ability of an enema to remove intes-
tinal remnants, desquamated mucosal cells and fasting basal
secretions from the gut or if it results from an effect of the
enema on gut microbiota composition.
The gut microbiota may also contribute to weight loss. In our
study, the taxonomic group which was most affected by the
reduction of nutrient supply was Lachnospiraceae, which is a
bacterial family known to have the highest energy-harvesting
properties
(59,60)
. Studies demonstrated that a 20 % increase in
Firmicutes (i.e. Lachnospiraceae) was associated with an
increased energy harvest of about 150 kcal (628 kJ)
(61)
.Since
the intrinsic production of energy from the Firmicutes comes
in addition to the general energy intake, it can be hypothesised
that shutting down Lachnospiraceae-mediated increase of
energy uptake had an influence on the weight decrease mea-
sured in this study. This is corroborated by the association
between Lachnospiraceae abundance and glucose levels. The
Fig. 4. Abundance of bacteria affected by fasting is associated with changes in health biomarkers. The abundance of bacterial species identified by 16S rRNA
sequencing was used as a predictor in linear mixed models to understand if they associate with health biomarkers. (A) Statistically significant associations were
found between markers of the energy metabolism switch and fasting-affected species. A total of five biochemical parameters are displayed along the bacteria asso-
ciated with their variations. All arrows indicate statistically significant associations (e.g. a decrease in glucose levels was associated with a decrease in
Lachnospiraceae abundance). (B) The abundance in Lachnospiraceae Coprococcus_2 eutactus (SV_299), Fusicatenibacter saccharivorans (SV_1010) and
Lachnospira pectinoschiza (SV_721) were consistently positively associated with plasma glucose levels, and negatively associated with branched-chain amino
acid (BCAA) levels. By opposition, the Bacteroidaceae Bacteroides dorei/fragilis (SV_96) and Bacteroides thetaiotaomicron (SV_600), as well as Bilophila wads-
worthia, presented the opposite trend and were negatively associated with plasma glucose levels, and positively associated with BCAA levels. EDN, eosinophil-
derived neurotoxin; sIgA, secretory IgA.
11
journals.cambridge.org/jns
Downloaded from https://www.cambridge.org/core. IP address: 80.148.22.98, on 13 Nov 2019 at 06:48:18, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/jns.2019.33
micro-organisms inhabiting the human gastrointestinal tract
have the ability to ferment indigestible complex carbohydrates
and produce energy substrates such as SCFA which are then
taken up by the host. Up to 5 to 10 % of human energy require-
ments are covered by SCFA
(62)
. In general, SCFA concentra-
tions remained stable during fasting which is surprising since
they are mainly produced by fermentation of dietary fibres
(63)
.
Furthermore, an increase in SCFA levels was noticed 3 months
after the intervention. Whether this is a consequence of the gut
microbiota structural changes due to fasting or a change in diet-
ary habits must be confirmed. Food protocols documented an
increase in fibre intake 3 months after fasting (Supplementary
Table S2).
The association between Bacteroidetes abundance and
faecal BCAA levels in our study suggests that energy require-
ments in absence of dietary nutrients during fasting could be
sustained by the use of host-derived compounds such as des-
quamated cells. The species which have their levels increased
during fasting in our study are also those species known to
have the highest proteolytic activity in the gut microbiota
(64)
.
The clinical significance of this switch in gut microbiome
metabolism is unclear. Experiments in mice showed that the
gut microbiome starts feeding on host mucins and degrade
the colonic mucus barrier when it is deprived of dietary
fibres
(65)
. Little is known about the fate of the mucus barrier
during fasting in humans. Starved broilers (also deprived of
water) have a thinner mucus adherent layer throughout the
small intestine
(66)
. However, this might not reflect physiology
during prolonged fasting since the gut is known to undertake
substantial structural changes
(11)
. It should also be taken into
account that the microbial density is severely reduced during
fasting. During hibernation, microbial density decreases by
93·7 % in the caecum of hamsters
(15)
. This cannot be
evaluated in our study because the sequencing of 16S rRNA
gene amplicons is not quantitative, and because faecal micro-
biota is not fully reflecting the intestinal microbiota.
