Macronutrient influences on the gut microbiota

Carbohydrate is critical for the host and is the major energy source for the gut microbiota. The principal carbohydrate sources for individuals consuming a western-style diet are resistant starch (RS) with usually lower amounts of NSP, which are consumed mainly in cereals. In addition, oligosaccharides such as fructans are consumed in foods including onions and artichokes. The extent of dietary starch breakdown in the colon is dependent on a number of factors that include amylose content and state of gelatinisation which is mainly determined by the cooking process⁽ Reference Englyst, Kingman and Cummings ^{31 )}. Starch is a complex polysaccharide consisting of a mixture of amylose (1,4-α-linked glucose residues) and amylopectin, a branched polymer composed of amylose chains linked to an amylose backbone by 1,6-α-linkages. The relative proportion of amylose and amylopectin has an effect on the availability of different types of starch for bacterial growth. In a model system, 80% of sequences recovered from starch particles belonged to Ruminococcus bromii, E. rectale and Bifidobacterium species⁽ Reference Leitch, Walker and Duncan ^{32 )}.

Moreover, bacteria related to R. bromii increased almost two-fold in faecal samples when the volunteers switched from normal diets to starch enriched diets⁽ Reference Abell, Cooke and Bennett ^{22 )}. Faecal populations of the Roseburia/E. rectale group⁽ Reference Aminov, Walker and Duncan ^{33 )} have been shown to respond to controlled variations in dietary carbohydrate intake⁽ Reference Duncan, Belenguer and Holtrop ^{2 ,} Reference Russell, Gratz and Duncan ^{8 )} that may be explained by a dependence on RS for growth. Another recent study compared the effects of diets supplemented either with a type three RS or a source of NSP (wheat bran) in obese subjects. The groups showing the most significant responses to RS on average were Firmicutes bacteria related to R. bromii and E. rectale, with the former group increasing 10-fold on average as a percentage of total bacteria. Two of the fourteen individuals showed no detectable ruminococci in faecal samples, and carbohydrate analysis of these samples revealed <40% fermentation of RS compared with >95% in the other subjects⁽ Reference Walker, Ince and Duncan ^{1 )}.

High fibre diets increase faecal bulking, SCFA production and transit rates along the large intestine⁽ Reference Lewis and Heaton ^{34 ,} Reference Lampe, Slavin and Melcher ^{35 )}. Faecal butyrate concentrations have been shown to correlate with the abundance of the Roseburia/E. rectale group⁽ Reference Duncan, Belenguer and Holtrop ^{2 )} which is likely to be a major contributor to butyrate formation from starch. This group is stimulated at mildly acidic pH in vitro largely because of the inhibition of competing Bacteroides species⁽ Reference Duncan, Louis and Thomson ^{36 )} suggesting that its preferred niche may be in the proximal colon where rapid fermentation creates mildly acidic conditions. Active fermentation of other fibre sources and prebiotics also tends to decrease pH in the proximal large intestine⁽ Reference Bown, Gibson and Sladen ^{37 )} and these shifts in pH can impact the gut microbial community structure⁽ Reference Duncan, Louis and Thomson ^{36 ,} Reference Walker, Duncan and Leitch ^{38 ,} Reference Belenguer, Duncan and Holtrop ^{39 )}.

