UEF//eRepository
DSpace https://erepo.uef.fi
Rinnakkaistallenteet Terveystieteiden tiedekunta
2020
Food ingredients in human health:
Ecological and metabolic perspectives implicating gut microbiota function
Wu, Qinglong
Elsevier BV
Tieteelliset aikakauslehtiartikkelit
© 2020 Elsevier BV
CC BY-NC-ND https://creativecommons.org/licenses/by-nc-nd/4.0/
http://dx.doi.org/10.1016/j.tifs.2020.04.007
https://erepo.uef.fi/handle/123456789/8233
Downloaded from University of Eastern Finland's eRepository
Journal Pre-proof
Food ingredients in human health: Ecological and metabolic perspectives implicating gut microbiota function
Qinglong Wu, Tingtao Chen, Hani El-Nezami, Tor C. Savidge
PII: S0924-2244(20)30436-2
DOI: https://doi.org/10.1016/j.tifs.2020.04.007 Reference: TIFS 2820
To appear in: Trends in Food Science & Technology Received Date: 24 October 2019
Revised Date: 30 March 2020 Accepted Date: 11 April 2020
Please cite this article as: Wu, Q., Chen, T., El-Nezami, H., Savidge, T.C., Food ingredients in human health: Ecological and metabolic perspectives implicating gut microbiota function, Trends in Food Science & Technology (2020), doi: https://doi.org/10.1016/j.tifs.2020.04.007.
This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version will undergo additional copyediting, typesetting and review before it is published in its final form, but we are providing this version to give early visibility of the article. Please note that, during the production process, errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
© 2020 Published by Elsevier Ltd.
Review
Food ingredients in human health: Ecological and metabolic perspectives implicating gut microbiota function
Qinglong Wua,b, Tingtao Chenc, Hani El-Nezamid,e,*, Tor C. Savidgea,b,*
a Department of Pathology and Immunology, Baylor College of Medicine, Houston, TX 77030, USA
b Texas Children’s Microbiome Center, Department of Pathology, Texas Children’s Hospital, Houston, TX 77030, USA
c Institute of Translational Medicine, Nanchang University, Nanchang, Jiangxi 330031, China
d School of Biological Sciences, The University of Hong Kong, Pokfulam, Hong Kong S.A.R., China
e Institute of Public Health and Clinical Nutrition, University of Eastern Finland, FI-70211 Kuopio, Finland
* Correspondence: Tor.Savidge@bcm.edu (T.C.S.) and elnezami@hku.hk (H.E.N.)
Highlights
1) Dietary imbalance associated with consumption of processed foods alters gut microbiome homeostasis
2) Perturbation of gut microbiome composition and function is mediated by specific food ingredients
3) Technical recommendations for evaluating food ingredient effects on gut microbiome 4) Potential strategies for reversing diet/food ingredient-induced gut dysbiosis
5) Consider gut microbiome as part of safety assessments of food ingredients
Abstract 1
Background: Dietary imbalance and harmful food substances are well established risk factors 2
that can adversely impact human health. The gut microbiome is emerging as a new metabolic 3
organ that can be serendipitously linked to these poor dietary outcomes. Modern manufactured 4
foods and process ingredients can significantly alter gut microbiome composition and function, 5
leading investigators to conclude that disruption of host-microbiome commensalism is a key 6
mechanism in human disease linked to imbalanced diets or processed foods.
7
Scope and Approach: In this review, we highlight disease-associated perturbations and 8
precision manipulation of the human gut microbiome in the context of food and nutrition. We 9
detail technical recommendations for evaluating food ingredient effects on gut microbiome 10
composition and function, following Koch’s principles of commensal postulates to evaluate 11
whether reported findings can be implicated in disease causation. Strategies for minimizing 12
and/or reversing diet/food ingredient-induced gut dysbiosis linked with human disease are also 13
considered.
14
Key Findings and Conclusions: Current findings should encourage us to reevaluate the potential 15
health hazards of modern food ingredients and processed foods by considering the 16
involvement of the gut microbiome through well-established pipelines for deciphering the 17
causality of a perturbated gut microbiome in disease induction.
18
Keywords: gut microbiome; diet; food ingredients; microbiome metabolism; host health 19
1. Introduction 20
From the standpoint of modern food science and technology, our diet commonly involves 21
mixing natural ingredients (agricultural products and by-products) with artificial substances 22
(preservatives, decorative food additives, fortifying agents, etc.) to improve the quality and 23
functionality of processed foods during manufacture (Carocho, Barreiro, Morales, & Ferreira, 24
2014; Carocho, Morales, & Ferreira, 2015). Termed “processed foods”, these popular modern 25
diets are frequently implicated in long-term consequences on human health and development 26
(Botelho, Araújo, & Pineli, 2018; Monteiro, Moubarac, Cannon, Ng, & Popkin, 2013; Moodie et 27
al., 2013; Stuckler & Nestle, 2012). For the purpose of this review, food and diet are collectively 28
considered as ingredients ingested by an individual and processed by the gastrointestinal (GI) 29
tract. The GI tract has naturally evolved to digest and absorb the foods that we eat by facilitating 30
coordinated transit and autonomic contractions that mix dietary ingredients with digestive fluids 31
secreted by the gut, gallbladder and pancreas. Food substances are also exposed to the 32
consortia of microbes (including bacteria, fungi and viruses), termed the microbiome, that aid in 33
nutrient digestion and biotransformation (Sonnenburg & Bäckhed, 2016).