In conclusion, 10 d of fasting led to a profound change
within the gut microbiota composition and function. The
reductions in body weight and waist circumference were asso-
ciated with an enhancement of well-being and an improvement
of energy metabolism. Refeeding induced an immune reaction,
as shown by circulating cytokines. This study on healthy sub-
jects lays the foundation for further investigations on the
impact of fasting on the microbiota and their relationship
with metabolic and inflammatory mediators in diet-related
chronic diseases.
Acknowledgements
We thank the study patients for their participation and the
physicians and nursing staff of the BWC for their cooperation.
The present study was financed by Amplius GmbH,
Überlingen, Germany. This company has the task to develop
a research department for the BWC Überlingen and
Marbella who are the funders. Amplius GmbH had no role
in the design, analysis or writing of this article. No additional
external funding was received for this study.
F. W. T., F. G. and Y. L. M. conceived and conceptualised
the study. A. S. provided expert input into the study design
and performed parts of the laboratory analyses (sequencing
and all faecal parameters). F. G. was project manager and
coordinated study conduction and data collection. R. M. coor-
dinated the writing of the paper and performed the bioinfor-
matics and statistical analysis. R. M., F. G. and F. W. T.
drafted the manuscript. All authors contributed to data inter-
pretation and the revision and editing of the final manuscript.
F. W. T. is managing director of Amplius GmbH, in charge
of the scientific documentation for the BWC. Amplius GmbH
is a company that conceives, coordinates and develops fasting
research. All authors declare that no competing interests exist.
Supplementary material
The supplementary material for this article can be found at
https://doi.org/10.1017/jns.2019.33
References
1. Prentice AM, Whitehead RG, Roberts SB, et al. (1981) Long-term
energy balance in child-bearing Gambian women. Am J Clin Nutr
34, 2790–2799.
2. Venegas-Borsellino C & Martindale RG (2018) From religion to
secularism: the benefits of fasting. Curr Nutr Rep 7, 131–138.
3. Mattson MP, Longo VD & Harvie M (2017) Impact of intermittent
fasting on health and disease processes. Ageing Res Rev 39,46–58.
4. Wilhelmi de Toledo F, Grundler F, Bergouignan A, et al. (2019)
Safety, health improvement and well-being during a 4 to 21-day
fasting period in an observational study including 1422 subjects.
PLOS ONE 14, e0209353.
5. Wilhelmi de Toledo F, Buchinger A, Burggrabe H, et al. (2013)
Fasting therapy –an expert panel update of the 2002 consensus
guidelines. Res Complement Med 20, 434–443.
6. Carlson MG, Snead WL & Campbell PJ (1994) Fuel and energy
metabolism in fasting humans. Am J Clin Nutr 60,29–36.
7. Mattson MP, Moehl K, Ghena N, et al. (2018) Intermittent meta-
bolic switching, neuroplasticity and brain health. Nat Rev Neurosci
19,63–80.
8. Longo VD & Mattson MP (2014) Fasting: molecular mechanisms
and clinical applications. Cell Metab 19, 181–192.
9. Cahill GF Jr (1970) Starvation in man. N Engl J Med 282, 668–675.
10. Habold C, Reichardt F, Foltzer-Jourdainne C, et al. (2007)
Morphological changes of the rat intestinal lining in relation to
body stores depletion during fasting and after refeeding. Pflügers
Arch 455, 323–332.
11. Dunel-Erb S, Chevalier C, Laurent P, et al. (2001) Restoration of
the jejunal mucosa in rats refed after prolonged fasting. Comp
Biochem Physiol A: Mol Integr Physiol 129, 933–947.
12. Kohl KD, Amaya J, Passement CA, et al. (2014) Unique and shared
responses of the gut microbiota to prolonged fasting: a comparative
study across five classes of vertebrate hosts. FEMS Microbiol Ecol
90, 883–894.
