Review

Computational tools for modeling xenometabolism of the human gut microbiota Martina Klu¨nemann, Melanie Schmid, and Kiran Raosaheb Patil Structural and Computational Biology Unit, European Molecular Biology Laboratory, Meyerhofstrasse 1, 69117 Heidelberg, Germany

The gut microbiota is increasingly being recognized as a key site of metabolism for drugs and other xenobiotic compounds that are relevant to human health. The molecular complexity of the gut microbiota revealed by recent metagenomics studies has highlighted the need as well as the challenges for system-level modeling of xenobiotic metabolism in the gut. Here, we outline the possible pathways through which the gut microbiota can modify xenobiotics and review the available computational tools towards modeling complex xenometabolic processes. We put these diverse computational tools and relevant experimental findings into a unified perspective towards building holistic models of xenobiotic metabolism in the gut. Gut microbiota as a site of xenometabolism The gut microbiota has been shown to modify or metabolize several kinds of xenobiotics, from novel cancer drugs through millennia old analgesics to dietary components [1–7]. Recent studies have also highlighted the feasibility of exploiting and manipulating this microbe-mediated xenometabolism to improve the host health or to prohibit medicinal side effects. For example, Wallace et al. [6] recently showed that a deleterious biotransformation of the cancer drug irinotecan can be averted by inhibiting bacterial b-glucuronidase. On a more general level, probiotic bacteria like Lactobacillus sp. have been shown to ease Clostridium difficile-associated diseases, diarrhea, and other side effects of antibiotics [8,9]. Owing to the advances in various omics technologies, molecular pathways of xenometabolism in the gut microbiota have now started to unfold through the identification of responsible microorganisms and enzymes [6,10,11]. With the help of metagenomics tools, it is now possible to determine the identity of a large fraction of the microbial species colonizing the human gut [12,13]. These tools are also revealing the genetic repertoire of the gut microbiome Corresponding author: Patil, K.R. ([email protected]). Keywords: metabolic modeling; microbial communities; functional metagenomics; biotransformation. 0167-7799/$ – see front matter ß 2014 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.tibtech.2014.01.005

Glossary BRENDA: Braunschweig Enzyme Database – one of the main collections of enzyme function and activity data. The database contains several xenobiotic– enzyme interactions (for examples, see Figure 2). Constraint-based modeling: a metabolic modeling technique that uses mass balance, reaction directionality and metabolite uptake/secretion constraints to estimate intracellular fluxes (reaction rates) for a given metabolic network. Enterohepatic cycle: circulation of native or xenobiotic compounds between liver and gastrointestinal tract. Hepatic biotransformations that increase water solubility of the compounds, for example, glucuronidation of bilirubin, is a common denominator for most known enterohepatic circulations. Enzyme promiscuity: property of enzymes whereby reactions are catalyzed with a degree of non-specificity. A promiscuous enzyme can act on multiple substrates and/or produce different products starting with the same substrate. Genome-scale metabolic model: a comprehensive mathematical representation of metabolic capabilities of a cell. A typical bacterial model consists of a network of several hundreds of reactions that are compiled based on genomic, biochemical and other evidences. KEGG: Kyoto Encyclopedia of Genes and Genomes. A comprehensive database of high-level gene functions and their organization in a pathway or cellular context. The well-maintained and annotated pathway maps, reaction modules, and drug database of KEGG are of particular interest for modeling enzyme promiscuity and predicting xenometabolic pathways. Machine learning algorithms: computational methods aimed at prediction based on patterns learned from a given input dataset (training data). In the absence of mechanistic models, machine learning is a useful tool for predicting whether a xenobiotic compound of interest would be susceptible for chemical modification by a particular enzyme. Known enzyme–compound relations from databases like BRENDA, KEGG, and UM-BBD can be used as training data. Phase I and II metabolism: metabolic reactions that predominantly occur in the liver and gastrointestinal epithelial tissue. Whereas the phase I enzymes introduce reactive or polar chemical groups in the xenobiotic compounds, the phase II enzymes catalyze conjugation reactions. Together, phase I and II metabolism alters the activity of xenobiotics and/or converts them into more water-soluble compounds, often leading to detoxification and eventual excretion. Structure-activity relation (SAR) paradigm: SAR paradigm assumes that similar molecules have similar biological activities. Thus, the activity of a xenobiotic compound of interest can be described based on the activity of known compounds with similar physicochemical/molecular properties. Different biochemical activities depend on different physicochemical properties, therefore, the most critical step in the application of SAR paradigm is to identify a similarity metric that is most appropriate for the activity in question. A quantitative approach to structure–activity relation (QSAR) allows predicting the degree of expected activity based on the degree of similarity. Site of metabolism: site in a chemical compound at which the major chemical change takes place during a given biochemical transformation. Thermodynamic feasibility analysis: feasibility analysis for a particular biochemical transformation based on estimation of the overall change in the Gibb’s free energy (DG) accompanying the reaction. Only reactions with negative DG values are thermodynamically feasible. UM-BBD: University of Minnesota Biocatalysis/Biodegradation Database; contains extensive data on microbial biocatalytic reactions and biodegradation pathways. Xenobiotics: molecules of foreign origin encountered by the body, such as drugs and dietary compounds that are not naturally produced by the human body. Xenometabolism: enzyme-mediated biochemical transformation/degradation of xenobiotics. Trends in Biotechnology, March 2014, Vol. 32, No. 3