There is much interest in modulating microbial metabolism in the intestine through dietary additives such as prebiotics that are defined as ‘a selectively fermented ingredient that allows specific changes, both in the composition and/or activity in the GI microbiota that confers benefits upon host well being and health’⁽ Reference Gibson and Roberfroid ^{40 )}. Current prebiotics are mainly carbohydrates of low digestibility that are found naturally in foodstuffs. Candidate prebiotics include xylo-oligosaccharides and galacto-oligosaccharides⁽ Reference Macfarlane, Steed and Macfarlane ^{41 ,} Reference Macfarlane, Macfarlane and Cummings ^{42 )} with most studies focusing on the use of inulin and fructo-ologosaccharides. The high numbers of Bifidobacterium species in the faeces of breast-fed babies is thought to result from their ability to utilise oligosaccharides in breast milk, including galacto-oligosaccharides while formula fed infants tend to harbour a greater diversity of organisms including a higher abundance of enterococci⁽ Reference Harmsen, Wildeboer-Veloo and Raangs ^{43 )}. The metabolite profiles also differ with higher levels of NH₃, amines and phenols in bottle fed babies compared with breast-fed infants⁽ Reference Heavey, Savage and Parrett ^{44 )}. Not all oligosaccharides, however, are highly selective⁽ Reference Flint, Duncan, Scott and Louis ^{3 )} and fructans, for example, have been shown to promote one or more groups of bacteria in addition to their effects on bifidobacteria⁽ Reference Kruse, Kleessen and Blaut ^{45 –} Reference Apajalahti, Kettunen and Kettunen ^{47 )}. As there is extensive inter-individual variation in composition of the gut microbiota there is likely to be inter-individual variation in the response of the microbial communities to prebiotics⁽ Reference Ramirez-Farias, Slezak and Fuller ^{46 )}. Recent evidence supports the view that there are likely to be detrimental health consequences of reduced carbohydrate intake especially when combined with an increase in the consumption of protein and/or fat, both of which have been potentially associated with an increased risk of colon cancer⁽ Reference Russell, Gratz and Duncan ^{8 ,} Reference Bernstein, Bernstein and Payne ^{48 )}.

Many protein- and protein-derived metabolites are considered to be potentially mutagenic and genotoxic and these include heterocyclic amines and N-nitroso compounds⁽ Reference Russell, Gratz and Duncan ^{8 )}. Studies with germ-free animals implied that the gut microbiota was essential in the formation of N-nitroso compounds⁽ Reference Massey, Key and McWeeny ^{49 )} and the DNA damage by heterocyclic amines was decreased compared with that of conventional animals⁽ Reference Kassie, Rabot and Kundi ^{50 )}. Additional by-products of amino acid metabolism include polyamines (including putrescine, cadaverine, spermine, spermidine, pyrollidine and piperidine), indoles, NH₃, hydrogen sulphide, branched chain fatty acids and the aromatic amino acid metabolites. Various species have been shown to produce these products⁽ Reference Beerens and Romond ^{51 ,} Reference Smith and Macfarlane ^{52 )}, but comprehensive screening of the predominant gut species has not been performed.

Significant changes in the gut metabolome were observed for volunteers consuming diets in which the protein and carbohydrate ratios were modulated⁽ Reference Russell, Gratz and Duncan ^{8 )}. Using liquid chromatography–MS, derivatives of a wide range of plant phenolics considered to be cancer-protective were reduced in diets high in protein (137 g/d) and low in carbohydrate (22 g/d). Phenylacetic acid, a major metabolite related to protein metabolism was found to have significantly increased. Increasing the carbohydrate content (181 g/d) resulted in significantly increasing some phenolic acids and their derivatives, principally fibre-related phenolics, namely, ferulic acid, 4-hydroxy-3-methoxyphenylpropionic acid and 3-hydroxyphenylpropionic acid.

Bile metabolism, along with lipase activity, are likely to be major factors affecting the delivery of lipids to the colon. Bile acids are predominantly deconjugated and dehydroxylated to secondary bile acids and Bacteroides intestinalis, Bacteroides fragilis and Escherichia coli have been shown to be involved in the production of deoxycholic and lithocholic acid⁽ Reference Fukiya, Arata and Kawashima ^{53 )}. Two major lipid metabolism pathways have been identified; hydration and reduction. The species responsible for the formation of hydrated lipids (i.e. hydroxy fatty acids) were predominantly Clostridium perfringens ⁽ Reference Pearson, Drasar and Wiggins ^{54 )} and Roseburia species⁽ Reference Devillard, McIntosh and Duncan ^{55 )}, although some other species could produce these metabolites to a lesser extent⁽ Reference Devillard, McIntosh and Duncan ^{55 )}. The species catalysing fatty acid reduction are less well characterised; only species responsible for the conversion of linoleic acid to vaccenic acid have been identified, namely Butyrivibrio fibrisolvens and Roseburia species⁽ Reference McIntosh, Shingfield and Devillard ^{56 )}.

Micronutrient influences on the gut microbiota

It has been shown that the microbiota can synthesise some vitamin B (B₃, B₅ and B₆) and related molecules including biotin, tetrahydrofolate, vitamin K and the corrinoids⁽ Reference O'Keefe, Ou and Aufreiter ^{57 )}. Microbes are considered to affect absorption of certain dietary minerals, with many studies demonstrating that carbohydrate and associated SCFA modulate uptake of Na, Ca and K. Many bacteria also actively accumulate Fe as a siderophore complex⁽ Reference Neilands ^{58 )}. There is very little information regarding the direct effect of the gut microbiota on mineral uptake and the species associated have not been identified.