13. McCue MD (2012) An introduction to fasting, starvation, and food
limitation. In Comparative Physiology of Fasting, Starvation, and Food
Limitation, pp. 1–6. Heidelberg: Springer.
14. Riedl RA, Atkinson SN, Burnett CM, et al. (2017) The gut micro-
biome, energy homeostasis, and implications for hypertension. Curr
Hypertens Rep 19, 27.
15. Sonoyama K, Fujiwara R, Takemura N, et al. (2009) Response of
gut microbiota to fasting and hibernation in Syrian hamsters.
Appl Environ Microbiol 75, 6451–6456.
16. Stevenson TJ, Duddleston KN & Buck CL (2014) Effects of sea-
son and host physiological state on the diversity, density, and
12
journals.cambridge.org/jns
Downloaded from https://www.cambridge.org/core. IP address: 80.148.22.98, on 13 Nov 2019 at 06:48:18, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/jns.2019.33
activity of the arctic ground squirrel cecal microbiota. Appl Environ
Microbiol 80, 5611–5622.
17. Sommer F, Ståhlman M, Ilkayeva O, et al. (2016) The gut micro-
biota modulates energy metabolism in the hibernating brown bear
Ursus arctos.Cell Rep 14, 1655–1661.
18. Carey HV & Assadi-Porter FM (2017) The hibernator microbiome:
host–bacterial interactions in an extreme nutritional symbiosis.
Annu Rev Nutr 37, 477–500.
19. Remely M, Hippe B, Geretschlaeger I, et al. (2015) Increased gut
microbiota diversity and abundance of Faecalibacterium prausnitzii
and Akkermansia after fasting: a pilot study. Wien Klin Wochenschr
127, 394–398.
20. Jabbar A, Chang W-K, Dryden GW, et al. (2003) Gut immunology
and the differential response to feeding and starvation. Nutr Clin
Pract 18, 461–482.
21. Thaiss CA, Zmora N, Levy M, et al. (2016) The microbiome and
innate immunity. Nature 535,65–74.
22. Beli E, Yan Y, Moldovan L, et al. (2018) Restructuring of the gut
microbiome by intermittent fasting prevents retinopathy and pro-
longs survival in db/db mice. Diabetes 67, 1867–1879.
23. Mohajeri MH, Brummer RJM, Rastall RA, et al. (2018) The role of
the microbiome for human health: from basic science to clinical
applications. Eur J Nutr 57, Suppl. 1, 1–14.
24. Peltonen R, Kjeldsen-Kragh J, Haugen M, et al. (1994) Changes of
faecal flora in rheumatoid arthritis during fasting and one-year
vegetarian diet. Br J Rheumatol 33, 638–643.
25. Callahan BJ, McMurdie PJ, Rosen MJ, et al. (2016) DADA2:
high-resolution sample inference from Illumina amplicon data.
Nat Methods 13, 581–583.
26. Ohlsson B, Roth B, Larsson E, et al. (2017) Calprotectin in serum
and zonulin in serum and feces are elevated after introduction of a
diet with lower carbohydrate content and higher fiber, fat and
protein contents. Biomed Rep 6, 411–422.
27. Unger MM, Spiegel J, Dillmann K-U, et al. (2016) Short chain fatty
acids and gut microbiota differ between patients with Parkinson’sdis-
ease and age-matched controls. Parkinsonism Relat Disord 32,66–72.
28. D’Amore R, Ijaz UZ, Schirmer M, et al. (2016) A comprehensive
benchmarking study of protocols and sequencing platforms for
16S rRNA community profiling. BMC Genomics 17, 55.
29. Claesson MJ, O’Sullivan O, Wang Q, et al. (2009) Comparative ana-
lysis of pyrosequencing and a phylogenetic microarray for exploring
microbial community structures in the human distal intestine. PLoS
ONE 4, e6669.
30. Roy AS & Bandyopadhyay A (2015) Effect of Ramadan
intermittent fasting on selective fitness profile parameters in
young untrained Muslim men. BMJ Open Sport Exerc Med 1,
e000020.