34
Over the last two decades, tremendous effort has been invested in elucidating the role of 35
nutrient-microbiome cross-talk with the host, often termed functional foods. These nutrient-host- 36
microbiome interactions have co-evolved over millennia, for example by developing tolerant 37
host defense systems, i.e. innate (Thaiss, Zmora, Levy, & Elinav, 2016), adaptive (Cullender et 38
al., 2013), and mucosal immunity (Mcdermott & Huffnagle, 2014) to avoid exacerbated immune 39
responses to dietary components in genetically susceptible individuals. In infancy, this field 40
emerged from studies identifying adverse mucosal immune responses, for example to dietary 41
gluten (Galipeau & Verdu, 2015; Sapone et al., 2011) or transient cow’s milk protein allergy in 42
young children (Walker-Smith, 2003). Since these early pioneering studies, our understanding 43
of nutrient-host-microbiome commensalism has moved in leaps and bounds to now implicate 44
specific microbe-derived metabolite classes in diverse human diseases, for example, succinate 45
for colonic epithelial gluconeogenesis (De Vadder et al., 2016), prevention of diabetes by short- 46
chain fatty acids (SCFAs) (Wen & Wong, 2017; Zhao et al., 2018), and bile acid homeostasis 47
linked to errors in host metabolism (Shapiro, Kolodziejczyk, Halstuch, & Elinav, 2018;
48
Wahlström, Sayin, Marschall, & Bäckhed, 2016). Moreover, the association of altered gut 49
microbiome composition with host phenotypes, i.e. obesity (Christensen, Roager, Astrup, &
50
Hjorth, 2018; Vangay et al., 2018), diabetes (Vatanen et al., 2018; Zhao et al., 2018), 51
hypertension and cardiovascular disease (Brown & Hazen, 2018; Marques, Mackay, & Kaye, 52
2018; Wilck et al., 2017), inflammatory bowel diseases (Khalili et al., 2018; Zhu et al., 2018), 53
and cancer (Goodman & Gardner, 2018; Shono & Van Den Brink, 2018; Zitvogel, Ayyoub, 54
Routy, & Kroemer, 2016), has been proposed as pathogenic and in some cases is supported by 55
Koch's commensal postulates (Neville, Forster, & Lawley, 2018; Zhao, 2013). Through these 56
studies, dysbiosis has become a central finding in many human diseases with tentative causes 57
linked to poorly balanced diets, unexpected chemicals/ingredients or other environmental 58
exposures. Consumption of different formulated foods and diets might lead to microbiome 59
changes in altered host health status (Botelho et al., 2018). For example, the dysbiosis 60
associated with high-fat or high-sugar diets (even from natural ingredients) is implicated in 61
obesity (Schulz et al., 2014; Turnbaugh et al., 2006). Biotransformation of foods by gut bacteria 62
is even reported to generate disease-causing metabolites, as suggested for L-carnitine 63
catabolism to trimethylamine-N-oxide (TMAO); this bacterial metabolite is derived primarily from 64
red meat and although it may promote bacterial osmolyte health benefits, is linked with 65
development of atherosclerosis in susceptible individuals by altering host cholesterol transport 66
(Koeth et al., 2013; Ussher, Lopaschuk, & Arduini, 2013). This finding is notable in the context 67
of food ingredients which are accumulating evidence of their capacity to adversely alter gut 68
microbiome composition and function in a manner that could impact host health.
69
In this review, we highlight recent links between specific food ingredients and gut dysbiosis, 70
and we provide technical recommendations for studying these effects. We also consider 71
potential strategies for reconstituting a healthy gut microbiome after dietary perturbation. We 72
conclude that consideration of the gut microbiome is important when evaluating food safety, as 73
well as nutritional intervention and management.
74
2. Perturbation of the human gut microbiome by imbalanced diet formulations and food 75
ingredients 76
Despite known lifestyle effects (including diet) on the assembly of distinct gut microbiome 77
communities in humans (Jha et al., 2018), it is now widely recognized that global food trade has 78
minimized such differences in populations consuming similar types of food. Modern food 79
manufacture among countries tends to be standardized, especially in its use of food additives to 80
prolong shelf life, enhance flavors and improve bio-functionality of manufactured foods (Carocho 81
et al., 2014). This has led to an emerging concern that frequent exposure to conserved food 82
additives or ingredients may disrupt gut microbiome homeostasis, thereby increasing host 83
disease susceptibility (Figure 1). Representative studies of food ingredient effects on gut 84
microbiome are summarized in Table 1.
85
2.1. Diet imbalance 86
Through natural symbiotic design to facilitate efficient food digestion, the GI tract offers 87
residence to trillions of commensal microorganisms that are shaped by our dietary influences.
88
Drastic shifts in gut microbiome composition are evident following long-term changes in diets 89
and are potentially associated with disease development. For example, in an adult French 90
cohort of 33,343 participants, an increase in the proportion of ultra-processed foods in the diet 91
was associated with a higher risk of developing functional gastrointestinal disorders, such as 92
irritable bowel syndrome and functional dyspepsia (Schnabel et al., 2018); but further studies 93
are needed to support causation by linking these longitudinal microbiome compositional shifts 94
with disease induction and restoring clinical health benefits through dietary intervention, ideally 95
in blinded placebo controlled cross-over trials. In further support of this notion, long-term 96
consumption of Western diet (high consumption of red meat and processed foods, etc.), but not 97
Mediterranean diet (high proportion of vegetables and whole grains, etc.), is widely recognized 98
as a risk factor in chronic inflammatory disorders such as Crohn’s disease and ulcerative colitis;
99
plausible mechanisms are linked to gut dysbiosis and enhanced intestinal permeability 100
triggering pro-inflammatory development (Khalili et al., 2018).
101
Seasonal recovery of the gut microbiome in Hadza hunter-gatherers of Tanzania is also 102
supportive of dietary impact on human microbiome patterns, being distinctly different from 103
urbanized individuals representing 18 different populations in 16 countries (Smits et al., 2017).
104
These findings are strongly supportive of dietary alterations that can significantly induce 105
microbial adaptation linked to human health. This phenomenon is exemplified by Hmong and 106
Karen populations from Thailand after immigration to the US, and involves microbiome 107
enterotype switching from Prevotella to Bacteroides after Westernization. Notably, this 108
enterotype switch is functionally linked with decreases in microbial enzymes required for plant 109
fiber degradation and loss of microbiome diversity, which are features associated with 110
development of obesity across generations (Vangay et al., 2018); Knights and colleagues 111
reasoned this microbiome shift might be linked to dietary changes. In addition to these 112
population-scale longitudinal studies, short-term (within days) intake of entirely animal-based 113
foods enriches bile-tolerant Alistipes and Bacteroides while decreasing the abundance of 114
Eubacterium and Ruminococcus that are capable of catabolizing plant polysaccharides (David 115
et al., 2014). These long- and short-term investigations on human subjects demonstrate that 116
dietary imbalance promotes gut dysbiosis and is associated with higher risk or susceptibility to 117
disease development. Thus, different formulations of food ingredients in the diet can 118
dramatically influence the composition and metabolic potential of the gut microbiome.
119
2.2. Non-caloric artificial sweeteners 120
Dietary non-caloric/nutritive artificial sweeteners (NAS) such as saccharin and sucralose 121
have been Generally Recognized As Safe (GRAS) by the US Food and Drug Administration 122
(FDA), and are widely used for manufacturing diverse processed/packaged foods which are 123
consumed by millions of type II diabetics and obese individuals without increasing caloric intake 124
(Suez, Korem, Zilberman-Schapira, Segal, & Elinav, 2015). However, to date there has been no 125
detailed investigation on the association between NAS consumption and gut microbiome 126
composition. In 2014, it was reported that glucose intolerance developed in mice fed 127
commercial saccharin, sucralose and aspartame when compared with mice fed isocaloric 128
sucrose and glucose; this finding models long-term NAS consumption being associated with 129
impaired glucose tolerance in humans. High resolution shotgun metagenomic analysis of human 130
fecal microbiome transplanted (FMT) germ-free mice identified NAS-enriched gut commensals 131
such as Bacteroides vulgatus as potential causative targets among the altered gut microbiome 132
(Bokulich & Blaser, 2014; Suez et al., 2014). Interestingly, administration of ciprofloxacin, 133
metronidazole and vancomycin to NAS-fed mice suppressed the glycemic response to NAS 134
suggesting elimination of causative bacteria by these antimicrobial agents (Bokulich & Blaser, 135
2014). Similar findings are evident in humans where responders showing elevated glycemic 136
response to NAS consumption demonstrate gut microbiome changes, whereas a proportion of 137
non-responder subjects maintaining a normal glycemic response to short-term NAS 138
consumption demonstrated insignificant changes in gut microbiota community (Suez et al., 139
2014). Since this dysbiosis may represent a compensatory microbiome response to elevated 140
systemic glucose levels rather than being causative, it would be interesting to explore whether 141
transfaunation of gut microbiome communities from responders versus non-responders into 142
germ-free models supports the transmissive glycemic phenotype. Nevertheless, these studies 143
highlight the dynamic association of gut microbiome communities with personalized glycemic 144
phenotypes (Zmora, Zeevi, Korem, Segal, & Elinav, 2016). Further support of nutraceutical 145
impact on the gut microbiome is provided by a recent study highlighting that sucralose 146
consumption induces liver inflammation in C57BL/6 mice and is associated with fecal 147
proinflammatory metabolite shifts (Bian et al., 2017). When considering the extensive 148
consumption of NAS, especially by individuals with metabolic syndrome, it becomes clear that 149
this class of foods needs to be re-assessed for impact on the microbiome and long-term health 150
consequences.