Non-nutrient/phytochemical influences on the gut microbiota

Epidemiological studies suggest that there is an inverse association between the intake of phytochemical-rich diets and the incidence of CVD, diabetes and cancer⁽ Reference Boeing, Bechthold and Bub ^{59 )}. Edible plant material contains hundreds of compounds most of which play an important role for the plant including protection against pathogens. Biochemically, they can be broadly categorised according to their structure and biosynthetic pathways, but it should be appreciated that many secondary metabolites are derived by combining elements from all of these biosynthetic routes. Some of these compounds particularly if available as small molecules or as their aglycones, may be absorbed in the upper GI tract and directly enter systemic circulation. However, many and in particular those bound to other plant components such as carbohydrates will be available in the colon. Within the colon these phytochemicals are extensively metabolised. Gut microbiota are capable of performing many transformations including: hydrolysis, deamination, dehydrogenation, demethylation, decarboxylation, ring cleavage and chain shortening.

SCFA

Active fermentation of carbohydrates in the colon results in the formation of SCFA⁽ Reference Barry, Hoebler and Macfarlane ^{60 )} together with gases, mainly H₂, CO₂ and methane. The SCFA detected in stool samples are a sub group of fatty acids with aliphatic tails with less than six carbons and include fornic, acetic, propionic, butyric and valeric acid. Branched-chain examples include isovaleric and isobutyric acid. There are also substituted short-chain carboxylic acids such as lactic acid. The total concentration of SCFA in the large intestine may reach upwards of 100 mm ⁽ Reference Macfarlane and Macfarlane ^{61 )}. Dietary shifts can result in changes in SCFA production rates and in the molar proportions of different SCFA detected in faeces. Weight loss diets that are high in protein but low in carbohydrates, for example, were recently shown to reduce faecal butyrate up to four-fold⁽ Reference Duncan, Belenguer and Holtrop ^{2 )} while higher proportions of propionate and butyrate and lower acetate have been reported to result from increasing prebiotic or fibre intake⁽ Reference Queenan, Stewart and Smith ^{62 )}. SCFA are likely to have several effects upon health. Butyrate is largely considered beneficial for gut health as it is the major energy source for the colonocytes and has a role in CRC prevention as discussed later⁽ Reference Pryde, Duncan and Hold ^{63 –} Reference Sleeth, Thompson and Ford ^{65 )}. Propionate is metabolised in the liver and is gluconeogenic. Activation of the gut receptors G protein-coupled receptor (GPR) 41 and GPR43 (also known as NEFA receptors NEFA2 and NEFA3) by SCFA influences gut motility as well as reducing inflammatory responses⁽ Reference Brown, Goldsworthy and Barnes ^{66 ,} Reference Tazoe, Otomo and Kaji ^{67 )}. Acetate is metabolised in the peripheral tissues and is a precursor for cholesterol metabolism and lipid formation. A shift in fermentation products away from acetate (normally present at the highest concentration) towards propionate and butyrate may therefore be beneficial and explain the decrease in cholesterol levels in volunteers consuming fibre⁽ Reference Brown, Rosner and Willett ^{68 )}. Increased SCFA concentrations may also increase the solubility of certain minerals such as Ca, and enhance absorption and expression of Ca-binding proteins⁽ Reference Scholz-Ahrens and Schrezenmeir ^{69 )}. Changes in intestinal microbial metabolism following the consumption of inulin fructans has also been shown to improve bone health by increasing Ca absorption while β-glucans may lower total cholesterol levels⁽ Reference Smith, Queenan and Thomas ^{70 )}.