31. Morgan XC & Huttenhower C (2012) Chapter 12: human micro-
biome analysis. PLoS Comput Biol 8, e1002808.
32. McMurdie PJ & Holmes S (2013) Phyloseq: an R package for
reproducible interactive analysis and graphics of microbiome census
data. PLOS ONE 8, e61217.
33. Benjamini Y & Hochberg Y (1995) Controlling the false discovery
rate: a practical and powerful approach to multiple testing. J R Stat
Soc 57, 289–300.
34. Valdes AM, Walter J, Segal E, et al. (2018) Role of the gut micro-
biota in nutrition and health. BMJ 361, k2179.
35. Zarrinpar A, Chaix A, Yooseph S, et al. (2014) Diet and feeding pat-
tern affect the diurnal dynamics of the gut microbiome. Cell Metab
20, 1006–1017.
36. Mosca A, Leclerc M & Hugot JP (2016) Gut microbiota diversity
and human diseases: should we reintroduce key predators in our
ecosystem? Front Microbiol 7, 455.
37. Michalsen A & Li C (2013) Fasting therapy for treating and pre-
venting disease –current state of evidence. Res Complement Med
20, 444–453.
38. Kjeldsen-Kragh J, Borchgrevink C, Laerum E, et al. (1991)
Controlled trial of fasting and one-year vegetarian diet in rheuma-
toid arthritis. Lancet 338, 899–902.
39. Müller H, de Toledo FW & Resch K-L (2001) Fasting followed by
vegetarian diet in patients with rheumatoid arthritis: a systematic
review. Scand J Rheumatol 30,1–10.
40. Maxton D, Menzies I, Slavin B, et al. (1989) Small-intestinal func-
tion during enteral feeding and starvation in man. Clin Sci 77,
401–406.
41. Ott B, Skurk T, Hastreiter L, et al. (2017) Effect of caloric restric-
tion on gut permeability, inflammation markers, and fecal micro-
biota in obese women. Sci Rep 7, 11955.
42. Cheng C-W, Adams GB, Perin L, et al. (2014) Prolonged fasting
reduces IGF-1/PKA to promote hematopoietic-stem-cell-based
regeneration and reverse immunosuppression. Cell Stem Cell 14,
810–823.
43. Wells JM, Brummer RJ, Derrien M, et al. (2017) Homeostasis of the
gut barrier and potential biomarkers. Am J Physiol Gastrointest Liver
Physiol 312, G171–G193.
44. Beer A-M, Rüffer A, Balles J, et al. (2001) Verläufe des sekretor-
ischen Immunglobulins A des Darms und Befindlichkeit von
Patienten unter naturheilkundlicher Therapie und Heilfasten
(Course of secretory immunoglobulin A of the intestine and
condition of patients under naturopathic therapy and fasting).
Res Complement Med 8, 346–353.
45. Kjeldsen-Kragh J, Rashid T, Dybwad A, et al. (1995) Decrease in
anti-Proteus mirabilis but not anti-Escherichia coli antibody levels in
rheumatoid arthritis patients treated with fasting and a one year
vegetarian diet. Ann Rheum Dis 54, 221–224.
46. Wilmore JR, Gaudette BT, Atria DG, et al. (2018) Commensal
microbes induce serum IgA responses that protect against polymi-
crobial sepsis. Cell Host Microbe 23, 302–311.e3.
47. Campos-Rodríguez R, Godínez-Victoria M, Reyna-Garfias H, et al.
(2016) Intermittent fasting favored the resolution of Salmonella
typhimurium infection in middle-aged BALB/c mice. Age (Omaha)
38, 13.
48. Abu-Ghazaleh R, Fujisawa T, Mestecky J, et al. (1989) IgA-induced
eosinophil degranulation. J Immunol 142, 2393–2400.
49. Kim D, Zeng MY & Núñez G (2017) The interplay between host
immune cells and gut microbiota in chronic inflammatory diseases.
