A purified membrane protein from Akkermansia muciniphila or the 1
pasteurized bacterium improves metabolism in obese and diabetic mice 2
3
Hubert Plovier
1
, Amandine Everard
1*
, Céline Druart
1*
, Clara Depommier
1*
, Matthias Van Hul
1
, 4
Lucie Geurts
1
, Julien Chilloux
2
, Noora Ottman
3#
, Thibaut Duparc
4
, Laeticia Lichtenstein
4
, 5
Antonis Myridakis
2
, Nathalie M. Delzenne
1
, Judith Klievink
5
Arnab Bhattacharjee
5
, Kees C.H. 6
van der Ark
3
, Steven Aalvink
3
, Laurent O. Martinez
4
, Marc-Emmanuel Dumas
2
, Dominique 7
Maiter
6
, Audrey Loumaye
6
, Michel P. Hermans
6
, Jean-Paul Thissen
6
, Clara Belzer
3
, Willem M. 8
de Vos
3,5
, Patrice D. Cani
1$
9
10
1
Université catholique de Louvain, Louvain Drug Research Institute, WELBIO (Walloon 11
Excellence in Life sciences and BIOtechnology), Metabolism and Nutrition research group, B-12
1200 Brussels, Belgium, 13
2
Division of Computational and Systems Medicine, Department of Surgery and Cancer, Imperial 14
College London, Exhibition Road, South Kensington, London SW7 2AZ, United Kingdom, 15
3
Laboratory of Microbiology, Wageningen University, Wageningen, The Netherlands.
16
4
Institute of Metabolic and Cardiovascular Diseases, UMR1048, Inserm, Université de Toulouse, 17
Toulouse, France, 18
5
RPU Immunobiology, Department of Bacteriology & Immunology, University of Helsinki, 19
Finland.
20
6
Pole of Endocrinology, Diabetes and Nutrition; Institut de Recherche Expérimentale et Clinique 21
IREC, Cliniques Universitaires Saint-Luc, Université catholique de Louvain, Brussels, Belgium 22
23
* These authors contributed equally to this work 24
#
Current affiliation: Metapopulation Research Centre, University of Helsinki, Helsinki, Finland 25
26
$
Correspondence to: Patrice.cani@uclouvain.be
27
Prof. Patrice D. Cani, Université catholique de Louvain, LDRI, Metabolism and Nutrition 28
research group, Av. E. Mounier, 73 box B1.73.11, B-1200 Brussels, Belgium.
Phone: +32 2 764 29
73 97 30
31
32
33
2
Obesity and type 2 diabetes are associated with low-grade inflammation and specific 34
changes in gut microbiota composition
1-7
. We previously demonstrated that administration 35
of Akkermansia muciniphila prevents the development of obesity and associated 36
complications
8
. However, its mechanisms of action remain unclear, whilst the sensitivity of 37
A. muciniphila to oxygen and the presence of animal-derived compounds in its growth 38
medium currently limit the development of translational approaches for human medicine
9
. 39
Here we addressed these issues by showing that A. muciniphila retains its efficacy when 40
grown on a synthetic medium compatible with human administration. Unexpectedly, we 41
discovered that pasteurization of A. muciniphila enhanced its capacity to reduce fat mass 42
development, insulin resistance and dyslipidemia in mice. These improvements were 43
notably associated with a modulation of the host urinary metabolomics profile and 44
intestinal energy absorption. We demonstrated that Amuc_1100, a specific protein isolated 45
from the outer membrane of A. muciniphila, interacts with Toll-Like Receptor 2, is stable at 46
temperatures used for pasteurization, improves the gut barrier and partly recapitulates the 47
beneficial effects of the bacterium. Finally, we showed that administration of live or 48
pasteurized A. muciniphila grown on the synthetic medium is safe in humansThese findings 49
provide support for the use of different preparations of A. muciniphila as therapeutic 50
options to target human obesity and associated disorders. 51
52
Akkermansia muciniphila is one of the most abundant members of the human gut 53
microbiota, representing between 1 and 5% of our intestinal microbes
10,11
. We and others recently 54
observed that the abundance of Akkermansia muciniphila is decreased during obesity and 55
diabetes
2,8
and is significantly associated with the improvement of cardiometabolic parameters in 56
individuals with obesity undergoing caloric restriction
12
. Moreover, we found that daily 57
administration of live A. muciniphila grown on a mucus-based medium can counteract the 58
development of high-fat diet (HFD)-induced obesity and gut barrier dysfunction
8
, an observation 59
later confirmed by other groups
13,14
. However, the underlying mechanisms of these effects are 60
still unclear. In addition, the current growth requirements of A. muciniphila and its oxygen 61