157

Review

Trends in Biotechnology March 2014, Vol. 32, No. 3

in unprecedented detail. The resulting rich datasets are enabling the characterization of the gut microbial communities and their association with health [14]. Metagenomics datasets are also providing a starting point for modeling the collective metabolic behavior of the gut microbiota [15,16]. In parallel to these advances stemming from metagenomics studies, more experimental evidence is piling up supporting the key role of gut microbiota in xenometabolism [3–5]. Many of the aforementioned studies have gained useful insight from or were validated in animal models. Indeed, animal models have been fundamental for understanding the biology of gut microbiota in general [17]. Complementing metagenomics, metabolomics has made it possible to trace the metabolic fate of xenobiotic compounds [18–20], leading the recent resurge in the research on xenometabolism [21–23]. Role of modeling in tackling complexity of xenometabolic processes Xenometabolic processes in the gut can be highly complex due to three main reasons: the widespread promiscuity of metabolic enzymes; the compositional complexity of the

gut microbiota; and the interactions between the host and the microbe-mediated xenometabolism. The promiscuity of metabolic enzymes [24–26] means that the number of possible routes through which a xenobiotic compound can be metabolized or modified increases combinatorially with the enzymatic repertoire of the microbiota. The compositional diversity and spatial heterogeneity of the microbiota and the host–microbiota interaction through the enterohepatic cycle add another layer to this complexity. These inherent complexities and the limitations in obtaining in vivo data from human subjects raise the appeal for computational modeling of xenometabolism. A holistic modeling platform accounting for combined host– microbiota xenometabolism can be instrumental in drug development and for devising personalized medicine strategies. Such a platform would allow prediction of potential xenometabolic pathways and thereby generation of testable hypotheses. Although no single platform currently tackles all of the distinct challenges in modeling xenometabolism, several tools are available that are capable of addressing the key individual steps. These tools range from prediction of enzyme-level biotransformation [27–30] to

Table 1. Representative computational tools relevant for modeling xenometabolism by the gut microbiota Tool

Original scope

Single-step biotransformation prediction Sites of human phase I metabolism MetaPrint2D Subset of biotransformations in phase I metabolism Metabolites of phase I and phase II Meteor metabolism Generic MaRIboES Generic Chen, Feng et al. Sites of enzymatic attack Mu, Unkefer et al. Biotransformation pathway prediction Biodegradation by plants and MetabolExpert animals Biodegradation by aerobic bacteria UM-PPS Biodegradation by aerobic bacteria CRAFT Generic BNICE MetaPrint2D-React

Microbial biodegradation Microbial biodegradation to host native metabolites Species metabolic modeling Bacterial genome-scale metabolic Model SEED models Generic COBRA Toolbox PathPred Desharky

Generic OptFlux Microbial community modeling Generic; continuous growth OptCom KIitgord and Segre` Freilich et al.

Induction of cooperation in communities Competitive and cooperative interactions

Abbreviations: EC, Enzyme Commission; –, no information found. a

Tool is open source, database is excluded.

b

http://www.molecular-networks.com/products/craft.

c

Works within the licensed MATLAB framework.

158

Main features/techniques

Web access/ open source

Refs

Comparison to reaction centers from the proprietary Symyx Metabolite database Based on Metaprint2D

Yes/yes a

[30]

Yes/–

[30]

Reaction rules based on multiple knowledge bases

No/no

[44]

Generalized reactions based on BRENDA Machine learning Reaction rules based on KEGG; machine learning

No/yes No/upon request Yes/no

[28] [48] [50]

Knowledge based reaction rules

No/no

[45]

Reaction rules based on UM-BBD; expert knowledge Reaction rules based on UM-BBD Generalized reactions based on EC numbers; thermodynamic feasibility analysis Machine learning based on KEGG Pathway search using KEGG; host-organism specific

Yes/no No/yes No/no

[29]

Yes/no No/yes

[47] [85]

Constraint-based modeling; automated model construction Constraint-based modeling; MATLAB and Python based Constraint-based modeling; Java based

Yes/yes

[57]

No/yes c

[61]

No/yes

[86]

Constraint-based modeling; accounts for growth requirements of individual species Constraint-based modeling; focus on nutritional composition of growth media Constraint-based modeling; pairwise interactions

No/academic use

[31]

No/yes

[32]

No/–

[84]

b

[27]

Review

Trends in Biotechnology March 2014, Vol. 32, No. 3

modeling of metabolic interactions in microbial communities [16,31–33]. Here, we review these diverse tools and present the main concepts underlying their functioning (Table 1). We also present several examples of xenometabolism and discuss how computational tools that address different aspects of the xenometabolic processes could be combined into a single platform. Hierarchy of xenometabolic processes in the gut The general metabolic processes that a xenobiotic compound can undergo in the gut microbiota can be conceptually organized into three levels: community, species, and enzymes (Figure 1). Complex xenometabolic pathways often emerge through the functional interplay within and across these hierarchical levels. At the topmost level, the spatial and compositional structure of the microbial community influences the survival, activity, and procreation of species in the gut environment, and hence the overall xenometabolism. At the intermediate level, individual species determine and control the enzyme availability for the xenometabolism. At the bottom level, the enzymes perform the actual biotransformations owing to their

promiscuity. From a modeling perspective, the bottommost enzyme level forms the basis for simulating xenometabolic processes at higher levels. We therefore present the modeling tools for the three hierarchical levels in a bottomup order. Enzyme level: promiscuous enzymes drive and enlarge xenometabolism Enzymes can often catalyze more reactions than those listed in pathway databases like KEGG [34] and may exhibit functions and biochemical features beyond the current description [24,25]. For example, xenobiotic metabolism in the liver is driven by highly promiscuous enzymes like cytochrome P450 oxidases and glutathione S-transferases [35]. Indeed, given the prominent role of the liver in xenobiotic detoxification (Box 1), numerous enzyme–xenobiotic relationships have been described in the context of liver metabolism [35–37]. In microbial systems, enzyme promiscuity has been so far investigated mainly in the context of bioremediation of toxic compounds from the environment [38], or in the context of biotechnological production of valuable chemical compounds [39].

(B)

(A)

(C)

H

CH3 H3C

H3C OH H3C

CH3

OH

H3C

OH

H CH3 CH3 HO

CH3 OH

OH

H3C H3C

OH OH

H3C

CH3 O

H3C

CH3

OH

Key:

CH3 CH3 CH3 O

Orally administered xenobioc

Absorpon into organism

Metabolic enzymes

Degradaon intermediates

Biotransformaon

Transport protein

Gut microbes

Inhibion of growth

Transport reacon TRENDS in Biotechnology

Figure 1. Hierarchical organization of xenobiotic metabolism in the gut microbiota. (A) Community-level xenometabolism. A xenobiotic or its derivatives can be absorbed and/or modified by the host, excreted from the gut, or modified by the gut microbiota in many alternative ways. For host contribution to the xenometabolism, which is not shown here for the sake of clarity, see Box 1. Different species in the microbiota can transform a given xenobiotic into different compounds, which can be further metabolized by the same or different microbes. Depending on the metabolic status of certain bacteria, a xenobiotic might be degraded or not. The xenobiotics and the degradation intermediates are also affected by the structure of the microbiota and vice versa. (B) Species-level xenometabolism. Xenometabolic enzymes are usually found in the cytosol of microbial cells, but some can be secreted as well. A xenobiotic compound can undergo different biotransformations within a microbe before its metabolites are exported into the gut lumen, or used by the microorganism as a nutrient. (C) Enzyme-level xenometabolism. Promiscuous enzymes like cytochrome P450 have broad substrate specificities and can biotransform different xenobiotics. Enzyme promiscuity can also lead to different modifications of a given xenobiotic compound.

