Published May 1, 2015

RUMINANT NUTRITION SYMPOSIUM: Use of genomics and transcriptomics to identify strategies to lower ruminal methanogenesis1,2,3 T. A. McAllister,*4 S. J. Meale,* E. Valle,* L. L. Guan,† M. Zhou,† W. J. Kelly,‡ G. Henderson,‡ G. T. Attwood,‡ and P. H. Janssen‡ *Agriculture and Agri-Food Canada, Lethbridge Research Centre, Lethbridge, AB T1J 4B1, Canada ; †Department of Agricultural, Food and Nutritional Science, University of Alberta, Edmonton, AB, T6G 2P5 Canada; and ‡Grasslands Research Centre, AgResearch Ltd., Private Bag 11008, Palmerston North 4442, New Zealand

ABSTRACT: Globally, methane (CH4) emissions account for 40% to 45% of greenhouse gas emissions from ruminant livestock, with over 90% of these emissions arising from enteric fermentation. Reduction of carbon dioxide to CH4 is critical for efficient ruminal fermentation because it prevents the accumulation of reducing equivalents in the rumen. Methanogens exist in a symbiotic relationship with rumen protozoa and fungi and within biofilms associated with feed and the rumen wall. Genomics and transcriptomics are playing an increasingly important role in defining the ecology of ruminal methanogenesis and identifying avenues for its mitigation. Metagenomic approaches have provided information on changes in abundances as well as the species composition of the methanogen community among ruminants that vary naturally in their CH4 emissions, their feed efficiency, and their response to CH4 mitigators. Sequencing the genomes of rumen methanogens has provided insight into surface proteins that may prove useful in the development of vaccines and has

allowed assembly of biochemical pathways for use in chemogenomic approaches to lowering ruminal CH4 emissions. Metagenomics and metatranscriptomic analysis of entire rumen microbial communities are providing new perspectives on how methanogens interact with other members of this ecosystem and how these relationships may be altered to reduce methanogenesis. Identification of community members that produce antimethanogen agents that either inhibit or kill methanogens could lead to the identification of new mitigation approaches. Discovery of a lytic archaeophage that specifically lyses methanogens is 1 such example. Efforts in using genomic data to alter methanogenesis have been hampered by a lack of sequence information that is specific to the microbial community of the rumen. Programs such as Hungate1000 and the Global Rumen Census are increasing the breadth and depth of our understanding of global ruminal microbial communities, steps that are key to using these tools to further define the science of ruminal methanogenesis.

Key words: archaea, chemogenomic, global rumen census, Hungate1000, interspecies H2 transfer © 2015 American Society of Animal Science. All rights reserved. J. Anim. Sci. 2015.93:1431–1449 doi:10.2527/jas2014-8329 1Based on a presentation at the Ruminant Nutrition Symposium

titled “The Rumen Microbiome and Nutritional Health and Production” at the Joint Annual Meeting, July 20–24, 2014, Indianapolis, IN. 2The Global Rumen Census and Hungate1000 projects are funded by the New Zealand government in support of the Livestock Research Group of the Global Research Alliance on Agricultural Greenhouse Gases (http://www.globalresearchalliance.org) to support international efforts to develop CH4 mitigation and rumen adaptation technologies. The Global Rumen Census is a collaborative project that relies on the participation of researchers from 70 organizations in 35 countries. The sequencing effort for the Hungate1000 is supported by the U.S. Department of Energy Joint Genome Institute Community Sequencing Program (CSP 612). The CSP proposal

involved 23 international research groups from 14 countries, and the overall project is a collaboration between members of the Rumen Microbial Genomics Network. The AgResearch authors are also funded under contract by the Pastoral Greenhouse Gas Research Consortium (PGgRc) and the New Zealand Agricultural Greenhouse Gas Research Centre (NZAGRC). The Canadian authors would also like to recognize support for work in the area under the NorwegianCanadian BILAT program and the Alberta Livestock and Meat Agency for funding of their research program. 3All authors contributed equally to this review. 4Corresponding author: [email protected] Received July 28, 2014. Accepted November 1, 2014.

1431

1432

McAllister et al.