151
2.3. Nutritive sweeteners 152
Trehalose is a non-reducing and temperature-stable disaccharide making it a preferred 153
baking ingredient. Its use as a stabilizer is now also extended to thousands of other common 154
foods, including ground beef and ice cream. Although the microbiome is reported to utilize 155
trehalose, the metabolic advantage that this disaccharide confers to epidemic strains of the 156
pathogen Clostridioides difficile has drawn a lot of clinical and press attention (Collins et al., 157
2018). Two epidemic ribotypes (RT027 and RT078) of C. difficile are able to utilize levels of 158
trehalose which are expected from the consumption of trehalose-rich food products, promoting a 159
selective advance for pathogen colonization and virulence in the intestine. This finding is 160
interesting since the emergence of these epidemic strains co-occurred around the time that 161
trehalose was introduced in bulk as a food supplement around the world. The pathogenic fitness 162
displayed by epidemic C. difficile strains was linked to two independently acquired mutations 163
that confer an enhanced ability to utilize trehalose in the diet. Specifically, in RT027 strains a 164
conserved L1721 amino acid substitution in TreR, a repressor of the treAR operon and an 165
effector of the phosphotransferase system for trehalose translocation was identified. As a result, 166
the elevated induction of the treA gene increased pathogen sensitivity to dietary trehalose. In 167
RT078 strains, a horizontally acquired gene cluster consisting of a second copy of TreA and 168
TreR, and a trehalose-specific phosphotransferase system IIBC component confer enhanced 169
trehalose catabolism for energy generation needed for bacterial expansion (Collins et al., 2018).
170
When considering the high trehalose content of modern Western diets, these findings raise a 171
concern that epidemic C. difficile (RT027 and RT078) infections will remain prevalent until food 172
manufacturing practices are changed (Britton & Young, 2014; Collins et al., 2018). Recently, 173
degradation-resistant trehalose analogues (trehalase inhibitors) have been synthesized that 174
block trehalose utilization by hypervirulent C. difficile strains (Danielson et al., 2019), but use of 175
analogues as co-additives needs further evaluation in terms of safety. Importantly, these 176
findings set precedent that other GRAS-granted dietary sweeteners might confer a fitness 177
advantage to some certain infectious pathogens.
178
More recently, lactose, the natural sugar found in raw milk and processed dairy products, 179
has been reported to be associated with the expansion of Enterococcus faecium, but not other 180
enterococci in the large intestine of patients who underwent allogeneic hematopoietic stem cell 181
transplantation (allo-HSCT); such enterococcal expansion exacerbated graft-versus-host 182
disease (GVHD) and patient mortality in both human and mouse models (Stein-Thoeringer et al., 183
2019). It is well known that various antibiotics including vancomycin have been used to 184
eliminate opportunistic pathogens during the conditioning process prior to stem cell infusion and 185
even during cell engraftment; this treatment selectively enriches vancomycin-resistant 186
enterococci including Enterococcus faecium in the GI (Taur et al., 2018). Not surprisingly, allo- 187
HSCT patients with lactase deficiency showed significantly higher abundance of Enterococcus 188
compared with lactose absorbers (Stein-Thoeringer et al., 2019). To overcome such potentially 189
detrimental microbiome responses associated with lactose consumption, dietary manipulation 190
resulting in lactose depletion decreased enterococcal abundance and attenuated GVHD-related 191
mortality (Stein-Thoeringer et al., 2019). Lactose is a common dietary component in Western 192
diets, and consumption of lactose-rich diet favors dysbiosis with outgrowth of Enterococcus 193
faecium. Instead of using the Leloir metabolic pathway for exclusively processing glucose 194
moieties from lactose, this organism adopts a tagatose-6-phosphate (T6P) pathway for 195
catabolizing lactose thus utilizing both glucose and galactose moieties of lactose (Wu, Cheung, 196
& Shah, 2015). However, our early genetic survey indicated that not all strains of Enterococcus 197
faecium and Enterococcus faecalis possess intact T6P metabolic pathways, and the T6P 198
pathway in Enterococcus faecium could be either plasmid-encoded or present in its 199
chromosome (Wu et al., 2015). Thus, further investigation of the functional analysis of the T6P 200
pathway in clinically relevant strains of Enterococcus faecium isolated from allo-HSCT patients 201
are merited to profile such differences.
202
2.4. Dietary emulsifying agents 203
Dietary emulsifiers are detergent-like agents that are common ingredients in modern 204
processed foods. Although these agents are approved by the FDA, their influence on the gut 205
microbiome remains unclear. In 2015, carboxymethylcellulose and polysorbate-80 were shown 206
to alter gut microbiome composition in mice, associated with a significant reduction in the alpha- 207
diversity metric namely observed Operational Taxonomic Units (OTUs) that was linked with 208
blooming of Akkermansia muciniphila and inflammation-promoting Clostridium perfringens.