159

Review

Box 1. Host–microbiota co-metabolism of xenobiotics After ingestion and passage of xenobiotics through the stomach, the alkalization of the intestinal content is crucial for the enzyme activity and hence for xenometabolism within the small intestine. Absorption into the bloodstream can occur by many different ways such as active transport, facilitated diffusion, pinocytosis, or passive diffusion. Absorbed substances are transported via the portal vein to the liver, where metabolism of most xenobiotics takes place. The bioavailability of a xenobiotic can be greatly reduced by the extent of intestinal absorption and the so-called first-pass metabolism occurring in both the liver and the small intestine. Biotransformation reactions at these sites are generally divided into phase I and phase II metabolism, which together convert xenobiotics to more watersoluble compounds [87–89]. Additionally, during the enterohepatic circulation, an unchanged xenobiotic or its biotransformed metabolite can be excreted back into the small intestines via the bile. Xenobiotics can thus be further biotransformed through repetitive cycling [4,6] or secreted out. Together, the intestinal absorption barrier, phase I and II metabolism, and excretion constitute the defense mechanism of the human body against foreign (toxic) substances. The intestinal microbiota adds to the possibilities and complexity of xenometabolism (see main text). The interconnectivity between the intestinal tract and other metabolic compartments makes it essential to view xenometabolic processes as co-metabolism by the host and microbiota. The examples of such co-metabolism include prodrugs like mesalazine, which are activated by the gut microbiota and then detoxified by the liver [2]. Another prominent example is irinotecan, a cancer drug, which is first glucuronidated by the liver and then, through enterohepatic circulation, transferred back to the gut, where it is further metabolized by the gut microbiota [6]. Intriguingly, the xenobiotic co-metabolism can be further interconnected by the mutual regulation of host and microbiota gene expression in response to a xenobiotic. Xenobiotics can alter the composition of the intestinal microbiota as well as the microbial gene expression [64]. Furthermore, the intestinal microbiota can mediate changes in the hepatic gene expression and the xenometabolism thereof in response to a xenobiotic [90].

Specific links between metabolism of a xenobiotic and the responsible enzyme are still relatively scarcely known for the gut microbes. For known examples of biotransformation of xenobiotics mediated by the gut microbial enzymes, and its similarity to bioremediation, see the review by Haiser and Turnbaugh [11]. Relatively more cases of enzyme–xenobiotic interactions are known in bacterial systems outside the context of the gut microbiota. Figure 2 shows selected examples of such interactions, highlighting the promiscuous action of enzymes as the key molecular mechanism responsible for xenometabolism. Owing to their promiscuity, enzymes with similar biochemical functionality can process structurally similar molecules. The currently known examples of xenobiotic biotransformations thus represent templates for predicting new biotransformations. Predicting single-step biotransformations The propensity of enzymes to biotransform structurally similar compounds is central to the modeling of xenobiotic metabolism. Indeed, several approaches for predicting enzyme–compound interactions aim at exploiting the socalled structure-activity relation (SAR) paradigm [40] (see Glossary). The SAR paradigm postulates that, given an established enzyme–compound relationship (such as inhibition, activation or biotransformation), other chemically similar compounds will have the same relation to the same 160

Trends in Biotechnology March 2014, Vol. 32, No. 3

or similar enzymes. Although theoretically attractive, the applicability of SAR-based methods is often limited due to difficulties in defining and quantifying chemical similarity and in relating it to the functional similarity [41]. For example, negatively charged oxygen ions and positively charged nitrogen ions are similar in terms of their polarity, whereas they are very different in terms of electrostatic potential or as hydrogen acceptors. Nevertheless, several quantitative SAR (QSAR) modeling approaches are available for predicting enzyme–compound relations and have shown some degree of predictive power [42]. QSAR approaches are at present mainly used for predicting inhibition of enzyme activity by a given compound, and less commonly for predicting biotransformations [43]. A common approach towards predicting biotransformations is based on identifying sites of metabolism or reaction centers in the query compound. MetaPrint2D [30] is an open-source tool utilizing this approach. In particular, it identifies the sites of oxygen addition and predicts the elimination reactions of the phase I metabolism. Its experimental sister program, MetaPrint2D-React, is aimed at predicting the types of transformation that can take place at each site of the phase I metabolism and the resulting metabolite formed. For an extensive review on QSAR and sites of metabolism tools see Kirchmair et al. [43]. Another approach that is commonly used for predicting biotransformation is the use of reaction rules, which is discussed below in the context of xenometabolic pathway prediction. Predicting biotransformation pathways The large enzymatic repertoire and the complexity of the gut microbiota require consideration of xenobiotic metabolism beyond single-step biotransformations (Figure 1). The currently available tools that are specifically aimed at predicting whole pathways of xenobiotic metabolism primarily concern human or mammalian enzymes. Tools like Meteor [44] and MetabolExpert [45] are used for predicting degradation of drug-like compounds by the human metabolism and formation of potential toxic metabolites. These and other similar tools generally rely on curated databases of reaction rules describing the biochemical reaction capabilities of the enzymes, and some also take into account the drug-like features of the query molecules, for example, lipophilicity. Pathway prediction tools are also emerging for designing synthetic biochemical routes to enable production of industrially valuable compounds and for predicting biodegradation of harmful chemicals [46]. As these tools focus on microbial enzymes, they bear a more direct relevance for xenometabolism by the gut microbiota. Prediction tools specialized for biodegradation include UM-PPS [29] and (http://www.molecular-networks.com/products/ CRAFT craft), which use a comprehensive biocatalysis/biodegradation database from the University of Minnesota [38] as a knowledge base for bacterial degradation reactions. Tools more geared towards novel or synthetic pathway prediction include BNICE [27], MaRIboES [28], PathPred [47], and others [48–50]. In general, the pathway prediction algorithms use a core module based on reaction rules to predict likely products of biotransformation for a given query compound. Subsequently, these products are used as