Phytochemicals

There is a vast amount of literature suggesting that plant secondary metabolites have properties beneficial to health⁽ Reference Russell and Duthie ^{71 )} In particular, almost all plant foods considered to have cancer-preventative properties are rich in compounds derived from the phenylpropanoid pathway. Most of this information is obtained from in vitro data and there is very little evidence from both pre-clinical and human interventions to support this. Bioavailability may be defined as the fraction of an ingested nutrient or metabolite that reaches systemic circulation and specific sites where it can exert specific biological activity. The most abundant phytochemicals consumed may not, however, be the most bio-available to the host. Bioavailability may be influenced by food composition as most of the phytochemicals are likely to be complexed with other plant components. Host factors will also influence bioavailability including enzyme activity and gut transit time. Following ingestion, the absorption of certain phytochemicals will occur in the small intestine where some glycosides will be hydrolysed. Rapidly absorbed plant metabolites will enter systemic circulation following methylation, sulphation and/or glucuronidation. However, many phytochemicals escape absorption early in the GI tract and in particular those bound to plant polymers reach the colon and are metabolised and released by the gut microbiota⁽ Reference Russell, Scobbie and Labat ^{72 )}. Data regarding the metabolism and bioavailability of these products are severely lacking, particularly with reference to the mechanisms of transformation and the species responsible. The involvement of the gut microbiota in conversion of these plant metabolites is demonstrated by the fact that their metabolites appear after 6–8 h in systemic circulation⁽ Reference Russell, Gratz and Duncan ^{8 )}. The majority of plant metabolites present in the form of glycosides will be converted to aglycones prior to further transformation. It is likely that the compounds that reach the functional sites such as cells and tissues will be chemically different from those consumed in the diet. To evaluate their potential role in cancer prevention, the structure, concentration and site of action must be established. In the colon, they can exert a direct action such as an anti-inflammatory effect on the gut mucosa or be absorbed from the colon via hepatic circulation and suppress low-level chronic systemic inflammation.

Both the parent compounds and the metabolites produced have the potential to influence specific microbial groups. Of the secondary metabolites, the group most widely studied are products of the phenylpropanoid pathway, as nearly all plant foods considered to have cancer-preventative properties are rich in these compounds⁽ Reference Russell, Scobbie and Labat ^{73 )} including cinnamic, phenylacetic, phenylpropionic, coumarins, flavonoids and anthocyanidins⁽ Reference Russell, Scobbie and Labat ^{72 )}.

Although these phenylpropanoid derivatives are most widely studied, many other secondary metabolites and their derivatives are present in the human colon. Moreover, glucosinolates and their metabolites (isothiocyanates and indoles) have also been extensively studied in relation to protection against carcinogenesis and mutagenesis⁽ Reference Higdonm, Delage and Williams ^{74 ,} Reference Mithen, Dekker and Verkerk ^{75 )}, while many other N- and S-related compounds have been generally overlooked in terms of metabolism. For some studies, the presence of specific gut metabolites was detected in plasma and urine by MS indicating that these compounds have entered systemic circulation via enterohepatic circulation. For example, urinary phenolic acid metabolites, measured by GC–MS, were significantly increased following consumption of red wine and red grape juice extracts⁽ Reference van Dorsten, Grün and van Velzen ^{76 )}. The best markers of intake included syringic acid, 3- and 4-hydroxyhippuric acid and 4-hydroxymandelic acid. Reductions in p-cresol sulphate, 3-indoxylsulphuric acid and increases in indole-3-acetic acid and nicotinic acid were also observed in urine following consumption of red wine and grape extracts⁽ Reference Jacobs, Fuhrmann and van Dorsten ^{77 )}. In addition, sesamin, a major bioactive lignin found in sesame seeds was metabolised to enterolactone by in vitro incubation with mixed faecal microbiota. Sesamin consumption also demonstrated that this compound was a precursor to enterolactone in vivo ⁽ Reference Peñalvo, Heinonen and Aura ^{78 )}. Ingestion of a range of ellagitannin-rich foods, demonstrated that the microbial metabolite 3,8-dihydroxy-6H-dibenzo[b,d]pyran-6-one (urolithin B) conjugated with glucuronic acid was detected in urine by liquid chromatography–MS⁽ Reference Cerdá, Tomás-Barberán and Espín ^{79 )}. These urolithin metabolites were also presently found to be present in human plasma⁽ Reference Seeram, Henning and Zhang ^{80 )}.