INTRODUCTION Methane (CH4) production is an inescapable consequence of normal ruminal fermentation. It provides an essential means of H2 removal, as H2 in the rumen can inhibit hydrogenase activity and limit the oxidation of sugar if alternative means of H2 disposal are unavailable (McAllister and Newbold, 2008). Although CH4 production appears essential for efficient ruminal digestion, emissions also present an environmental concern because CH4 is a potent greenhouse gas with a global warming potential (over 100 yr) 28 times that of CO2 (Intergovernmental Panel on Climate Change, 2013). In light of the projected increase in demand for meat and milk, CH4 abatement strategies that reduce emissions are needed to curb the predicted rise in emissions associated with greater ruminant production. Consequently, significant effort has been directed toward improving our understanding of ruminal CH4 formation and the identification of strategies that reduce methanogenesis. Advances in our understanding of rumen microbial communities through genomics (Leahy et al., 2010; Attwood et al., 2011; Leahy et al., 2013a), metagenomics (Brulc et al., 2009; Hess et al., 2011; Morgavi et al., 2013), and metatranscriptomics (Qi et al., 2011; Shi et al., 2014) have revealed insights into species composition and natural variations in abundances, as well as interactions among members of this complex ecosystem. Several reviews on CH4 abatement have been published recently (Beauchemin et al., 2008; McAllister and Newbold, 2008; Hook et al., 2010; Martin et al., 2010; Meale et al., 2012; Kumar et al., 2014). Here, we focus on the roles that genomics, transcriptomics, and metatechniques may play in identifying avenues to mitigate ruminal methanogenesis. Ongoing sequencing projects, such as the Hungate1000 and Global Rumen Census, are providing the foundational knowledge that will be required to formulate viable mitigation strategies. METHANOGENS Methanogenic archaea (methanogens) lack the peptidoglycan found in the cell walls of bacteria, but rather contain pseudomurein, heteropolysaccharide, or protein in their cell walls (Hobson and Stewart, 1997). Methanogens also possess distinctive cofactors, including coenzyme F420, which displays blue-green autofluorescence at 470 nm (Ashby et al., 2001), and coenzyme M, which is methylated to CH4 (Hobson and Stewart, 1997; Hook et al., 2010). It has been predicted that there may be as many as 120 species of methanogens representing 33 genera (Wright and Klieve, 2011), but only a fraction of these species have been identified

in the rumen. The majority of rumen methanogens belong to groups closely related to Methanobrevibacter gottschalkii and Methanobrevibacter ruminantium, a few species of Methanosphaera, Methanomicrobium mobile, and a number of as-yet undefined and unnamed species of the class Methanomassiliicoccales (Janssen and Kirs, 2008). Other species have been isolated as pure cultures (Table 1) and are detected at low abundance (Janssen and Kirs, 2008; Kim et al., 2011). On the basis of culturing techniques, only 2, M. ruminantium (Smith and Hungate, 1958) and Methanosarcina spp., have been found in the rumen at populations greater than 106/mL (Lovley et al., 1984). However, these data probably misrepresent the true abundances of different rumen methanogens because cultivation-based methods often underestimate microbial population sizes. Difficulties in culturing rumen methanogens are attributed to their fastidious anaerobic nature with requirements of a redox potential below −300 mV (Stewart and Bryant 1988) and optimal growth conditions within a pH of 6.0 to 8.0 (McAllister et al., 1996; Kumar et al., 2009). METHANOGENS AND THEIR INTERACTIONS Methanogens exist within a number of niches in the rumen ecosystem, including association with the rumen epithelium (Janssen and Kirs, 2008; Pei et al., 2010), integration into biofilms (Fig. 1A), and acting as ecto- and endosymbionts of rumen protozoa (Fig. 1B). They are also closely associated with rumen fungi (Fig. 1C), from which are thought to derive reducing equivalents from organelles known as hydrogenosomes. Biofilm Following feed ingestion, a consortium of microbes forms a structured biofilm on the surface of feed particles, producing the myriad of enzymes required for feed digestion. Methanogens within the biofilm use H2 produced by other microorganisms, resulting in the terminal reduction of CO2 to CH4 (Qi et al., 2010). This process, known as interspecies H2 transfer, reduces the partial pressure of H2 within the biofilm (Wolin, 1979; McAllister and Newbold, 2008; Janssen, 2010; Leng, 2014), thus enabling the recycling and oxidation of electron carriers such as NADH (Hungate et al., 1970; Joblin et al., 1989; McAllister et al., 1996). Cross-feeding reactions among community members alter the end products of digestion, which, in turn, are absorbed by the host, thereby preventing their accumulation from inhibiting fermentation (Qi et al., 2010). For example, coculture of Ruminococcus albus or Ruminococcus flavefaciens with M. ruminantium