Review

Trends in Biotechnology March 2014, Vol. 32, No. 3

S I

I

I I

P P P P

I

I

P A

I

I

S

I

I

I S P I

S

I P A IPS S A A I

A

I S

IS P IS

I

IPS A S

S PS P

S

S A

P I

I

I

No. of Enzyme– compound interacons 1–3 4–7 8+

P P

S I

S

I

I

S A

IP

S I IPS

I

I

I

P I

S S S S I

I I

I

S

S I

I

IS I S

I S P

IPS

Ac

n g Ac on Ac ng CHn on OH Ac g d g n Ac on iph rou go n pe en p np g o rox ols of d Ac air hy ide or on n ed dr as th or go do og a e l s A n a c no en cce ike Ac lde ng rs, a p Ac ng hyd on oxy Oxy s do tor n o e/ CH ge ge no n n o Ac g on CH xo or in nas r n C -C gr CH vo es l o H Ac g on -N H gr up 2 groved n CH H2 oup of d up g g s Ac on Ac-NH rou of dono n ot ng gr p o on rs Tr g o her o ou f d or an n N n N p o o n s Tr sfe he g A f o r an rri me rou DH don s sfe ng g ps /N o rri on rou as AD rs ng e- p do PH alk c o Tr y an l/a G Ac arbo f donors Tr sf ry lyc yl n n o a e l t Tr nsfe rrin gro osyl rans grou rs an rr g up tra fe p sfe in nit s ns ra s rri g P rog (no fer ses ng -co en t m as S n o e es Ac -contain us g thy n t a ing ro l) g o i n i gr up n e n g ou s Ac s gr p n Gl ter o u p s Ac g o yc bo s n n g o oth Peosyl nds n a er p ase A Ac cn cid C-N das s n g an bo es g o o n h y nd n h C - C dri s ali b des de o n C- b o d s C n C- lya d s O se C -N lyas s Ep C- lyases Int ime S e ra ras m es O P-O lyas s ole a th ly es n cu d e as Fo lar racr lya es Fo r t e s rm Fo min rans ma es ing s r g F ph o min C-Ofera es os rm g C b ses ph ing - S on or C b ds ic -N o n es b d s te on r b ds on ds

EC EC 1.1 1 EC .10 1 EC .11 1 EC .12 1 EC .13 1 EC .14 1.1 EC 7 1 E C .2 1 EC .3 1 EC . 4 1.5 EC EC1.6 1 EC .7 1 EC .9 2 EC .1 EC2.3 2 EC .4 2 EC .5 2 EC .6 E C 2.7 2 EC . 8 EC3.1 3 E C .2 3 EC .4 3 EC . 5 3 E C .6 3.7 EC EC3.8 4 E C .1 E C4 . 2 4 EC . 3 EC4.4 EC 4.6 4 EC.99 5 EC .1 5.4 EC 6 E C .1 6 EC . 2 6 EC . 3 6.5

S

S S I

S

S

I

Drugs

5-Aminosalicylic acid Acetaminophen Chloramphenicol Genistein Glyceryl trinitrate Hesperidin Lactulose S Methotrexate Methylmercuric chloride Metronidazole Nicone I Phenacen

IS AIS P PS I I IS S S P IS P P S S IPS IS AS A S P I I P P S

Dietary compounds

Benzoic acid P Bilirubin Caffeine Carnine PS Catechin Chloroacec acid Chlorogenic acid I S L-Dopa S PS Phosphadylcholine A A Phthalate P P Propyl gallate S Vanillin PS

TRENDS in Biotechnology

Figure 2. Examples of enzyme–xenobiotic interactions in bacteria. Each column corresponds to a different enzyme class according to the Enzyme Commission (EC) nomenclature. Enzyme promiscuity is the key driver of xenobiotic metabolism, whereby a xenobiotic compound can often be transformed by several different enzymes and vice versa. The shown examples were obtained from the BRENDA database [52]. Abbreviations: A, activating; I, inhibiting; P, product; S, substrate.

new query compounds for identifying the next-step products, and this prediction cycle is continued until the desired pathway length or a target molecular structure is reached. The number of predicted products often increases combinatorially with each successive round, thus posing a challenge for reducing the number of identified pathways that are biologically irrelevant. To prohibit this combinatorial explosion, the prediction systems often rely on expert user input for prioritizing feasible pathways and for estimating the likelihood of transformation depending on the expected chemical environment [29]. In a more unbiased approach, the BNICE framework [27] from the Hatzimanikatis laboratory uses thermodynamic feasibility analysis to rank the identified pathways. Their methodology allows considering a single biotransformation step in the context of a whole pathway and can thereby help in eliminating biologically unfeasible pathways. Chemical similarity of query compounds to the natural (or known) substrates of enzymes has also been suggested as a filter for circumventing the problem of combinatorial explosion [51]. In an approach using this concept, De Groot et al. [28] used biochemical similarity of the enzymecatalyzed chemical transformations from BRENDA [52] to identify groups with distinct reaction patterns. Subsequently, they trained a machine-learning algorithm,

termed MaRIboES, with structural and stereochemical similarity metrics and molecular fingerprints of the known substrates. If the query molecule is sufficiently similar to a known substrate of an enzyme, MaRIboES predicts that it will be transformed accordingly. The web-based tool PathPred [47] also uses reaction patterns; however, it relies on chemical similarity of substrate–product pairs as found in the KEGG database. An additional strategy for reducing the combinatorial space of the predicted biotransformation pathways could be to take advantage of the evolutionarily conserved patterns of successive biochemical reactions. In a recent study, Muto et al. [53] found several such conserved sequences of reactions among the KEGG metabolic pathways. These conserved reaction modules can include up to six consecutive reactions and are now part of the KEGG database. These modules can be seen as building blocks of the enzyme-catalyzed reaction networks and could be used to screen for biologically relevant xenometabolic pathways. Species level: enzyme availability and interaction between xeno- and native bacterial metabolism One of the well-known examples of xenometabolism that is specific to a particular gut bacterium is the metabolism of digoxin by Eggerthella lenta [10]. Association between 161