Obesity

Obesity is a major health problem both in developed and in developing nations and arises when energetic content of food ingested is in excess of energy expenditure. Excess body fat is associated with a number of metabolic diseases such as diabetes, CVD and cancer and can also have a major impact on longevity and quality of life. There has been much interest in the potential role of non-digestible dietary carbohydrates for body weight control and obesity⁽ Reference Nilsson, Ostman and Holst ^{90 )}. Drastic reduction in total dietary carbohydrate intake in weight loss diets alters the composition of the colonic microbial community as well as faecal metabolite profiles⁽ Reference Duncan, Belenguer and Holtrop ^{2 ,} Reference Duncan, Lobley and Holtrop ^{91 )}. Colonic fermentation provides an additional source of energy to the host via absorption of SCFA that is estimated to contribute about 10% of dietary energetic intake⁽ Reference Bergman ^{92 )} and bacterial fermentative activity in the colon may contribute to fat deposition⁽ Reference Backhed, Ley and Sonnenburg ^{93 ,} Reference Turnbaugh, Ley and Mahowald ^{94 )}. The energy recovered from ingested sugar by this route is, however, less than that for sugar directly absorbed in the small intestine⁽ Reference Roberfroid ^{95 )}. The net effect must therefore depend largely on how alternative sources of dietary carbohydrate influence satiety. High intakes of monosaccharides such as glucose and fructose present in soft drinks appear to increase serum ghrelin, activating hunger signals and decreasing satiety⁽ Reference Dornonville de la Cour, Lindqvist and Egecioglu ^{96 ,} Reference Robertson, Bickerton and Dennis ^{97 )}. It has been suggested that fructo-ologosaccharides intake, on the other hand, results in decreased ghrelin levels that may help in the control of food intake⁽ Reference Delzenne, Cani and Neyrinck ^{98 )}.

Obese human subjects on weight loss diets were shown to have altered microbial profiles⁽ Reference Ley, Backhed and Turnbaugh ^{20 ,} Reference Ley ^{99 )}. Also, drastic reduction in total dietary carbohydrate intake in weight loss diets alters the composition of the colonic microbial community as well as faecal metabolite profiles⁽ Reference Duncan, Belenguer and Holtrop ^{2 ,} Reference Duncan, Lobley and Holtrop ^{91 )}. Microbial metabolites associated with increased weight gain include increased excretion of hypoxanthine, hippurate, dimethylglycine and creatinine in the urine⁽ Reference Zhang, Yan and Gao ^{100 )}. Studies where weight loss was achieved via gastric bypass surgery demonstrated that asparagine, lysophosphatidylcholine (C18:2), nervonic (C24:1) acid, p-cresol sulfate, lactate, lycopene, glucose and mannose were all significantly reduced⁽ Reference Mutch, Fuhrmann and Rein ^{101 )}.

Cancer

CRC accounts for approximately 17 000 deaths in the UK annually and progression of the disease is likely to result from a combination of genetic and environmental factors. The majority of CRC is thought to be sporadic in origin with the most common form, adenocarcinoma, developing from glandular cells lining the colonic wall. There are several genetic events commonly occurring in CRC at the molecular level that include inactivation of the tumour suppressor genes such as p53 along with activation of oncogenes including the ras family of genes. Not all colorectal polyps progress to cancer, suggesting that perhaps other factors can influence malignant transformation. There is increasing evidence supporting the role of inflammation in the pathogenesis of the gut, including diseases such as CRC. Compared with healthy subjects, patients suffering from inflammatory bowel disease are considerably more likely to develop CRC⁽ Reference Rhodes and Campbell ^{102 )}. One of the similarities between inflammatory bowel disease-associated and sporadic CRC is the importance of cyclooxygenase-2 induction and this enzyme is induced in response to mediators of inflammation, cytokines and endotoxins. In the healthy colon, the thick mucin layer helps to protect the epithelial cells from direct contact with bacterial cells⁽ Reference Johansson, Johansson-Haque and Okret ^{103 )}. Moreover, tight junctions are located in the epithelium that create a barrier which regulates permeability of the epithelial layer in response to various signals including cytokines. Maintenance of an intact epithelial layer is one of the mechanisms that limit bacterial translocation. Bacterial cells may also provide a source of regulatory signals that, for example, direct the differentiation of T-helper cells producing IL-17 (T-helper 17 cells) and T-regulatory cell activity. These signals influence the maturation of the gut and the immune system. In the absence of bacteria these defence systems are likely to be weakened.