1433

Genomic strategies to lower methanogenesis

Table 1. Methanogens isolated from the rumen Organism Methanosarcina sp. M. barkeri M. mazei Methanobacterium formicicum Methanobacterium bryantii Methanomicrobium mobile Methanobrevibacter sp. M. ruminantium

Morphology1

Substrate

Reference

Sequence status

Irregular cocci large clusters, immobile, HPS + PR Cocci, immobile, HPS

H2, methanol methylamines, acetate Methanol methylamines, acetate H2, formate

Beijer (1952)

Complete (Strain CM1; William et al., 2014)

Mah (1980)

In progress (Strain S-6)

Oppermann et al. (1957) Joblin (2005) Paynter and Hungate (1968) Lovley et al. (1984) Smith and Hungate (1958)

Complete (Strain BRM9)

Long rods and filaments, immobile, PS PS Short, curved rods, motile, PR Short rods, synthesizes CoM, PS Short rods, requires CoM,3 PS, variably motility

H2, formate H2,formate, acetate H2, formate H2, formate

M. wolinii M. gottschalkii M. boviskoreani

Lee et al., 2013a

M. millerae M. olleyae

Rea et al. (2007) Rea et al. (2007)

Methanoculleus olentangyi2

H2, formate, acetate

Draft (Strain YE299) Draft genome available (DSM1539) Complete (Strain M1; Leahy et al., 2010) Draft genome available (DSM11976) Closed (Strain SM9) Draft (Strain JH1; Lee et al., 2013b); Complete (Strain AbM4; Leahy et al., 2013a) Draft genome available (Strain DSM16643) Draft genome available (Strain DSM16632) Draft (Strain YLM1, Strain YE286)

Joblin (2005)

1Abbreviations:

CoM = coenzyme M; PS = pseudomurein; HPS = heteropolysaccharide; PR = protein. from cervid rumen. 3In strain M1 only. 2Cultured

shifted the end products of digestion from acetate, H2, CO2, and ethanol (R. albus) and acetate, succinate, formate, and H2 (R. flavefaciens; Miller and Wolin, 1973; Latham and Wolin, 1977; McAllister et al., 1996) to mainly acetate and CH4, an outcome that enhances the yield of ATP to both the cellulolytic bacterium and the methanogen (Wolin and Miller, 1988; McAllister et al., 1996). Similarly, coculturing Methanobrevibacter smithii and R. flavefaciens dramatically increased the rate and extent of cellulose digestion compared with R. flavefaciens monocultures (Beaudette, 1994; Williams et al., 1994). Epifluorescence microscopy showed colonies of M. smithii dispersed across the entirety of the

solid cellulose substrate and situated proximally to R. flavefaciens colonies, an association that presumably facilitates interspecies H2 transfer (Beaudette, 1994). Protozoa Methanogens form symbiotic associations with H2-producing protozoa by floc formation (Conrad et al., 1985; Thiele et al., 1988; Lange et al., 2005), by adherence to the pellicles (Vogels et al., 1980; Stumm et al., 1982), and as endosymbionts within ciliates (Finlay et al., 1994). This association appears to occur preferentially with ciliate protozoa because methanogens

Figure 1. Methanogen interactions documented in different rumen microenvironments using confocal laser scanning microscopy (CLSM) and scanning electron microscopy. Characteristic blue-green autofluorescence of coenzyme F420 identified (A) methanogens associated with feed/particles in biofilm and (B) endosymbiotic methanogens associated within an Ophryoscolex sp., using CLSM. (C) Scanning electron microscopy image of methanogens (a) attached to fungal rhizoids (b). The physical proximity between hydrogenosomes and methanogens is considered to facilitate H2 transfer. Adapted from Jin et al. (2011).