Review bacterial species and specific metabolites have also been observed in several other studies [5,6,54,55]. Although species-level specificity of xenometabolism is known only for a small number of compounds, the known associations can be used to narrow down the list of candidate biotransforming species for structurally similar xenobiotics. Whereas the nature and the abundance of different enzymes harbored by a species will determine the possibilities and limits of the xenobiotic metabolism, the interactions between xeno- and native metabolism will affect the dynamics and efficiency of the actual xenometabolic pathways. The first step towards modeling species-level xenometabolism will be to assess the enzymatic repertoire and to map the corresponding metabolic network. Several software tools are available for rapid gene annotation and for obtaining draft metabolic network reconstructions based on genomic or metagenomic data [56,57]. Reconstruction of high quality genome-scale models still requires several manual curation steps, which limits their application to complex gut microbial communities. So far, genomescale models are available for only about 10 gut microbial species, representing commensal, probiotic, and pathogenic bacteria [16]. Genome-scale metabolic models have been in use for several years and excellent computational tools are readily available for various analyses [58–61]. Among other applications, these models can be used to assess metabolic capabilities of the microbes and to simulate effects of genetic or environmental perturbations on their growth. For example, metabolic models can help pinpointing the nutrients that a bacterium feeds on and the metabolites it excretes under a given nutritional environment [60,61]. For an extensive review on constraintbased metabolic modeling see Lewis et al. [61] and Zomorrodi et al. [59]. The challenge of modeling interactions between the microbial native metabolism and xenometabolic pathways is yet to be directly addressed. A recent study on synthetic pathway design using the BNICE framework [62] provided insight into the nature of such interactions. In that study, the authors assessed the impact of expressing different synthetic pathways for biodegradation of 1,2,4-trichlorobenzene on the native metabolism of Pseudomonas putida. The analysis identified key features of synthetic routes, such as reducing power requirement and oxygen utilization, which would likely affect the bacterial growth. Similar analysis can be used to evaluate the impact of a given xenometabolic pathway on the growth of gut microbes. Moreover, the reduction potential and other physicochemical parameters may vary within the gut microbiota, inciting the need for using thermodynamic considerations suggested by BNICE. The metabolic activity status of a bacterium can also have a strong influence on the likelihood of biotransformations, which can subsequently affect the entire community [63]. Some xenobiotics can directly influence microbial metabolism, for example, by invoking changes in gene expression [64], thereby increasing the challenges for modeling. Metatranscriptomic and metaproteomic studies [65,66] can, in principle, help identifying the metabolic state of species of interest and in restricting the biotransformation possibilities to the expressed enzymes. Further 162

Trends in Biotechnology March 2014, Vol. 32, No. 3

advances in these meta-omics technologies are needed before species-level expression can be inferred with coverage sufficient to allow modeling. Algorithms based on genome-scale metabolic models [58,67,68] can offer additional help in understanding the response of microbial native metabolism to the perturbations introduced by xenobiotics. Community level: community structure determines the xenometabolic pathways A typical gut microbiota consists of hundreds of diverse microbial species [13,69,70]. This compositional complexity, together with the spatial heterogeneity of the microbiota [71–73], poses arguably the biggest challenge for modeling xenometabolism in the gut. A microbial consortium can transform a certain xenobiotic compound in qualitatively different ways than any single species (Figure 1). A community is especially more likely to perform multiple consecutive transformation steps due to the larger enzymatic repertoire. Moreover, multiple species can harbor a particular biotransforming enzyme, thereby increasing the likelihood that the enzyme would be expressed by at least some of the species under a given physiological condition. The spatial differences of the gut microbiota within the intestinal tract, from mucus layer to lumen, or proximal to distal intestines, is another crucial factor for xenometabolism [71–73]. The composition of the gut microbiota is strongly influenced by several environmental and hostdependent factors including nutrient supply, peristaltic movements, and the host immune system [74,75]. In turn, the gut microbiota can act as an ecosystem engineer influencing some of these factors [76]. In terms of species composition, individual gut microbiota are often considerably different from each other [13]. Interestingly, these diverse microbiotas might converge regarding their functional repertoire, for example, when seen from the viewpoint of represented metabolic capabilities [13,77]. In a recent metaproteomic analysis, Kolmeder et al. [78] observed temporally stable expression for a core protein pool of the human intestinal microbiota. These observations suggest that different microbiota may exhibit common functionalities in terms of xenometabolism despite compositional dissimilarities. However, the dependency of xenometabolism on the microbial community composition can be highly complex and hence the functional implications of the convergent metabolic potential remain to be evaluated. A source of diversity in xenometabolism that can arise even between species with similar metabolic capabilities is the disparity in their ability to secrete enzymes, and to uptake/excrete xenobiotics and derivative xenometabolites. A given xenometabolic process may involve a complex combination of intra- and extracellular biotransformation processes (Figure 1). In the gut lumen, secreted enzymes can transform the original xenobiotic compound or its metabolic derivatives secreted by other microbes. Inside the cells, biotransformation is limited to the enzyme repertoire of the respective bacterium, but likelihood of biotransformation may be higher due to closer proximity between the xenobiotic and the transforming enzymes.

Review

Trends in Biotechnology March 2014, Vol. 32, No. 3

Secreted enzymes, native metabolites and xenometabolites can positively or negatively affect the whole community [79,80]. Thus, the xenometabolic processes and the gut microbiota can reciprocally affect each other. This interdependency is currently beyond the reach of modeling tools due to the lack of knowledge of the involved mechanisms. Yet, computational tools are emerging that are able to tackle the first challenge in this direction, namely modeling of metabolic interactions in microbial communities [18]. These tools either use network topology to identify potential metabolic links between the community members [15,81,82] or extend the constraint-based modeling approaches for simulating interspecies metabolic fluxes [31–33,83,84]. The constraint-based methods have an advantage in that they account for the mass balance and also take into account the fundamental growth and maintenance requirements of the community members. In essence, the community metabolic models simulate exchanges of metabolites within the member species, subject to the constraints of nutrient availability from the abiotic environment. These models often assume that the individual species, at least to some extent, maximize

Input data Database of known transformaon rules

their growth efficiency [31]. Although promising, these models are currently limited to relatively small communities in comparison to the scale of the human microbiota. Increasing scientific and technological interest in this field is expected to bring these models up to the challenge of tackling the microbiota-scale communities and eventually that of simulating xenometabolic pathways. Limitations of the current tools for modeling xenometabolism Although the above-discussed computational tools can, in principle, be used to model the individual steps of xenometabolism, there are several critical limitations that must be overcome before their application. Most of these tools originate from diverse disciplines like bioremediation and industrial biotechnology and are currently restricted to applications in their respective fields, and sometimes restricted to even narrower contexts (Table 1). Their power for predicting gut microbe-mediated biotransformation thus remains to be tested and may hinge upon the availability of new training datasets more relevant to the gut microbiota.