Most of the bacteria in the colon possess a number of microbial-associated molecular pattern moieties that are well recognised by cells of the innate immune system. These molecular motifs are recognised by TLR. For example, bacterial lipopolysaccharide, an endotoxin is recognised by TLR4, a recognition receptor of the innate immune system. Bacterial flagellin is recognised by TLR5 and lipoteichoic acid by TLR2. Deficiency in TLR2 has been reported to lead to both increased tumour burden and size in mice with dextran sodium sulphate-induced colitis. This has been related to increased cytokine levels of, for example, IL-6; TNFα that may drive inflammation and induce CRC. Separately, both TLR2 and TLR5 play a role in the suppression of tumour formation, via the activation of specific anti-tumour immunity. Mice deficient in TLR4 were protected against the development of neoplasia as TLR4 signalling can promote colon carcinogenesis by stimulating tumour infiltration of T-helper 17 cells. One of the key lipopolysaccharide producing species in the colon is E. coli ⁽ Reference Erridge, Duncan and Bereswill ^{104 )} although the abundance of this species is likely to be a small proportion of the total bacterial load. There have been a number of bacterial species associated with CRC including the Gram-positive Streptococcus species and the Gram-negative Helicobacter species, B. fragilis and E. coli. It is not only the balance of bacterial species in the colon that is important for maintaining gut health; in addition the impact of changing bacterial composition in response to dietary intake also drives bacterial metabolite formation.

Microbial metabolites may be a key factor in regulating inflammatory and immunological responses in the colon. Metabolism of dietary carbohydrates will result in the formation of SCFA and as discussed earlier butyrate plays an important role in health maintenance. Increasing the protein content of the diet particularly by increasing red meat intake is likely to result in increased levels of toxic metabolites that include heterocyclic amines, fecapentaenes, nitrosamines, super oxide radicals and hydrogen sulphide⁽ Reference Gill and Rowland ^{105 )}. Fecapentaenes are mutagens that are reportedly synthesised by Bacteroides species that may alkylate DNA to form mutagenic adducts. Hydrogen sulphide is also a toxic microbial metabolite formed by sulphate reducing bacteria, including Desulfovibrio piger, in the colon. D. piger can metabolise lactic acid that may accumulate in bowel disease⁽ Reference Vernia, Caprilli and Latella ^{106 ,} Reference Vernia and Cittadini ^{107 )} while reducing sulphate to sulphide⁽ Reference Marquet, Duncan and Chassard ^{108 )} and meat is a major source of sulphur that promotes the growth of sulphate reducing bacteria⁽ Reference Christl, Gibson and Cummings ^{109 )}. The genotoxic potential of hydrogen sulphide is in part mediated by oxidative free radicals and cyclooxygenase-2 is up-regulated in epithelial cells following administration of hydrogen sulphide at physiological concentrations⁽ Reference Magee, Richardson and Hughes ^{110 )}. Colonic bacteria also play a role in the formation of N-nitroso compounds, the levels of which are elevated following intake of high protein diets, particularly meat⁽ Reference Russell, Gratz and Duncan ^{8 )}. Cooking meat generates heterocyclic amines that can be further transformed to genotoxic intermediates. The production of these products is likely to be linked to increased risk of CRC⁽ Reference Hughes, Magee and Bingham ^{111 )}.

Secondary bile acids, mainly deoxycholic acid and chenodeoxycholic acid, are formed by microbial conversion of the primary acids that are formed in the liver and secreted into the duodenum. Approximately one litre of bile enters the duodenum each day; however, bile acid excretion is related to fat and red meat intake that are potential risk factors for CRC⁽ Reference Reddy ^{112 )}. Epidemiological studies reported higher concentrations of secondary bile acids in CRC patients compared with healthy controls⁽ Reference Gill and Rowland ^{105 )}. Secondary bile acids can cause DNA damage⁽ Reference Gill and Rowland ^{105 )} by the production of oxygen radicals and reactive nitrogen species⁽ Reference Payne, Bernstein and Bernstein ^{113 ,} Reference Dvorak, Payne and Chavarria ^{114 )}. Bile acids enter enterohepatic circulation therefore the concentration decreases along the GI tract, however, elevated bile acid levels may modulate the abundance of certain bacterial species including F. prausnitzii ⁽ Reference Lopez-Siles, Khan and Duncan ^{115 )} that has been reported to have potent anti-inflammatory activity⁽ Reference Sokol, Pigneur and Watterlot ^{116 )}.