1434

McAllister et al.

can be seen at up to 104 per protozoa only 1 to 2 h after ingestion of feed particles (Vogels et al., 1980; Leng, 2014). A study by Vogels et al. (1980) reported an ectosymbiotic relationship between 11 species of rumen ciliates (Ophryoscolecidae) and methanogens, where epifluorescence microscopy revealed that almost all prokaryotes attached to the protozoan cell surface were methanogens. The cells occurred as clusters or long chains of rods, whereas single methanogens were abundant in the rumen fluid and chains were seldom found free floating or attached to substrates other than ciliates. A similar pattern of attachment was observed in the association of methanogens with Eudiplodinium magii and Diplodinium dentatum (Stumm et al., 1982). Vogels et al. (1980) further observed that although all of the species examined exhibited associations with methanogens, the degree of association varied according to species. In excess of 50% of Ostracodinium obtusum and Eudiplodinium magii cells had methanogens attached, but only 8% of Entodinium longinucleatum had attached methanogens. The authors suggested that the physical structure of the pellicles, in particular the body striations, plays a role in the level of attachment, as the species with the lowest frequency of attachment (E. longinucleatum and Entodinium simplex) had the finest striations, whereas species exhibiting a high frequency of association (Entodinium caudatum and Polyplastron multivesiculatum) exhibited a coarse surface structure. Similarly, Tokura et al. (1997) suggested that cells with a larger volume and greater metabolic activity may be more heavily colonized by methanogens. Such variation could also be attributable to dietary differences. For example, Stumm et al. (1982) reported a reduction in the number of ciliates associated with methanogens (65% to 25%) following a period of fasting in cattle grazing meadow grass. The authors observed a similar reduction with grain feeding, although the decrease in association was more rapid. As a result of the changing physiochemical conditions in the rumen, predominantly the partial pressure of H2, the number of methanogens associated with the protozoan cell surface varies throughout the day (Stumm et al., 1982). Ruminal methanogens and ciliate protozoa also exist in endosymbiotic relationships. Finlay et al. (1994) reported that the majority of endosymbionts within the cytoplasm of Dasytricha ruminantium and Entodinium spp. in the rumen of sheep displayed autofluorescence at F420 and were, in fact, more numerous than those attached to the external cell surface of ciliates. Not all interactions are physical, as association of Methanosarcina barkeri with Isotricha spp. reduced the production of alternative H2 sinks and increased CH4 formation by M. barkeri without a physical inter-

action. This indicates that metabolic exchange of hydrogen can occur without protozoa and methanogens being intimately associated (Hillman et al., 1988). Similarly, there are several species of methanogens (Methanomicrobiales) that have not been shown to associate with ruminal protozoa (Sharp et al., 1998). Despite being a dominant group in the rumen, Methanobrevibacter spp. failed to establish interspecies H2 transfer with P. multivesiculatum (Ushida et al., 1995), perhaps indicating that ciliate-associated ruminal methanogens possess special characteristics that enable such interactions to occur (Tokura et al., 1997). Fungi Rumen methanogens have been shown to associate closely with the hyphae of fungi (Bauchop and Mountfort, 1981), forming stable cocultures that shift the flow of electrons generated in glycolysis away from the formation of lactate and succinate to CH4 (Bauchop and Mountfort, 1981). As with rumen protozoa, rumen fungi produce H2 in membrane-bound organelles called hydrogenosomes (Yarlett et al., 1984, 1986), and the physical proximity of these organelles to methanogens is thought to facilitate H2 transfer (Krumholz et al., 1983; Wolin and Miller, 1988; McAllister et al., 1996). This is a general phenomenon in cocultures of H2 or H2 plus formate-consuming methanogens and several genera of anaerobic fungi (Marvin-Sikkema et al., 1990). Coculturing M. ruminantium or M. smithii with Neocallimastix spp. (Bauchop and Mountfort, 1981) or M. smithii with Neocallimastix frontalis (Joblin et al., 1990) increased acetate and CH4 production and reduced ethanol, formate, lactate, and H2 formation, indicative of interspecies H2 transfer (Bauchop and Mountfort, 1981; Joblin et al., 1990). Similarly, coculturing N. frontalis or Piromyces communis with M. smithii on barley straw decreased ethanol, lactate, and H2 and increased CH4 formation (Joblin et al., 1989). Fermentation of cellulose by N. frontalis, Neocallimastix patriciarum, or P. communis alone resulted in the formation of H2, CO2, formate, acetate, lactate, succinate, and ethanol (Marvin-Sikkema et al., 1990). However, when anaerobic fungi were incubated in coculture with Methanobrevibacter arboriphilus or Methanobacterium bryantii, CH4, CO2, formate, and increased amounts of acetate were the only end products of cellulose digestion. This is comparable to the effect of methanogens on H2-producing bacteria. When M. smithii was used in cocultures, a similar shift was observed with a rise in the production of CO2 and complete consumption of formate with no detection of H2 either during or after growth.