209
Transfaunation of the emulsifier-induced dysbiotic microbiota in germ-free mice induced low- 210
grade gut inflammation and promoted metabolic syndrome with increased adiposity and 211
dysglycaemia linking causation of these disease states with emulsifier consumption and impact 212
on the host microbiome (Chassaing et al., 2015). To investigate emulsifier effects directly on the 213
microbiome, a simulator of the human intestinal microbial ecosystem lacking host influence was 214
used: carboxymethylcellulose and polysorbate-80 not only increased alpha- and beta-diversity 215
(UniFrac distance) metrics in human fecal microbiome communities, but also induced microbial 216
gene expression resulting in higher levels of pro-inflammatory lipopolysaccharide and flagellin 217
FliC (Chassaing, Van De Wiele, De Bodt, Marzorati, & Gewirtz, 2017). Microbiome perturbations 218
caused by carboxymethylcellulose and polysorbate-80 are also reported to drive alterations in 219
proliferation and apoptosis pathways promoting colon carcinogenesis (Viennois, Merlin, Gewirtz, 220
& Chassaing, 2017). Similarly, carrageenan – a sulfated polysaccharide – is reported to induce 221
intestinal inflammation associated with a decreased relative abundance of Akkermansia 222
muciniphila in mice (Shang et al., 2017). However, further causation studies are required to 223
demonstrate the importance of Akkermansia muciniphila as an anti-inflammatory commensal in 224
response to dietary carrageenan. Since other dietary emulsifiers, for example glycerol 225
monolaurate, are also reported to induce gut dysbiosis, metabolic syndrome and low-grade 226
inflammation, even in a low-fat diet mouse model (Jiang et al., 2018), further evaluation of 227
detrimental emulsifier-host-microbiome interactions are merited.
228
2.5. Salt 229
Dietary salt (sodium chloride) is a ubiquitous component in modern diets. High salt 230
consumption is one of the causative factors contributing to cognitive impairment (Faraco et al., 231
2018, 2019) and cardiovascular diseases driven by hypertension (Mozaffarian et al., 2014).
232
However, the impact of this common food additive on the gut microbiome has not been well 233
characterized. Recently, a high salt diet was shown to alter mouse fecal microbiome community 234
composition with decreased relative abundances of Lactobacillus, especially Lb. murinus which 235
recovered after introduction of normal salt chow (Wilck et al., 2017); interestingly, administration 236
of Lb. murinus to high salt-challenged mice reduced the levels of hypertension-contributing TH17 237
lymphocytes (Wilck et al., 2017). In an open-label clinical pilot study, 14 days of increased salt 238
intake depleted several species of Lactobacillus including Lb. casei group, Lb. plantarum and Lb.
239
brevis as assessed via shotgun metagenomic analysis; high dietary salt also lead to elevated 240
nocturnal systolic and diastolic blood pressure, and reduced circulating TH17 lymphocytes 241
(Wilck et al., 2017). Although mouse and human differences exist in the specific Lactobacillus 242
sp. that are adversely impacted by a high salt diet, it does link Lactobacillus as a genus with 243
blood pressure abnormalities. This association could be due to altered host physiology caused 244
by high salt, for example intestinal transit is commonly linked with blood pressure and 245
Lactobacillus abundance. However, causation is indicated by a recent meta-analysis of nine 246
human trials showing that consumption of probiotic Lactobacillus reduces human systolic and 247
diastolic blood pressure (Khalesi, Sun, Buys, & Jayasinghe, 2014). Potentially, altered 248
microbiome function could mediate these changes in host physiology through modulation of 249
vagal nerve and immune signaling. For example, a high-salt diet administrated to mice 250
decreased the production of protective microbiome-derived short-chain fatty acids e.g. butyrate 251
and altered gut immune homeostasis involving enhanced pro-inflammatory gene expression 252
which exacerbated experimental colitis in mice. Transfaunation of this high salt diet-associated 253
microbiome into germfree mice demonstrated that changes in community composition are 254
dependent on continuous dietary salt exposure and that changes may be transient (Miranda et 255
al., 2018).
256
2.6. Dietary soluble fiber and carbohydrates 257
Soluble and insoluble dietary fibers are generally thought to be beneficial to human health 258
via sustaining a healthy gut microbiome and gastrointestinal homeostasis (Makki, Deehan, 259
Walter, & Bäckhed, 2018). Western diet which is typically a low-fiber diet shown to deplete gut 260
commensal bacteria that are capable of producing key physiological metabolites such as short- 261
chain fatty acids (SCFAs) (Deehan & Walter, 2016; Makki et al., 2018). In other words, dietary 262
fiber is vital for SCFAs production by the healthy colonic microbiome. Consumption of fiber-rich 263
diet also demonstrates other potentially notable health-promoting effects including improved 264
glucose tolerance through regulating the ratio of Bacteroides/Prevotella (Kovatcheva-Datchary 265
et al., 2015), facilitating clearance of C. difficile via possible link to elevated levels of SCFAs 266
(Hryckowian et al., 2018), and reducing risk of hepatocellular carcinoma (Yang et al., 2019).
267
There are a few exceptions to this consensus, for example Vijay-Kumar and the team 268
demonstrated that soluble fibers (inulin, pectin and fructo-oligosaccharides), but not insoluble 269
fiber counterpart cellulose, induced hepatocellular carcinoma in a subset of Toll-like receptor 5 270
(TLR5)-null mice, a model which is deficient in immune responses to bacterial flagellin (Singh et 271
al., 2018). They found that diet enriched in soluble fibers promoted γ-Proteobacteria relative 272
abundance as a potential pathogenic source of bacterial flagellin and lipopolysaccharides. The 273
soluble fiber diet also enriched for fermenters of the Clostridium XIVa cluster that was 274
associated with high levels of fecal butyrate and serum secondary bile acids linked to liver 275
fibrosis disease progression (Singh et al., 2018). Hepatocellular carcinoma was not observed in 276
germ-free, wild-type, and metronidazole-treated TLR5 deficient mice fed soluble fiber, which the 277
authors claim support a role for microbial-derived mediators in the disease pathogenesis (Singh 278
et al., 2018). It is also possible that the microbiome community shift represents a compensatory 279
response to liver injury and further work is needed to delineate disease causation. The impact of 280
food on secondary bile acids is an area of intense research currently, especially since colonic 281
7α-dehydroxylating bacteria which convert primary to secondary bile acids are also emerging as 282
potentially important modulators of human disease, including infectious colitis (Buffie et al., 2015;
283
Savidge & Sorg, 2019).
284
In addition to the unexpected side effects of soluble fiber (long-chain polysaccharides) 285
mentioned above, soluble short-chain carbohydrates are reported to exacerbate diseases such 286
as irritable bowel disease. Diets containing a high proportion of FODMAPs (Fermentable 287
Oligosaccharides, Disaccharides, Monosaccharides, And Polyols), such as in a normal Western 288
diet, can trigger gastrointestinal symptoms in patients with irritable bowel syndrome and have 289
been linked with altered microbiome fermentation (McIntosh et al., 2017; Shepherd, Lomer, &
290
Gibson, 2013). Symptom relief and improved quality of life can be achieved in a proportion of 291
irritable bowel syndrome patients on a diet that is low in FODMAPs (Camilleri & Acosta, 2014;
292
Shepherd et al., 2013). Much public attention has been levied on this dietary management as 293
this disease impacts up to 20% of the western population (Oshima & Miwa, 2015).
294
Carbohydrate-trigged diseases are likely highly dependent on microbiome activity, but 295
identifying the culprit species and disease mechanisms have been difficult. Another important 296
consideration here that applies broadly to the analysis of fecal specimens from human subjects 297
is that these may represent poor surrogates to investigate a disease process that happens in 298
the small intestine or at gut mucosal surfaces.