sensitivity
9
render this bacterium unsuitable for human investigations and putative therapeutic 62
opportunities. 63
3
Therefore, in HFD-fed mice, we compared the effects of daily administration of A. 64
muciniphila grown either on a mucus-based medium (HFD Live Akk Mucus) or a synthetic 65
medium where mucin was replaced by a combination of glucose, N-acetylglucosamine, soy 66
peptone and threonine (HFD Live Akk Synthetic). This medium allowed us to culture A. 67
muciniphila with the same efficiency as the mucus-based medium while being exempt of any 68
compound incompatible with human administration. We observed that live A. muciniphila 69
treatment tended to reduce HFD-induced body weight and fat mass gain (by about 40-50%) and 70
to improve glucose intolerance and insulin resistance regardless of the growth medium used and 71
independently of food intake (Fig. 1a-g). 72
We previously showed that autoclaving A. muciniphila abolished its beneficial effects
8
. 73
However, recent investigations suggest that probiotics inactivated by pasteurization for 30 74
minutes at 70°C, a less extreme treatment limiting the denaturation of their cellular components, 75
could partly or fully retain their beneficial effects
15,16
. Hence, we assessed the effects of A. 76
muciniphila grown on a synthetic medium and inactivated by pasteurization (HFD Pasteurized 77
Akk). Unexpectedly, in two separate sets of experiments, we found that pasteurized A. 78
muciniphila exerted stronger effects than the live bacterium, as HFD-fed mice treated with 79
pasteurized bacteria showed similar body weight and fat mass gain to mice fed with a control diet 80
(ND), independently of food intake (Fig. 1a-c and Supplemental Fig. 1a-c). In both sets of 81
experiments, we found that mice treated with pasteurized A. muciniphila displayed a much lower 82
glucose intolerance and insulin concentration when compared to the HFD group, resulting in a 83
lower insulin resistance (IR) index (Fig. 1d-g and Supplemental Fig. 1d-g). Treatment with 84
pasteurized A. muciniphila also led to greater goblet cell density in the ileum when compared to 85
ND-fed mice (Fig. 1h), suggesting a higher mucus production, while normalizing the mean 86
adipocyte diameter (Fig. 2a-b) and significantly lowering plasma leptin when compared to HFD-87
fed mice (Fig. 2c). These effects were not observed in mice treated with live A. muciniphila. A 88
similar trend could be observed for plasma resistin (Supplemental Fig. 1h), thereby suggesting 89
improved insulin sensitivity, while plasma adiponectin remained unaffected in all conditions 90
(Supplemental Fig. 1i). We found that mice treated with pasteurized A. muciniphila had a higher 91
fecal caloric content when compared to all other groups (Fig. 2d), suggesting a lower energy 92
absorption. This could contribute to the further reduction in body weight and fat mass gain 93
observed in this group. Similar effects of A. muciniphila on energy absorption were previously 94
4
reported in mice undergoing cold exposure
17
. Altogether, these data suggest that pasteurization 95
enhances the beneficial effects of A. muciniphila on HFD-induced metabolic syndrome. This 96
could be due to increased accessibility of specific bacterial compounds involved in the positive 97
effects of A. muciniphila on its host. Conversely, pasteurization of A. muciniphila could prevent 98
the production of metabolites or factors mitigating its beneficial effects. 99
We next tested whether treatment with A. muciniphila could reduce the HFD-induced shift 100
in the host urinary metabolome
18
HFD was the main factor influencing
1
H NMR-based 101
untargeted metabolic profiles on the first O-PLS-DA score (Tpred1) while treatment with 102
pasteurized A. muciniphila clustered separately from all other groups regarding the second score 103
(Tpred2) (Fig. 2e-g). This resulted in a normalization of the HFD-induced shift of 37% with the 104
pasteurized bacterium, and 17% with the live bacterium (Fig. 2f). 105
By comparing the metabolic profiles of the different groups, we found that the shift 106
induced by pasteurized A. muciniphila was mainly associated with trimethylamine (TMA) and 107
trimethylamine-N-oxide (TMAO) according to the OPLS-DA model coefficients (Supplemental 108
Fig. 2). While HFD feeding severely lowered the abundance of TMA compared to ND-fed mice, 109