Genomic strategies to lower methanogenesis

1435

Figure 2. Development of sequencing technologies and their use in identifying novel CH4 mitigation strategies. Early sequencing technologies and those developed more recently are colored grey and white, respectively. aSequencing of single genes facilitates the identification of enzymes of industrial importance (e.g., cellulases) and can lead to improvements in feed efficiency. bGlobal Rumen Census will increase our understanding of the microbial diversity, resulting in the application of broad spectrum CH4 mitigation strategies. cHungate1000 aims to strengthen current databases. dFunctional “omics” aims to improve the linkage between genomics, transcriptomics and proteomics as posttranslational processing results in a small subset of proteins being transcribed as outlined by the transcriptome.

In a study by Bauchop and Mountfort (1981), a significant increase in cellulose degradation was observed, such that 82% of initial cellulose was degraded after incubation in coculture, compared with 53% in monoculture. Fungi of the genus Neocallimastix have a greater cellulose-digesting capacity than many of the cellulolytic bacteria (Fibrobacter succinogenes and R. flavefaciens). However, in coculture with R. flavefaciens, the cellulose-digesting capacity of Neocallimastix appeared to be reduced, indicating antagonistic bacterial-fungal interactions also occur in the rumen (Qi et al., 2010). As acetate formation is coupled to ATP synthesis in many anaerobes, greater production of acetate in the presence of methanogens could have significant effects on energy metabolism and growth of fungi (Marvin-Sikkema et al., 1990). In cocultures of N. frontalis and methanogens, a greater rate and degree of cellulose fermentation and greater cellulolytic activity were observed (Marvin-Sikkema et al., 1990) when it was grown in the presence of M. bryantii or M. smithii, resulting in a 15% to 25% increase in the rate of cellulose degradation. RUMEN MICROBIAL GENOME SEQUENCING Molecular studies on the structure and function of the rumen microbiome have moved from the laborious sequencing of single genes extracted from genomic

DNA libraries of rumen microorganisms to the almost routine sequencing of entire genomes of individual organisms using next-generation, high-throughput sequencing technologies. Furthermore, these massively parallel sequencing technologies have recently made possible the deep sequencing of the rumen metagenome (Bult et al., 1996; Brulc et al., 2009; Hess et al., 2011) and metatranscriptome (Qi et al., 2011; Shi et al., 2014) and in the future promise to allow retrieval of all microbial sequences from this complex and dense ecosystem (Fig. 2). The first rumen bacterial genome sequenced was that of F. succinogenes S85, which revealed an abundance of genes involved in plant cell wall breakdown, including 33 cellulases, 24 xylanases, and 14 carbohydrate esterases (Jun et al., 2007). Since then, genome sequencing of individual rumen bacteria has accelerated steadily and has focused mainly on understanding the genomes of rumen fiber-degrading organisms for applications in animal nutrition and feedstock depolymerization for biofuel production or on rumen methanogens for mitigating CH4 emissions. The recent commencement of the Hungate1000 project (described subsequently) has accelerated the genome sequencing effort targeting rumen microbes. Currently, 204 rumen microorganisms have completed or produced draft genome sequences from the study. These genome sequences will underpin future functional “omics” proj-

1436

McAllister et al.

ects that initially aim to define the specific function of each gene within these organisms but ultimately will lead to an understanding of each organism’s contribution to rumen function. Genomics of Rumen Methanogens Currently known methanogens are members of 7 orders of the phylum Euryarchaeota in the domain Archaea. Although there is considerable species diversity among all methanogens, the range of rumen methanogens appears to be quite small (Janssen and Kirs, 2008). Methanogens are the only significant group of ruminal archaea and are united by a common underlying biochemistry, not only in their methanogenesis pathway but also in regard to other aspects of their biochemistry (Hedderich and Whitman, 2013). They are also evolutionarily distinct from the other members of the rumen ecosystem, which is made up of members of the domains Eukarya (host animal, ingested plant feed, protozoa, and fungi) and Bacteria. This makes the development of methanogen-specific inhibitors especially attractive. A small molecule that targets an enzyme unique to methanogens or that targets an archaeal enzyme that displays considerable divergence from its eukaryotic and bacterial homologs could potentially selectively inhibit ruminal methanogenesis. To devise approaches to mitigating enteric methanogenesis, rumen methanogens must be fully characterized, with genomic sequencing playing an important role in this process. Methanogens are difficult organisms to work with, and almost every aspect of genome sequencing presents a challenge. Rumen methanogens grow slowly, are extremely sensitive to O2, and have unusual growth requirements, characteristics that usually result in low growth densities and, as a result, low yields of nucleic acids. They are also difficult to preserve, and revival from stored stocks can be problematic. Consequently, few stable cultures of rumen methanogens are available, making the number of candidate strains for genome sequencing projects limited. Methanogens also have cell walls that are insensitive to the enzymes commonly used to extract DNA from bacteria such as lysozyme and mutanolysin. Consequently, cells must be lysed by physical disruption, a process that shears genomic DNA into small fragments. Smaller fragments make subsequent library construction more difficult and often preclude the construction of the large insert libraries required for scaffolding the genome. Furthermore, the molar percentage of guanine + cytosine (G+C) content of DNA from members of the family Methanobacteriaceae is typically low, in the range of 29% to 35% G+C, making sequencing more prone to early termination, resulting in poor se-