299
As shown in Table 1, food ingredients may demonstrate specific microbiome alterations 300
which are not in common across studies. One explanation for such diverse changes could be 301
linked to the metabolic traits of gut bacteria on those food ingredients, i.e. competitive growth 302
fitness conferred by specific ingredients such as simple sugars; however, for food ingredient- 303
induced depletion of beneficial commensal bacteria, indirect influences driven by alterations in 304
host and dysbiosis-associated chemical niches should not be ignored. Importantly, many 305
experimental validation studies transferred ingredient-induced microbiome communities into 306
germ-free or conventional mouse model and recapitulate the phenotype confirming the 307
microbiome dysbiosis-driven disease outcome. Although this highlights the potential 308
cooperation among microbes for phenotype induction, few of these studies have confirmed 309
disease causality of individual microbes associated with disease phenotypes. This merits further 310
effort to decipher microbe-microbe interactions to establish disease causation.
311
Long-term frequent consumption of food ingredients-containing foods and diets could be 312
cause diverse gut microbiome dysbiosis resulting in disease induction and susceptibility.
313
Modern food manufacturing should rethink the content of food additives and re-innovate such 314
process would improve food quality. Considering ingredient-specific induction of microbiome 315
alterations, strategies to reverse such dysbiosis might vary but the replacement with healthy gut 316
microbiome should be a universal and effective approach to recover healthy state. However, 317
microbiota replacement seems to be not a broadly acceptable way at current stage because our 318
knowledge on gut microbiome is still expanding and more extensive investigations are still 319
needed to illustrate microbial dark matter of gut microbiome. Thus, the best preventative 320
strategy so far is to follow certified dieting guidelines to avoid excessive consumption of food 321
additives. More precisely, we need to know what we are feeding, and thus microbiome 322
monitoring should aid such decision on diet choice and disease treatment.
323
3. Association to causation – technical recommendations for evaluating the effects of 324
food ingredients on host health with a focus on the gut microbiome 325
Numerous next generation sequencing techniques are established to profile the human 326
gut microbiome, especially targeted and shotgun metagenomic sequencing which are culture- 327
independent strategies. An important feature of these methods is that they are able to detect or 328
assemble a large proportion of uncultured or unknown microbes in the host GI tract beyond our 329
current culture collections and databases. Thus, the term “culturomics”, a conventional culture- 330
dependent method using varying nutrients for isolation and culture, has been introduced to 331
advance this field (Browne et al., 2016; Lagier et al., 2018, 2016). This labor-intensive approach 332
identified over 200 previously “unculturable” species from the human GI tract (Lagier et al., 333
2016), and generated an archive of gut bacteria representing 90% of GI bacterial species 334
commonly identified using next generation sequencing (Browne et al., 2016). These single 335
bacterial isolates can even be assembled together to perform functional dynamic assays of key 336
human microbiome consortia identified (Samuel, Rowedder, Braendle, Félix, & Ruvkun, 2016).
337
Although these culture-dependent efforts are meaningful, microbe-microbe and host-microbe 338
interactions still need to be considered when performing food/nutrition-associated microbiome 339
studies, such as toxicology of food additives and evaluation of nutritional interventions. To best 340
achieve this type of approach, we present a basic experimental workflow incorporating three 341
typical models for microbiome research as illustrated in Figure 2.
342
3.1. Consider the resilience and individuality of the human gut microbiome 343
The temporal stability of the human gut microbiome as measured by its taxonomic 344
composition over time has been unclear for a long time. Recent longitudinal studies have shed 345
some important new light on this question: in a Chinese longitudinal pilot study, bi-directional 346
plasticity and resilience of the human gut microbiome were observed across a longer time scale 347
(6 months) and was associated with multiple dietary shifts (Liu et al., 2019). Similar 348
observations were re-captured in another longitudinal (more than one year) and geographically 349
diverse study on Hadza hunter-gatherers in Tanzania showing annual cyclical reconfiguration of 350
microbiome community compositions that are linked to seasonal availability of different types of 351
food (Smits et al., 2017). Geography also shows a strong association with host microbiome 352
variation as demonstrated in a Chinese regional trial composing over 7,000 individuals from 14 353
districts in Guangdong province (He et al., 2018). Notably, ethnic origins contribute to the inter- 354
individual dissimilarities in gut microbiome composition in Amsterdam, Netherlands revealed in 355
a trial of over 2,000 participants (Deschasaux et al., 2018); these reports demonstrate the 356
contribution of geography and ethnicity (i.e. genetics) in explaining microbiome variation (Gupta, 357
Paul, & Dutta, 2017), but variation is also influenced by diet and lifestyle indicating the 358
importance of understanding the personalized microbiome in terms of individuality. Large 359
pediatric and adult cohort studies of healthy populations, including MetaHIT, Human Microbiome 360
Project, American Gut Project and TwinsUK, endorse these findings by demonstrating 361
population-scale stratification based on distinct microbiome community clusters termed as 362
“enterotypes” (Arumugam et al., 2011; Costea et al., 2017). In adults, this variation is 363
contributed to by genetics and epigenetics, although the basis for enterotyping an individual 364
based on their gut microbiome composition remains controversial. Further, enterotypes have not 365
been reported in infants, and the TEDDY cohort, a large longitudinal study composing over 900 366
infants with follow-up from 3 to 46 months of age, showed that the early assembly of the infant 367
microbiome underwent distinct developmental, transitional and stable phases (Stewart et al., 368
2018). Infant microbiome composition was largely associated with breastfeeding, while birth 369
mode, geography and household exposure also contributed to microbial variation (Stewart et al., 370
2018). These population scale studies have thought us that the individuality of the human gut 371
microbiome is associated with age, diet, and environmental exposures. These are important 372
lessons since we need to consider host genetics and epigenetics when assessing the impact of 373
food ingredients on the human gut microbiome, as well as microbiome-based interventions for 374
personalized nutrition and disease management (Christensen et al., 2018; He et al., 2018).
375
In the context of personalized nutritional management, a pioneering study measured host 376
glycemic responses to dietary carbohydrates such as bread (Zeevi et al., 2015). Elevated post- 377
meal blood glucose levels are normally a major risk factor for developing type II diabetes, yet 378
postprandial glycemic responses among individuals consuming the same identical meals are 379
highly variable (Zeevi et al., 2015). One major factor to be considered for such variability is the 380
individuality of the gut microbiome (Korem et al., 2017), which may be considered a significant 381
metabolic organ in personalized nutrition (Bashiardes, Godneva, Elinav, & Segal, 2018). Such 382
an approach potentially has significant implications for general published guidelines on diet and 383
lifestyle aimed at educating populations on healthy food choices. Because an individual’s gut 384
microbiome may require deviations from general dietary guidelines to accommodate an 385
optimized lifestyle based on diet (David et al., 2014; Deschasaux et al., 2018; Jha et al., 2018;
386
Petersen et al., 2017), geography (Gupta et al., 2017; He et al., 2018; Yatsunenko et al., 2012), 387
age (Yatsunenko et al., 2012) and environmental conditions (Worthmann et al., 2017), this 388
provides new challenges to nutritionists providing dietary recommendations. This becomes 389
especially taxing with patients seeking clinical advice on microbiome profiling provided by 390
emerging commercial vendors, who often refer the client back to their own physician for correct 391
interpretation. A pragmatic approach is needed, with clinical associations with certain microbial 392
taxa being recognized as such until causation can be demonstrated using the approaches 393
outlined in Figure 2.