Predicon of biotransformaon

Pathway predicon

Pathway analysis

Microbiome-mediated xenometabolism

Xenometabolism pathway reconstrucon

Targets for modificaon of xenometabolism

• Based on reacon rules and molecular similarity • Filtered by thermodynamic feasibility analysis

Metagenomics/transcriptomics m

AC TATAT... AGG T CC ... G C G TT A...

Metabolites

• Enzyme repertoire p nz abundance • Species/enzyme -1 1 0 -1 0 0 -1 1 1 -1 0 1 Reacons

Subset of transformaon rules Xenobioc compounds

Predicted xenometabolites

In vivo assessment and applicaon

Physiological status of gut

In silico assessment Xenobioc compound of interest

Liver-mediated xenometabolism

• Toxicity • Binding to host proteins • Pharmacokinecs

Assessment of impact on host TRENDS in Biotechnology

Figure 3. Proposed framework for computational prediction and analysis of xenometabolic pathways in the gut microbiota. The figure illustrates how existing bioinformatics and modeling tools from diverse disciplines can be brought together towards holistic modeling of xenometabolism. Metagenomics and other metaomics datasets will form one of the main inputs for the iterative cycle of biotransformation predictions. Degradation/modification of xenobiotics by the host will need to be an integral part of this cycle (Box 1). The predicted xenometabolic steps can be used to infer higher order pathways. Subsequent identification of key nodes (enzymes and bacteria) for predictive manipulation of the xenometabolic pathway of interest will provide testable hypotheses for in vivo assessment.

163

Review Box 2. Outstanding questions  What is the contribution of spatial organization and heterogeneity of the gut microbiota to xenometabolism?  How can enzyme availability for xenometabolism be assessed?  How transferable is the knowledge from bioremediation/biodegradation to the xenometabolism performed by gut microbes?  What are the principles underlying crosstalk between the microbial native metabolism and xenometabolism?  How can the number of biologically irrelevant solutions be reduced while predicting potential biotransformation/biodegradation pathways for xenometabolites?  How can modeling tools from different hierarchical levels be seamlessly integrated?

Concluding remarks and future perspectives Despite the proven relevance of the microbiota-mediated xenometabolism to human health, there are only a few published approaches that specifically tackle the task of modeling the underlying complex biochemical processes. Although several challenges remain (Box 2), integration of the available tools into a single platform can be a powerful approach towards holistic modeling of xenometabolism. We envision such an integrative approach as illustrated in Figure 3. For physiologically relevant predictions, the iterative process of xenometabolic modeling of the gut microbiota will need to be tightly coupled to the established liver biotransformation prediction tools. Predicted xenometabolic reactions should be further analyzed by reconstructing the biochemical, species-level and communitylevel pathways. These xenometabolic reconstructions will form the basis for assessing the impact on the host and ultimately for identifying the intervention nodes (enzymes or microbial species) towards controlling and manipulating a given xenometabolic pathway. Acknowledgments We thank A. Zelezniak and S. Andrejev for comments on the manuscript. M.S. was supported by the EMBL interdisciplinary postdoctoral program.

References 1 Goldman, P. et al. (1974) Metabolism of drugs by microorganisms in the intestine. Am. J. Clin. Nutr. 27, 1348–1355 2 Azad Khan, A.K. et al. (1983) Tissue and bacterial splitting of sulphasalazine. Clin. Sci. 64, 349–354 3 Sousa, T. et al. (2008) The gastrointestinal microbiota as a site for the biotransformation of drugs. Int. J. Pharm. 363, 1–25 4 Clayton, T.A. et al. (2009) Pharmacometabonomic identification of a significant host–microbiome metabolic interaction affecting human drug metabolism. Proc. Natl. Acad. Sci. U.S.A. 106, 14728–14733 5 Zheng, X. et al. (2013) Melamine-induced renal toxicity is mediated by the gut microbiota. Sci. Transl. Med. 5, 172ra22 6 Wallace, B. et al. (2010) Alleviating cancer drug toxicity by inhibiting a bacterial enzyme. Science 330, 831–835 7 Wang, Z. et al. (2011) Gut flora metabolism of phosphatidylcholine promotes cardiovascular disease. Nature 472, 57–63 8 Cimperman, L. et al. (2011) A randomized, double-blind, placebocontrolled pilot study of Lactobacillus reuteri ATCC 55730 for the prevention of antibiotic-associated diarrhea in hospitalized adults. J. Clin. Gastroenterol. 45, 785–789 9 Hickson, M. (2011) Probiotics in the prevention of antibiotic-associated diarrhoea and Clostridium difficile infection. Ther. Adv. Gastroenterol. 4, 185–197 10 Haiser, H.J. et al. (2013) Predicting and manipulating cardiac drug inactivation by the human gut bacterium Eggerthella lenta. Science 341, 295–298