quence length and making genome assemblage more challenging. However, despite these problems, several genome sequencing projects on rumen methanogens have been completed or are underway. These projects are contributing new methanogen gene sequences to an expanding database, which is widening the possible methanogen-specific gene targets for testing, and improving our understanding of the diversity and metabolic capacity of methanogenic archaea in the rumen. Methanobrevibacter ruminantium M1 was chosen as the first rumen methanogen for genome sequencing mainly for pragmatic reasons: a stable culture was available from DSMZ (Deutsche Sammlung von Mikroorganismen und Zellkulturen; German Collection of Microorganisms and Cell Cultures, Braunschweig, Germany), Methanobrevibacter spp. are the most commonly isolated methanogens from the rumen, and the most abundant archaeal 16S rRNA gene sequences retrieved were affiliated with Methanobrevibacter species. Given the difficulties indicated previously, completing the M1 genome sequence was a significant achievement and serves as a platform on which subsequent methanogen genome sequencing projects have built. The genome sequence confirmed several previous observations related to this organism and uncovered several significant new features (Leahy et al., 2010). It was confirmed that M1 occupied a hydrogenotrophic lifestyle in the rumen, encoding all of the genes needed for the reduction of CO2 to CH4 using H2. However, it was found that methanogenesis in M1 was facilitated by only 1 form of the coenzyme M (CoM) reductase enzyme, encoded by the Methyl Coenzyme M Reductase (mcr) operon (mcrBCDGA), unlike in other hydrogenotrophic methanogens that also encoded the Mcr II isozyme, encoded by the mrt operon (mrtBCDA). In Methanothermobacter thermoautotrophicus, mrt is differentially expressed according to H2 partial pressure (Reeve et al., 1997; Luo et al., 2002) and is upregulated during growth at high H2 concentrations. The rumen usually has low H2 partial pressures; thus, M1 appears to be adapted for these conditions using only the mcr system. Accessory features of the genome indicated that M1 was also capable of growth on a wider range of energy sources in addition to H2. The presence of formate utilization genes (fdhAB) in the M1 genome indicated that formate can also serve as an energy source for CH4 formation. In coculture experiments, it was found that these M1 genes were upregulated during growth with a xylan-degrading rumen bacterium, Butyrivibrio proteoclasticus, which indicated that formate was an important methanogenic substrate for this syntrophic interaction (Leahy et al., 2010). Two genes encoding NADPH-dependent F420 dehydrogenase (npdG1, 2) and 3 genes encod-