394
3.2. In vitro bioreactor studies 395
In vitro models have been extensively used to monitor the taxonomic and functional 396
changes in microbiome communities perturbed by food or specific nutrients. Different types of 397
bioreactors have been developed to maintain the complex structure and diversity of a 398
microbiome community. A well-designed model is the Simulator of the Human Intestinal 399
Microbial Ecosystem (SHIME), a 5-step multi-chamber bioreactor, which was developed two 400
decades ago for mimicking the progressive digestive processes of the stomach, small intestine 401
and colon (Molly, Vande Woestyne, & Verstraete, 1993). The spatiotemporal model enables a 402
complex and stable microbiome community structure to form, and a specific model termed M- 403
SHIME also incorporated an unstirred mucus layer for adhering microbiome biofilm (Van den 404
Abbeele et al., 2012). A more simplistic model – Mini-BioReactor Arrays (MBRAs), was 405
developed for rapid cultivation of microbiome communities from human fecal specimens 406
(Auchtung, Robinson, & Britton, 2015). This MBRAs model enables medium throughput studies, 407
which have been applied to investigate for example trehalose effects on C. difficile growth in 408
complex microbiota communities (Collins et al., 2018). A similar model strategy was used to 409
establish symbiotic (prebiotic and probiotic combination) mechanisms that drive antimicrobial 410
activity against C. difficile (Steiner, 2010). However, significant limitations using these in vitro 411
models include lack of host influences e.g. antibodies or antimicrobial, or limited experimental 412
throughput especially in SHIME models which provide a more stable microbiome community 413
structure (Mcdermott & Huffnagle, 2014; Pabst, Cerovic, & Hornef, 2016; Zeng et al., 2016).
414
Although this in vitro approach provides a clearer understanding of the perturbations that 415
directly target gut microbiome structure-function relationships, host-microbiome interactions 416
should not be neglected since these may exert profound effects on host physiology.
417
3.3. Ex vivo or in vivo studies 418
Ex vivo approaches are not extensively reported for gut microbiome studies. When this 419
strategy is adopted, it usually involves time-series fecal cultivation in combination with dietary 420
perturbations (Collins et al., 2018; Lamendella et al., 2018). Results from this type of study 421
design should be interpreted with caution since the stability of the microbiome community 422
structure during experimental anaerobic cultivation is uncertain and not well controlled.
423
An advantage of using animal models (typically mouse) for studying the gut microbiome is 424
that perturbation can be controlled using standardized feed and housing conditions. A 425
translational limitation that needs to be considered is the major species difference between 426
murine and human microbiome community composition and function. To overcome this, 427
significant effort has been levied to generate humanized models by introducing an individual’s 428
gut microbiome communities into the germ-free mice, a process termed transfaunation (Collins, 429
Auchtung, Schaefer, Eaton, & Britton, 2015; Faith et al., 2010). However, this model often does 430
not contain the same diversity and structure of human gut microbiome and emerging data 431
indicates that certain host responses e.g. immune responses are not appropriately developed in 432
the presence of human-derived microbial species as these have not evolutionary adapted to the 433
mouse gut environment or genetic determinants (Zhou, Chow, Fleming, & Oh, 2019).
434
3.4. Human cohort studies 435
Generally, clinical trials should be carried out when expected outcomes are validated in 436
preclinical models, but with major microbiome compositional differences existing across species 437
that possibly complicate experimental findings, exceptions might be applied for GRAS-granted 438
or Qualified Presumption of Safety (QPS)-listed food ingredients and probiotics. The 439
observation of region-dependent microbiome differences among age-matched human 440
individuals has been linked to diets contributing to rapid transit of the gut microbiome (David et 441
al., 2014; Deschasaux et al., 2018; He et al., 2018; Yatsunenko et al., 2012). Hence, any pilot 442
study involving human subjects for microbiome-based assessment needs to control for or 443
document dietary differences. A diet provided by the clinical coordinator with a detailed food and 444
symptoms diary is the preferred documentation for such clinical studies since recall memory is 445
often reported to provide inaccurate information. Although representing the gold standard, new 446
electronic applications are assisting with dietary assessment and confer some advantages since 447
time stamped entries of snacks and meals may provide a more accurate approach to food 448
questionnaires and diaries as these provide gross-level estimates. Smartphone applications, for 449
example MyFitnessPal provides automated energy and macro- and micronutrient intake values 450
based on foods consumed (Teixeira, Voci, Mendes-Netto, & da Silva, 2018). Longitudinal 451
measurements of dietary and gut microbiome data, ideally as a double-blinded cross-over or 452
parallel design trials, may assist in powering the study and to control for individual specific 453
responses. Data analysis of responder versus non-responder phenotypes should be performed 454
separately and blindly to avoid possible interpretation bias.
455
3.5. Analytical methods for profiling microbiome composition and inferring metabolic 456
function 457
Several strategies as shown in Figure 3 are used for profiling the gut microbiome and 458
include: (1) targeted metagenomic sequencing, i.e., 16S rDNA amplicon or full-length 459
sequencing; (2) shotgun (deep or shallow) metagenomic sequencing; (3) shotgun 460
metatranscriptomic sequencing; (4) mass spectrometry-based metaproteomic analysis. Among 461
these high-throughput assays, targeted metagenomics – commonly referred to as 16S rDNA- 462
based amplicon sequencing is a more affordable approach for gut microbiome analysis.