164

Trends in Biotechnology March 2014, Vol. 32, No. 3

11 Haiser, H.J. and Turnbaugh, P.J. (2013) Developing a metagenomic view of xenobiotic metabolism. Pharmacol. Res. 69, 21–31 12 Qin, J. et al. (2010) A human gut microbial gene catalogue established by metagenomic sequencing. Nature 464, 59–65 13 Human Microbiome Project Consortium (2012) Structure, function and diversity of the healthy human microbiome. Nature 486, 207–214 14 Blaser, M. et al. (2013) The microbiome explored: recent insights and future challenges. Nat. Rev. Microbiol. 11, 213–217 15 Levy, R. and Borenstein, E. (2013) Metabolic modeling of species interaction in the human microbiome elucidates community-level assembly rules. Proc. Natl. Acad. Sci. U.S.A. 110, 12804–12809 16 Thiele, I. et al. (2013) A systems biology approach to studying the role of microbes in human health. Curr. Opin. Biotechnol. 24, 4–12 17 Kostic, A.D. et al. (2013) Exploring host-microbiota interactions in animal models and humans. Genes Dev. 27, 701–718 18 Segata, N. et al. (2013) Computational meta’omics for microbial community studies. Mol. Syst. Biol. 9, 666 19 Wikoff, W. and Anfora, A. (2009) Metabolomics analysis reveals large effects of gut microflora on mammalian blood metabolites. Proc. Natl. Acad. Sci. U.S.A. 106, 3698–3703 20 Van Duynhoven, J. et al. (2011)Metabolic fate ofpolyphenols in the human superorganism. Proc. Natl. Acad. Sci. U.S.A. 108 (Suppl), 4531–4538 21 Gill, S.R. et al. (2006) Metagenomic analysis of the human distal gut microbiome. Science 312, 1355–1359 22 Nicholson, J.K. et al. (2012) Host-gut microbiota metabolic interactions. Science 336, 1262–1267 23 Johnson, C.H. et al. (2012) Xenobiotic metabolomics: major impact on the metabolome. Annu. Rev. Pharmacol. Toxicol. 52, 37–56 24 Khersonsky, O. and Tawfik, D.S. (2010) Enzyme promiscuity: a mechanistic and evolutionary perspective. Annu. Rev. Biochem. 79, 471–505 25 Ekins, S. (2004) Predicting undesirable drug interactions with promiscuous proteins in silico. Drug Discov. Today 9, 276–285 26 Oguri, K. (1994) Regiochemistry of cytochrome P450 isozymes. Annu. Rev. Pharmacol. Toxicol. 34, 251–279 27 Hatzimanikatis, V. et al. (2005) Exploring the diversity of complex metabolic networks. Bioinformatics 21, 1603–1609 28 De Groot, M.J.L. et al. (2009) Metabolite and reaction inference based on enzyme specificities. Bioinformatics 25, 2975–2982 29 Gao, J. et al. (2011) The University of Minnesota Pathway Prediction System: multi-level prediction and visualization. Nucleic Acids Res. 39, W406–W411 30 Boyer, S. et al. (2007) Reaction site mapping of xenobiotic biotransformations. J. Chem. Inf. Model. 47, 583–590 31 Zomorrodi, A.R. and Maranas, C.D. (2012) OptCom: a multi-level optimization framework for the metabolic modeling and analysis of microbial communities. PLoS Comput. Biol. 8, e1002363 32 Klitgord, N. and Segre`, D. (2010) Environments that induce synthetic microbial ecosystems. PLoS Comput. Biol. 6, e1001002 33 Stolyar, S. et al. (2007) Metabolic modeling of a mutualistic microbial community. Mol. Syst. Biol. 3, 92 34 Kanehisa, M. et al. (2012) KEGG for integration and interpretation of large-scale molecular data sets. Nucleic Acids Res. 40, D109–D114 35 Jakoby, W.B. and Ziegler, D.M. (1990) The enzymes of detoxication. J. Biol. Chem. 265, 20715–20718 36 Holzhu¨tter, H-G. et al. (2012) The virtual liver: a multidisciplinary, multilevel challenge for systems biology. Wiley Interdiscip. Rev. Syst. Biol. Med. 4, 221–235 37 Valerio, L.G. and Long, A. (2010) The in silico prediction of humanspecific metabolites from hepatotoxic drugs. Curr. Drug Discov. Technol. 7, 170–187 38 Gao, J. et al. (2010) The University of Minnesota Biocatalysis/ Biodegradation Database: improving public access. Nucleic Acids Res. 38, D488–D491 39 Nobeli, I. et al. (2009) Protein promiscuity and its implications for biotechnology. Nat. Biotechnol. 27, 157–167 40 Williams, D.A. ed. (2007) Foye’s Principles of Medicinal Chemistry, Lippincott Williams & Wilkins 41 Nikolova, N. and Jaworska, J. (2003) Approaches to measure chemical similarity – a review. QSAR Comb. Sci. 22, 1006–1026 42 Nantasenamat, C. et al. (2010) Advances in computational methods to predict the biological activity of compounds. Expert Opin. Drug Discov. 5, 633–654