Genomic strategies to lower methanogenesis

ing NADP-dependent alcohol dehydrogenases (adh1, 2, and 3) were also unexpectedly discovered. These genes are typically only found in methanogens using short-chain alcohols, and subsequent tests showed that M1 growth on H2 was stimulated by the addition of methanol or ethanol but that short-chain alcohols alone could not support the growth of M1. The previously observed growth requirements of M1 for acetate, 2-methylbutyrate and CoM were also explained from analysis of the genome sequence. Acetate is needed for cell carbon biosynthesis via activation to acetyl CoA (acs, acsA) and reductive carboxylation to pyruvate (porABCDEF); 2-methylbutyrate is used for making isoleucine, as homoserine kinase is missing from the usual biosynthetic pathway from threonine, and coenzyme M addition is essential as several genes involved in its biosynthesis (comADE) are absent. Combined, these discoveries from a single rumen methanogen genome, begin to reveal the subtleties of methanogen life in the rumen, and add a degree of complexity when considering which genes to target for inhibiting CH4 emissions. Although both methyl-CoM reductase II and CoM biosynthetic genes are highly conserved and specific to methanogens, they are not necessarily good choices for inhibiting rumen methanogens given their absence in the M1 genome. Methanogen Genomics and CH4 Mitigation As indicated previously, the rationale for sequencing methanogen genomes is to identify essential, methanogen-specific features that can be targeted for CH4 mitigation. As gene functions are mediated by the activities of their encoded proteins, converting such sets of conserved, methanogen-specific, and functionally essential gene targets into practical, effective technologies requires interventions at either the transcription or the protein level. There are 2 main avenues that are being followed in this regard, a chemogenomics approach in which the protein products of the target genes are expressed and used in enzyme assays to screen chemical libraries to identify compounds with inhibitory effects against target proteins and a reverse vaccinology approach in which extracellular proteins are identified and screened for antigenicity in a suitable host. Either of these approaches may also open avenues to developing additives that inhibit the transcription of functional proteins in methanogens. In addition to these 2 approaches, there are other ways in which methanogen genomes can and are contributing to CH4 mitigation research. By virtue of uncovering the entire gene set encoded by a particular organism, the genome sequence allows discovery of novel genes with new functions (Fig. 3). For example, the M1

1437

Figure 3. Stages in the development of CH4 mitigation strategies arising from sequencing methanogen genomes.

genome revealed, for the first time in archaea, genes encoding nonribosomal peptide synthetases (NRPS; Leahy et al., 2010). The NRPS are well known in other microorganisms for their involvement in the synthesis of small molecule natural products that have found a variety of pharmaceutical uses as antibiotics, iron-chelating siderophores, immunosuppressants, or antitumor drugs. The M1 genome encodes 2 NRPS genes (mru0068, mru0351), which contain 2 and 4 peptide synthesis modules, respectively. The genes surrounding these NRPS genes are putatively involved in environmental sensing, NRPS gene regulation, and export of the synthesized peptide(s). The mru0068 protein shows strong similarity to a NRPS from Syntrophomonas wolfei, a fatty acid–oxidizing bacterium known to participate in interspecies H2 transfer with methanogens (McInerney et al., 1979). In S. wolfei, the NRPS peptide product is posited to act as a signaling molecule to communicate with its methanogenic partners during the initiation of syntrophy (Sieber et al., 2010). It is tempting to speculate that the M1 NRPS peptide products have a similar function in the rumen. If nonribosomally synthesized peptides are important in establishing syntrophic interactions in the rumen, then developing interventions against these peptides may offer a new approach to interrupting interspecies H2 transfer between H2-producing bacteria and methanogens. Prevention or reduction of the efficiency of these syntrophic interactions could have profound effects on the supply of methanogenic substrates to methanogens and therefore on ruminal CH4 production. Inhibitors Halogenated compounds were shown a number of years ago to reduce CH4 formation when dosed to animals (Johnson et al., 1972; Clapperton, 1974;

1438

McAllister et al.

Czerkawski and Breckenridge, 1975; McCrabb et al., 1997) but are considered unacceptable for use on farm because of their environmental or toxicological impacts. A range of other inhibitors have been identified using traditional in vitro screening techniques, but their effects are generally inconsistent or short-term (Soliva et al., 2011). Targeted discovery of novel small molecule inhibitors can be achieved by in silico modeling of genomically derived enzyme structures and subsequent in silico screening of large libraries of chemical compounds or by in vitro screening of chemical compound libraries using enzyme assays or directly against methanogens. The former offers the possibility of screening literally millions of compounds by testing their potential to interact with critical catalytic or regulatory sites in the enzyme using a computer-based model of the enzyme. The development of targets for the in silico approach relies on bioinformatic identification of conserved target enzymes across multiple methanogen genomes (Carbone et al., 2013; Leahy et al., 2013b) and uses gene sequence data for direct cloning, expression, and purification of the target enzymes. Although enzyme structural data, derived from X-ray crystallographic structures, are required for effective in silico modeling, gene sequence data (Leahy et al., 2010) can be applied to existing structures to develop models that can also be used to explore inhibitor docking (McMillan et al., 2011). Lead compounds can be further explored by analyzing the docking ability of structural homologs and substituted derivatives. Studies by Soliva et al. (2011), Martínez-Fernández et al. (2014), and Reynolds et al. (2014) describe the discovery of several compounds that are potential inhibitors of methyl-CoM reductase. These compounds show structural similarities to CoM, the cosubstrate of this enzyme, and to its analog bromoethanesulfonate, a well-known inhibitor of methanogens (Gunsalus et al., 1978). In silico screening against the methyl-CoM reductase structure identified new compounds that may inhibit methanogens (Martínez-Fernández et al., 2014; Reynolds et al., 2014). Testing revealed that 1 of the newly identified compounds, 3-nitrooxypropanol, inhibited CH4 formation by rumen fluid in vitro (Martínez-Fernández et al., 2014) and in vivo (Haisan et al., 2014; Martínez-Fernández et al., 2014; Reynolds et al., 2014), clearly demonstrating the utility of this approach. Vaccines Vaccination represents a very cost-effective and acceptable approach to control methanogenesis in the rumen and is 1 of the few strategies easily applied to free-ranging ruminants. Methanogen cells and their components induce specific antibodies in ruminants,