463
Amplicon sequence data has been generated for thousands of projects, but each project is 464
designed with little consideration to preference for selecting variable 16S rDNA region that are 465
commonly used i.e. V4, V1-V3, V3-V5 or V6-V9. Sequencing strategies are often influenced by 466
large studies such as Human Microbiome, Earth Microbiome and American Gut Projects since 467
validations have been conducted before the production phase of these studies. However, the 468
influence of amplicon length (after quality trimming), sequence orientation and 16S region on 469
taxonomic calls has not been benchmarked systematically. To ensure the same observations on 470
gut microbiome features are validated by previously published data, meta-analysis of gut 471
microbiome-associated studies might be performed via a closed-reference analysis strategy 472
allowing comparisons and concatenation of data (Wang et al., 2018), however, attention need to 473
be exercised when using such an approach: 1) quality filtering and trimming of amplicon 474
sequences should adopt expected error-based methods but not average quality-based filtering 475
(Edgar & Flyvbjerg, 2015); 2) for the sake of accuracy default similarity (97%) for clustering 476
amplicon sequences might be avoided since there is no strong evidence to support the even 477
distribution of 3% dissimilarity in nucleotides confined to the nine variable regions of the 16S 478
rRNA gene that are sequenced (Mysara et al., 2017; Nguyen, Warnow, Pop, & White, 2016;
479
Yang, Wang, & Qian, 2016), thus more stringent clustering or sequence denoising for meta- 480
analysis may generate more accurate profiles; 3) methods of taxonomic classification for the 481
feature sequences or representative sequences of OTUs should be benchmarked in detail for 482
selecting the confidence threshold for controlling misclassification and overclassification (Edgar, 483
2018; Murali, Bhargava, & Wright, 2018); 4) appropriate choice of reference database is 484
important for accurate taxonomy, for example Greengenes and SILVA databases do not have 485
precisely curated databases because those collections are usually derived from GenBank and 486
European Nucleotide Archives where world-wide users can deposit sequence data (Edgar, 487
2018). An authoritative database from the training set of Ribosomal Database Project (RDP), 488
NCBI 16S rRNA RefSeq Targeted Loci Project or Genome Taxonomy Database (GTDB) is 489
commonly recommended but users should be cautious in relying exclusively on these 490
databases. Another strategy that is emerging is applying exact amplicon sequence variants 491
(ASVs) introduced by DADA2 for the meta-analysis (Callahan et al., 2016). Moreover, multiple 492
copies (ranging from 5 to 10) of prokaryotic ribosomal RNA operons in a single genome are 493
commonly found (Stoddard, Smith, Hein, Roller, & Schmidt, 2015); ideally, individual 494
microbiome OTUs and ASVs profiles should be corrected for 16S copy number so that this is 495
matched to the ribosomal RNA database. A major challenge in conducting this type of 496
bioinformatics correction is that short amplicon sequences of 16S rRNA gene do not offer 497
confident calls at the rank of species because those amplicons may be derived from either 498
known species or uncultured bacteria sharing the same sequence identity of the sequenced 499
region of 16S rRNA gene. Even with these considerations, microbiome profiling bias might still 500
be there and this merits future bioinformatics and statistical efforts (McLaren, Willis, & Callahan, 501
2019). Importantly, microbiome community profiles built on calculating relative abundances do 502
not reflect the biomass of the microbial consortia, thus total 16S copy numbers should be 503
assessed, i.e. real-time quantitative PCR with bacterial universal primers for the same DNA 504
extracts.
505
Since 16S rDNA amplicon sequencing does not offer high resolution to taxon, especially at 506
species rank, shotgun metagenomic sequencing becomes a useful strategy to provide potential 507
whole-genome information for precise taxonomic profiling of the gut microbiome. In general, two 508
types of analysis have been performed as illustrated in Figure 3B: 1) read-mapping based 509
analysis: filtered and trimmed sequence reads are mapped to the curated reference microbial 510
genomes or a set of marker genes per bacterial genomes by using sequence alignment or k- 511
mers spectrum, this approach is adopted in many popular analysis packages – MetaPhlAn, 512
DIAMOND, Kraken and CLARK (McIntyre et al., 2017); 2) de novo assembly-based analysis:
513
individual or concatenated sequencing reads from all samples are first assembled prior to 514
taxonomic assignment and functional annotations. This type of analysis includes many 515
packages but the comprehensive Anvi’o platform is useful for downstream analysis (Eren et al., 516
2015). For de novo assembly, curations must be performed for misassemblies and misbinning 517
of metagenomic contigs; sometimes, authoritative genome databases are required to guide 518
those processes depending on the programs. Timely documentation from large and 519
comprehensive benchmarking studies on analytical strategies, performance of programs, and 520
databases have been generated (McIntyre et al., 2017; Sczyrba et al., 2017), and these 521
comparisons provide valuable information to users for selecting analysis strategies of shotgun 522
metagenomics.
523
Shotgun metagenomics is able to provide not only precise taxonomy but also the potential 524
functional dynamics of the gut microbiome. Although meta-transcriptomics is a preferable 525
approach providing better understanding on functional potential (Abu-Ali et al., 2018), unbiased 526
label-free metaproteome analysis of fecal specimens or biopsies also provides potential at 527
protein level to probe microbiome functional capacity (Zhang et al., 2016). A potential concern 528
using this latter technique is protein detection may be biased to abundant proteins as these are 529
more easily profiled; detection limits are associated with: 1) invalid methods for protein 530
extraction with high yields; 2) simple or lack of fractionation of tryptic peptides; 3) constraints 531
from database-driven peptide identification, while de novo peptide sequencing for tandem mass 532
spectra is still prone to errors (Muth & Renard, 2018). As such, paired metagenomics- 533
assembled genomes together with public protein databases including UniProt and Unified 534
Human Gastrointestinal Protein (UHGP) (Almeida et al., 2019) are commonly recommended for 535
the database search strategy during metaproteome analysis (Heyer et al., 2017).
536
4. Potential strategies for overcoming gut dysbiosis caused by specific food ingredients 537
Globalization of the food industry contributes to the wide spread consumption of common 538
food ingredients around the world. It is important for government agencies to rapidly update 539
public guidelines once GRAS status of specific food ingredients is revoked after re-evaluation of 540
possible impact on human health. To apply such guidelines to dietary effects on the human 541
microbiome, we must first understand how to define a “normal microbiome”, especially since 542
dysbiosis is a frequent manifestation without an associated health outcome. Recently, there is 543
increasing demand for manipulations and interventions that may counter and/or correct 544
dysbiosis. Often, little consideration is given to possible compensatory responses of the 545
microbiome being beneficial to the host, and thus the term dysbiosis may be misleading as it 546
implies a detrimental outcome to the host. For this reason, we and others emphasize the 547
importance of establishing disease causation before considering microbiome manipulation as a 548
form of therapy. If causation can be established, two strategies including ecological modelling 549
and non-modelling approaches may be considered for discovering effector strains that could be 550
considered to restore a healthy microbiome (Figure 4).
551
4.1. Dietary intervention 552
A healthy diet is key to maintaining gut microbiome homeostasis by supplying appropriate 553
nutrients and cofactors to benefit microbial function. Patients with type 2 diabetes (T2D) have a 554
dysbiosis that is linked to a deficiency in short-chain fatty acids (SCFAs) production (Tolhurst et 555
al., 2012); a high-fiber diet, demonstrated in a randomized clinical trial to selectively enrich 15 556
species with acetate or/and butyrate production potential. This finding was associated with 557
improved glucose homeostasis via triggering of intestinal glucagon-like peptide-1 production 558
that that enhances pancreatic insulin secretion in T2D participants (Zhao et al., 2018). This 559
benefit was validated in a mouse model with dysregulation of glucose homeostasis by 560
administrating acetate-producing Bifidobacterium pseudocatenulatum (Zhao et al., 2018).