Review 43 Kirchmair, J. et al. (2012) Computational prediction of metabolism: sites, products, SAR, P450 enzyme dynamics, and mechanisms. J. Chem. Inf. Model. 52, 617–648 44 Marchant, C.A. et al. (2008) In silico tools for sharing data and knowledge on toxicity and metabolism: derek for windows, meteor, and vitic. Toxicol. Mech. Methods 18, 177–187 45 Darvas, F. (1987) MetabolExpert, an expert system for predicting metabolism of substances. In QSAR in Environmental Toxicology–II (Kaiser, K.L.E. and Riedel, D., eds), pp. 71–81, Springer 46 Medema, M.H. et al. (2012) Computational tools for the synthetic design of biochemical pathways. Nat. Rev. Microbiol. 10, 191–202 47 Moriya, Y. et al. (2010) PathPred: an enzyme-catalyzed metabolic pathway prediction server. Nucleic Acids Res. 38, W138–W143 48 Chen, L. et al. (2010) Predicting the network of substrate-enzymeproduct triads by combining compound similarity and functional domain composition. BMC Bioinformatics 11, 293 49 Kuhn, M. et al. (2008) Large-scale prediction of drug-target relationships. FEBS Lett. 582, 1283–1290 50 Mu, F. et al. (2011) Prediction of metabolic reactions based on atomic and molecular properties of small-molecule compounds. Bioinformatics 27, 1537–1545 51 Oh, M. et al. (2007) Systematic analysis of enzyme-catalyzed reaction patterns and prediction of microbial biodegradation pathways. J. Chem. Inf. Model. 47, 1702–1712 52 Schomburg, I. et al. (2013) BRENDA in 2013: integrated reactions, kinetic data, enzyme function data, improved disease classification: new options and contents in BRENDA. Nucleic Acids Res. 41, D764–D772 53 Muto, A. et al. (2013) Modular architecture of metabolic pathways revealed by conserved sequences of reactions. J. Chem. Inf. Comput. Sci. http://dx.doi.org/10.1021/ci3005379 54 Shu, Y.Z. et al. (1991) Metabolism of levamisole, an anti-colon cancer drug, by human intestinal bacteria. Xenobiotica 21, 737–750 55 Li, M. and Wang, B. (2008) Symbiotic gut microbes modulate human metabolic phenotypes. Proc. Natl. Acad. Sci. U.S.A. 105, 2117–2122 56 Yamada, T. et al. (2012) Prediction and identification of sequences coding for orphan enzymes using genomic and metagenomic neighbours. Mol. Syst. Biol. 8, 581 57 Henry, C.S. et al. (2010) High-throughput generation, optimization and analysis of genome-scale metabolic models. Nat. Biotechnol. 28, 977– 982 58 Patil, K.R. and Nielsen, J. (2005) Uncovering transcriptional regulation of metabolism by using metabolic network topology. Proc. Natl. Acad. Sci. U.S.A. 102, 2685–2689 59 Zomorrodi, A.R. et al. (2012) Mathematical optimization applications in metabolic networks. Metab. Eng. 14, 672–686 60 Oberhardt, M.A. et al. (2009) Applications of genome-scale metabolic reconstructions. Mol. Syst. Biol. 5, 320 61 Lewis, N.E. et al. (2012) Constraining the metabolic genotypephenotype relationship using a phylogeny of in silico methods. Nat. Rev. Microbiol. 10, 291–305 62 Finley, S.D. et al. (2010) In silico feasibility of novel biodegradation pathways for 1,2,4-trichlorobenzene. BMC Syst. Biol. 4, 7 63 Allison, K. et al. (2011) Metabolite-enabled eradication of bacterial persisters by aminoglycosides. Nature 473, 216–220 64 Maurice, C.F. et al. (2013) Xenobiotics shape the physiology and gene expression of the active human gut microbiome. Cell 152, 39–50 65 Booijink, C.C.G.M. et al. (2010) Metatranscriptome analysis of the human fecal microbiota reveals subject-specific expression profiles, with genes encoding proteins involved in carbohydrate metabolism being dominantly expressed. Appl. Environ. Microbiol. 76, 5533–5540

Trends in Biotechnology March 2014, Vol. 32, No. 3

66 Jacobs, D.M. et al. (2008) (1)H NMR metabolite profiling of feces as a tool to assess the impact of nutrition on the human microbiome. NMR Biomed. 21, 615–626 67 Segre`, D. et al. (2002) Analysis of optimality in natural and perturbed metabolic networks. Proc. Natl. Acad. Sci. U.S.A. 99, 15112–15117 68 Brochado, A.R. et al. (2012) Impact of stoichiometry representation on simulation of genotype-phenotype relationships in metabolic networks. PLoS Comput. Biol. 8, e1002758 69 Modi, S.R. et al. (2013) Antibiotic treatment expands the resistance reservoir and ecological network of the phage metagenome. Nature 499, 219–222 70 Lepage, P. et al. (2013) A metagenomic insight into our gut’s microbiome. Gut 62, 146–158 71 Rey, F.E. et al. (2013) Metabolic niche of a prominent sulfate-reducing human gut bacterium. Proc. Natl. Acad. Sci. U.S.A. 110, 13582–13587 72 Dunne, C. (2001) Adaptation of bacteria to the intestinal niche: probiotics and gut disorder. Inflamm. Bowel Dis. 7, 136–145 73 Yang, H. et al. (2005) Niche heterogeneity determines bacterial community structure in the termite gut (Reticulitermes santonensis). Environ. Microbiol. 7, 916–932 74 Hao, W-L. and Lee, Y-K. (2004) Microflora of the gastrointestinal tract: a review. Methods Mol. Biol. 268, 491–502 75 Hooper, L. et al. (2012) Interactions between the microbiota and the immune system. Science 336, 1268–1273 76 Costello, E. and Stagaman, K. (2012) The application of ecological theory toward an understanding of the human microbiome. Science 336, 1255–1262 77 Abubucker, S. et al. (2012) Metabolic reconstruction for metagenomic data and its application to the human microbiome. PLoS Comput. Biol. 8, e1002358 78 Kolmeder, C.A. et al. (2012) Comparative metaproteomics and diversity analysis of human intestinal microbiota testifies for its temporal stability and expression of core functions. PLoS ONE 7, e29913 79 Lee, H. et al. (2010) Bacterial charity work leads to population-wide resistance. Nature 467, 82–85 80 Riley, M. and Wertz, J. (2002) Bacteriocins: evolution, ecology, and application. Annu. Rev. Microbiol. 56, 117–137 81 Janga, S.C. and Babu, M.M. (2008) Network-based approaches for linking metabolism with environment. Genome Biol. 9, 239 82 Borenstein, E. et al. (2008) Large-scale reconstruction and phylogenetic analysis of metabolic environments. Proc. Natl. Acad. Sci. U.S.A. 105, 14482–14487 83 Taffs, R. et al. (2009) In silico approaches to study mass and energy flows in microbial consortia: a syntrophic case study. BMC Syst. Biol. 3, 114 84 Freilich, S. et al. (2011) Competitive and cooperative metabolic interactions in bacterial communities. Nat. Commun. 2, 589 85 Rodrigo, G. et al. (2008) DESHARKY: automatic design of metabolic pathways for optimal cell growth. Bioinformatics 24, 2554–2556 86 Rocha, I. et al. (2010) OptFlux: an open-source software platform for in silico metabolic engineering. BMC Syst. Biol. 4, 45 87 Chhabra, R.S. (1979) Intestinal absorption and metabolism of xenobiotics. Environ. Health Perspect. 33, 61–69 88 Gad, S.C. ed. (2007) Toxicology of the Gastrointestinal Tract, CRC Press 89 Kaminsky, L.S. and Zhang, Q-Y. (2003) The small intestine as a xenobiotic-metabolizing organ. Drug Metab. Dispos. 31, 1520–1525 90 Bjo¨rkholm, B. et al. (2009) Intestinal microbiota regulate xenobiotic metabolism in the liver. PLoS ONE 4, e6958

165