both in serum and in saliva (Williams et al., 2008, 2009; Wedlock et al., 2010, 2013). Ruminants produce copious amounts of saliva, up to 3 rumen volumes per day (Kay, 1960; Bailey 1961), which contains antibodies that are continuously supplied to the rumen and have the potential to interact with ruminal microbes. Interaction of the antibodies with methanogens might cause cell aggregation, which may decrease the ability of individual cells to access H2 for CH4 formation, or might result in steric interference with surface functions like transporters or cell components that are responsible for the syntrophic interactions described above. The intention of vaccination is to produce antibodies that interfere with cellular processes that allow growth and survival of methanogens in the rumen. The growth rate of ruminal methanogens is estimated to be about half maximal (Janssen, 2010) but is sufficient to allow them to maintain a population by growing fast enough to compensate for population losses as a result of passage of solid and liquid rumen contents into the lower intestinal tract. In theory, interference that results in more than a 50% reduction in the growth rate may result in washout of methanogens from the rumen, making a killing action unnecessary to achieve some measure of a reduction in methanogenesis. Traditional approaches to vaccine development have already been used to formulate vaccines against rumen-dwelling microbes. Vaccines have been developed against bacteria implicated in ruminal acidosis, namely, Streptococcus bovis and Lactobacillus spp. Such a vaccine was reported to protect sheep and cattle against ruminal acidosis, resulting in higher ruminal pH values, less mortality, and lower ruminal lactate concentrations (Shu et al., 1999, 2000; Gill et al., 2000). These studies demonstrated proof of concept that salivary antibodies can be targeted and control the deleterious effects of ruminal microbes. Others have used similar approaches to develop vaccines against ruminal protozoa (Williams et al., 2008) and methanogens (Wright et al., 2004), but these preparations had minimal effects on targeted microbial populations or enteric CH4 emissions. Mining of genome sequence data is an approach to identifying specific surface proteins as antigens that could dramatically improve vaccine efficacy (Leahy et al., 2010). By comparison across multiple genomes (Leahy et al., 2013b), common antigens or conserved domains can be identified to develop vaccines with a broad enough specificity to target multiple species of ruminal methanogens. Experiments using avian antibodies indicated that some strains of methanogens can induce an antibody response that has a broad effect against the total mixed methanogen population in rumen contents in vitro, presumably across many different species (Cook et al., 2008).

Genomic strategies to lower methanogenesis

Because methanogens are very difficult to grow on a large scale and because multiple methanogen species will have to be targeted in a vaccine, it is likely that a vaccine formulation containing only conserved proteins or peptides would be far more effective than a preparation consisting of crude whole cells. Methanogens in the rumen fall into 2 broad physiological types (Janssen and Kirs, 2008). The first are H2- and formate-using methanogens, which can be termed hydrogenotrophs and belong largely to the genus Methanobrevibacter. Members of this genus are well adapted to gastrointestinal environments and appear to outcompete other hydrogenotrophs such as Methanobacterium and Methanomicrobium, which are found far less frequently in the rumen. The second group requires H2 plus methyl groups for growth and can be termed methylotrophs. These are predominately members of the genus Methanosphaera and a range of as yet poorly defined genera and species of the order Methanomassiliicoccales (Iino et al., 2013), also known as rumen cluster C (Janssen and Kirs, 2008). Methylotrophs are presumably limited by the supply of methyl donors rather than by H2 and, in total, generally account for