561
Moreover, a specialized diet containing acetate and butyrate provided protection against insulitis 562
in non-obese diabetic mice, a model that is prone to developing human T1D-like symptoms 563
(Mariño et al., 2017; Wen & Wong, 2017). This intervention has also been applied to lactose- 564
intolerant individuals administrated short-chain galacto-oligosaccharides that enriched lactose- 565
metabolizing commensal bacteria, which in turn improved clinical outcomes of host lactose 566
digestion and tolerance (Azcarate-Peril et al., 2017). Another conceptual study demonstrated in 567
mice showing that SCFAs generated from gut bacteria via fermenting microbiota-accessible 568
carbohydrates (MACs), i.e. dietary plant polysaccharides, reduced the fitness of C. difficile in the 569
gut (Hryckowian et al., 2018); such a dietary strategy might be appropriate in combination with 570
fecal microbiota transplant or the treatment and management of patients with complicated 571
recurrent CDI. The future holds promise to develop precise and personalized dietary 572
management using computational modelling, i.e. supervised machine learning, to identify 573
human gut microbiome functional features that are associated with clinical metadata; this type of 574
an approach has been demonstrated to aid in optimizing dietary interventions for patients with 575
Crohn’s disease, prediabetes and T2D-diabetes (Bauer & Thiele, 2018; Zeevi et al., 2015).
576
Loss of specific gut commensal bacteria is common in dysbiosis with the reduction of new 577
and often unexplored species; thus, recovery of a healthy gut microbiome is a time-consuming 578
process via dietary intervention and management. Such manipulations are important for us to 579
better understand the role of dysbiosis in clinical practice, but may not be a first-line strategy, 580
especially in patients requiring fast symptom relief from severe illness, such as cancer and 581
infectious diseases.
582
4.2. Fecal microbiome transplantation 583
Fecal microbiome transplantation (FMT) is originally designed as investigational therapy for 584
treating patients with gut dysbiosis-associated recurrent C. difficile infection (rCDI) (Britton &
585
Young, 2014). FMT is not an FDA-approved procedure and is being explored for treating 586
diseases with potentially serious health consequences, especially in the critically ill or 587
immunodeficient patient. Recent adverse events associated with FMT transfer of antibiotic- 588
resistant pathogens has prompted the FDA to restrict its use to clinical trials that have an 589
approved investigational new drug (IND) status (Kim & Gluck, 2019). Generally regarded as a 590
treatment after multiple recurrent disease episodes in CDI patients, a major reason for adopting 591
FMT is that conventional antibiotic treatment worsens or does not reverse the dysbiotic 592
microbiome (Kassam, Lee, Yuan, & Hunt, 2013). Unlike many other diseases, there is strong 593
evidence to support microbiota causation in rCDI and this justifies its use as investigational 594
microbiota therapy, especially since CDI can be rapidly fatal in these patients (Britton & Young, 595
2014; Buffie et al., 2015; Seekatz & Young, 2014).
596
FMT use has also recently been extended to the treatment of other gut dysbiosis-associated 597
diseases with mixed outcomes, including some efficacy in Crohn’s disease and ulcerative colitis 598
(Moayyedi et al., 2015; Paramsothy et al., 2017), improved cognition in cirrhosis with recurrent 599
hepatic encephalopathy (Bajaj et al., 2018, 2017), and decreased GVHD-related mortality in 600
recipients of allogeneic hematopoietic stem cell transplantation (Shono et al., 2016; Shono &
601
Van Den Brink, 2018; Taur et al., 2018). Use of immune checkpoint inhibitors for treating 602
patients with numerous cancer types may induce serious immune-associated adverse effects, 603
such as colitis in a subset of patients (Wang et al., 2018); but colitis may be abrogated by FMT 604
to restore regulatory T-cell function in cancer management (Wang et al., 2018). Several of these 605
clinical trials demonstrate rapid microbiome reconstitution following FMT and improved 606
symptoms. However, other studies demonstrate that FMT is contraindicated and should be used 607
with caution because the well reported safety concerns on the transfer of infectious agents (Kim 608
& Gluck, 2019; Zuo et al., 2018).
609
4.3. Targeted microbial therapeutics – next-generation probiotics 610
Targeted microbial therapeutics are being developed to remove or inhibit certain features of 611
the dysbiotic microbiome, i.e. antimicrobial resistance genes and infectious pathogens. As 612
illustrated in Figure 4, ecological modelling and non-modelling approaches might prove useful 613
in the discovery of effector strains during dietary interventions and FMT; those effector strains 614
would be potential probiotics for precisely reversing host disease states. Recently, such an 615
approach was used to identify Clostridium scindens as candidate probiotic to protect against C.
616
difficile infection via secondary bile acid-mediated resistance and tryptophan-derived antibiotics 617
(Buffie et al., 2015; Kang et al., 2019). Other examples include consumption of probiotic Bacillus 618
subtilis to abolish gut colonization of pathogenic Staphylococcus aureus via fengycins-trigged 619
inhibition of S. aureus quorum sensing (Piewngam et al., 2018); reduced intestinal inflammation 620
through tungstate-mediated microbiota editing that specifically inhibits activity of gut 621
Enterobacteriaceae enriched during colonic inflammation (Byndloss et al., 2017; Litvak, 622
Byndloss, & Bäumler, 2018; Zhu et al., 2018); oral antibiotic administration also sustained 623
intestinal T cell dysfunction in the host even after FMT-based microbiome reconstitution, but the 624
immune defect was spared after supplying exogenous butyrate to the host (Scott et al., 2018);
625
Bacteroides ovatus alleviated gut inflammation induced by lipopolysaccharide and dextran 626
sodium sulfate (Ihekweazu et al., 2019; Tan, Zhao, Zhang, Zhai, & Chen, 2019); Bacteroides 627
uniformis with expanded glycolytic capability exhibited pre-clinical efficacy on obesity-associated 628
metabolic and immune dysfunction (Benítez-Páez, Gómez del Pulgar, & Sanz, 2017);
629
Clostridium butyricum also demonstrated potential to attenuate acute colitis via regulating TLR2- 630
MyD88 anti-inflammatory responses (Kanai, Mikami, & Hayashi, 2015); supplementation of 631
ethanol-depleted Akkermansia muciniphila conferred protection against ethanol-associated gut 632
leakiness by increasing mucus thickness and elevating tight-junction protein expression 633
(Grander et al., 2017, 2018). Interestingly, a mixture of probiotics suppressed the growth of 634
hepatocellular carcinoma tumors by modulating the gut microbiome regulation of Th17 cell 635
differentiation and IL-17 cytokine bioavailability in the tumor microenvironment (Li et al., 2016).
636
Over the last decade, newly-identified commensal strains with probiotic potential have been 637
identified and include Bacteroides xylanisolvens, Bacteroides ovatus, Bacteroides dorei, 638
Bacteroides fragilis, Faecalibacterium prausnitzii, Eubacterium hallii, Eubacterium rectale, 639
Anaerostipes hadrus and Roseburia hominis, demonstrating efficacy against cancer, gut 640