treatment with pasteurized A. muciniphila significantly offset this reduction (Fig. 2h). A similar 110
trend was observed for urinary TMAO (Fig. 2i). This relative increase in TMA abundance was 111
not observed in mice treated with live A. muciniphila. Treatment with pasteurized A. muciniphila 112
also modulated hepatic expression of Fmo3, encoding the Flavin monooxygenase 3 that converts 113
TMA to TMAO, a metabolite associated with atherosclerosis
19,20
. While exposure to a HFD led 114
to a two-fold higher Fmo3 expression when compared to ND-fed mice, treatment with 115
pasteurized A. muciniphila reversed this effect (Fig. 2j). Of note, knockdown of Fmo3 by the use 116
of specific antisense oligonucleotides can increase serum concentration of TMA
20
and protects 117
mice against the development of atherosclerosis and insulin resistance
21
. Moreover, recent 118
findings suggest that oral administration of live A. muciniphila can impede atherosclerosis 119
development in Apoe
-/-
mice
22
. In our study however, the effects observed on urinary TMA and 120
Fmo3 expression were not associated with a modification of plasma TMA and TMAO, as all 121
HFD-fed group displayed similar concentrations for both metabolites (Fig. 2k,l). This suggests 122
that the observed decrease in Fmo3 expression is not sufficient to inhibit the conversion of TMA 123
in TMAO, and that metabolic effects of pasteurized A. muciniphila are not related to these 124
metabolites. 125
5
Toll-like receptors (TLRs) regulate bacterial recognition, intestinal homeostasis and can 126
also shape the host metabolism
23-25
. To identify how A. muciniphila interacts with its host, we 127
performed in vitro experiments to evaluate its TLR signaling potential. Previous results suggested 128
that A. muciniphila lipopolysaccharide (LPS) differs structurally from that of E. coli and is not a 129
powerful TLR4 agonist
26
. Here, we observed that A. muciniphila specifically activated cells 130
expressing TLR2 (Fig. 3a), but not cells expressing TLR5, TLR9 or the NOD2 receptor (Fig. 3b-131
d). 132
Genomic and proteomic analyses of A. muciniphila identified proteins encoded by a 133
specific Type IV pili gene cluster in fractions enriched for outer membrane proteins
27
. Among 134
these, Amuc_1100 was one of the most abundant. Additionally, its presence on a gene cluster 135
related to pilus formation suggests that it could be involved in the crosstalk with the host. To test 136
this hypothesis, we showed that a His-tagged Amuc_1100 produced in E. coli (hereafter called 137
Amuc_1100*) could signal to TLR2-expressing cells in a similar manner to A. muciniphila (Fig. 138
3a). Furthermore, Amuc_1100* appeared relatively thermostable as differential light scattering 139
analysis indicated its melting temperature was 70°C (Fig. 3e), which is the temperature applied 140
for pasteurization. Therefore, Amuc_1100 could still be active in pasteurized bacteria and 141
contribute to the observed signaling. 142
Consequently, we compared the effects of the live and pasteurized bacterium to those of 143
Amuc_1100* in HFD-fed mice. Similarly to what was observed with the pasteurized bacterium, 144
treatment with Amuc_1100* induced a lower body weight and fat mass gain when compared to 145
untreated HFD-fed mice, independently of food intake (Fig. 3f-h). It also tended to correct HFD-146
induced higher adipocyte diameter (Supplemental Fig. 3a,b). Treatment with A. muciniphila or 147
Amuc_1100* corrected HFD-induced hypercholesterolemia, with significantly lower plasma 148
HDL-cholesterol concentrations and a similar trend for LDL-cholesterol (Fig. 3i). Mice treated 149
with pasteurized A. muciniphila displayed significantly lower plasma triglycerides concentrations 150
when compared to either untreated mice or HFD-fed mice treated with live A. muciniphila, again 151
suggesting an increased potency of the bacterium after pasteurization (Fig. 3j). No differences 152
were observed in the distribution of triglycerides and cholesterol in lipoproteins (Supplemental 153
Fig. 3c-d). Amuc_1100* also improved glucose tolerance with the same potency as the live and 154
pasteurized bacterium (Fig. 3k,l). To further investigate the effects of A. muciniphila on insulin 155
sensitivity, we analyzed insulin-induced phosphorylation of the insulin receptor (IR) and its 156