Exp Mol Med 49, e339.
50. Roberfroid M, Gibson GR, Hoyles L, et al. (2010) Prebiotic effects:
metabolic and health benefits. Br J Nutr 104, Suppl. 2, S1–S63.
51. Devkota S, Wang Y, Musch MW, et al. (2012) Dietary-fat-induced
taurocholic acid promotes pathobiont expansion and colitis in
Il10
−/−
mice. Nature 487, 104–108.
52. Sassone-Corsi M, Nuccio S-P, Liu H, et al. (2016) Microcins medi-
ate competition among Enterobacteriaceae in the inflamed gut.
Nature 540, 280–283.
53. Segata N (2018) On the road to strain-resolved comparative meta-
genomics. MSystems 3, e00190–e00117.
54. Sleiman SF, Henry J, Al-Haddad R, et al. (2016) Exercise promotes
the expression of brain derived neurotrophic factor (BDNF)
through the action of the ketone body β-hydroxybutyrate. Elife 5,
e15092.
55. Marosi K, Kim SW, Moehl K, et al. (2016) 3-Hydroxybutyrate reg-
ulates energy metabolism and induces BDNF expression in cerebral
cortical neurons. J Neurochem 139, 769–781.
56. Schéle E, Grahnemo L, Anesten F, et al. (2013) The gut microbiota
reduces leptin sensitivity and the expression of the obesity-
suppressing neuropeptides proglucagon (Gcg) and brain-derived
neurotrophic factor (Bdnf) in the central nervous system.
Endocrinology 154, 3643–3651.
57. Messaoudi M, Lalonde R, Violle N, et al. (2011) Assessment of
psychotropic-like properties of a probiotic formulation
(Lactobacillus helveticus R0052 and Bifidobacterium longum R0175) in
rats and human subjects. Br J Nutr 105, 755–764.
58. Jalanka J, Salonen A, Salojärvi J, et al. (2015) Effects of bowel
cleansing on the intestinal microbiota. Gut 64, 1562–1568.
59. Shabat SKB, Sasson G, Doron-Faigenboim A, et al. (2016) Specific
microbiome-dependent mechanisms underlie the energy harvest
efficiency of ruminants. ISME J 10, 2958–2972.
13
journals.cambridge.org/jns
Downloaded from https://www.cambridge.org/core. IP address: 80.148.22.98, on 13 Nov 2019 at 06:48:18, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/jns.2019.33
60. Méndez-Salazar EO, Ortiz-López MG, de los Ángeles
Granados-Silvestre M, et al. (2018) Altered gut microbiota and
compositional changes in Firmicutes and Proteobacteria in
Mexican undernourished and obese children. Front Microbiol 9,
2494.
61. Jumpertz R, Le DS, Turnbaugh PJ, et al. (2011) Energy-
balance studies reveal associations between gut microbes,
caloric load, and nutrient absorption in humans. Am J Clin Nutr
94,58–65.
62. McNeil N (1984) The contribution of the large intestine to energy
supplies in man. Am J Clin Nutr 39, 338–342.
63. Brial F, Le Lay A, Dumas M-E, et al. (2018) Implication of gut
microbiota metabolites in cardiovascular and metabolic diseases.
Cell Mol Life Sci 75, 3977–3990.
64. Macfarlane G, Allison C, Gibson S, et al. (1988) Contribution of the
microflora to proteolysis in the human large intestine. J Appl
Bacteriol 64,37–46.
65. Desai MS, Seekatz AM, Koropatkin NM, et al. (2016) A dietary
fiber-deprived gut microbiota degrades the colonic mucus barrier
and enhances pathogen susceptibility. Cell 167, 1339–1353.21.
66. Smirnov A, Sklan D & Uni Z (2004) Mucin dynamics in the chick
small intestine are altered by starvation. J Nutr 134, 736–742.
14
journals.cambridge.org/jns
Downloaded from https://www.cambridge.org/core. IP address: 80.148.22.98, on 13 Nov 2019 at 06:48:18, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/jns.2019.33
