Rumen microbiota and dietary fat: a mutual shaping
Summary
Although fat content in usual ruminant diets is very low, fat supplements can be given to farm ruminants to modulate rumen activity or the fatty acid (FA) profile of meat and milk. Unsaturated FAs, which are dominant in common fat sources for ruminants, have negative effects on microbial growth, especially protozoa and fibrolytic bacteria. In turn, the rumen microbiota detoxifies unsaturated FAs (UFAs) through a biohydrogenation (BH) process, transforming dietary UFAs with cis geometrical double-bonds into mainly trans UFAs and, finally, into saturated FAs. Culture studies have provided a large amount of data regarding bacterial species and strains that are affected by UFAs or involved in lipolysis or BH, with a major focus on the Butyrivibrio genus. More recent data using molecular approaches to rumen microbiota extend and challenge these data, but further research will be necessary to improve our understanding of fat and rumen microbiota interactions.
Introduction
In ruminant animals, most of the digestive processes, including carbohydrate fermentation, dietary protein degradation and microbial protein synthesis, are due to the ruminal microbiota, which comprises bacteria, protozoa, fungi and archaea. This symbiotic microbiota provides available nutrients to the host, especially volatile fatty acids (VFA) and microbial proteins and vitamins, but also results in energy waste due to methane production, making it a major determinant of feed efficiency. Despite a high functional stability due to functional redundancy and the resilience of microbial ecosystem (Weimer 2015), the rumen microbiota exhibits large interindividual variations (Jami and Mizrahi 2012) and can be disturbed by abrupt or major dietary changes, for example regarding starch or fat content.
In most ruminant diets, fat represents under 5% of total dry matter, and linoleic acid (LA; cis-9,cis-12-C18:2) and α-linolenic acid (ALA; cis-9,cis-12,cis-15-C18:3) are the major fatty acids (FA) in forage and most concentrates. Oilseeds such as linseeds, rapeseeds, soybeans and sunflower seeds can be used in farm ruminant diets to carry unsaturated FA (UFA), including oleic acid (OLA; cis-9-C18:1), LA and ALA. Fat supplements were originally developed to improve the energetic values of diets to meet the energy requirements of dairy cows or fattening cattle in intensive farming systems. Fat supplementation can also modify the FA profile of meat (Wood et al. 2008) or milk (Chilliard et al. 2007), modulating their dietetic, organoleptic and technological properties (i.e. by decreasing the saturated FA/UFA ratio in ruminant products). Fat addition can also result in negative effects (i.e. decreased intake and milk fat content) (Rabiee et al. 2012). Moreover, the addition of fat to the diet can modulate rumen function, mitigating methane emissions (Martin et al. 2016).
A strong limit to fat supplementation of ruminant diets is its negative effects on ruminal bacterial degradation, especially with UFA (Brooks et al. 1954). In turn, the ruminal microbiota extensively hydrogenates UFA (Reiser 1951), which is considered as a detoxifying adaptation (Kemp et al. 1984; Maia et al. 2010), and also marginally contributes to the disposal of reducing equivalents produced by rumen fermentation (Lourenço et al. 2010). This biohydrogenation (BH) process comprises several steps, depending on UFA, and several pathways, depending on diet and ruminal environment (Griinari et al. 1998).
This paper provides an overview of existing knowledge about the interactions between dietary fat and rumen microbiota, from early culture studies to results obtained with currently available molecular methods that point out the limits of previous studies and offer new insights and perspectives into applications.
Dietary fat shapes rumen microbiota
A pioneer study (Brooks et al. 1954) showed that both in vitro and in vivo, corn oil decreases ruminal cellulose degradation and VFA concentration and affects the microbiota by increasing small cocci and decreasing small rods. It also showed that lard, which is more saturated than corn oil, results in a smaller decrease in cellulose degradation, and that alfalfa ash addition partly alleviates the negative effects of fat. Similarly, Ikwuegbu and Sutton (1982) found decreased fibre degradability, a decreased proportion of acetate and butyrate, and an increased proportion of propionate among rumen VFA when using linseed oil.
It should be noted that most experiments trials were performed with oil addition, which contrasts with field conditions where oilseeds are the most common fat source, and most experimental amounts exceeded common practice. Consequently, the experimental results regarding the effects of added fat on microbiota and its activity must be taken with caution when extrapolated to field conditions.
Effects on protozoa, archaea and fungi
In vivo studies have shown either no change or a decrease in total and major protozoa genera, with the greatest effects observed with linseed oil addition, especially in high concentrate diets (Table 1). Lauric acid (C12:0) strongly decreased protozoa counts compared to myristic (C14:0) and stearic acids (C18:0) (Hristov et al. 2012). More generally, Oldick and Firkins (2000) showed that increasing the degree of unsaturation decreases protozoal count, but emphasized that this change can be difficult to assess due to large random and animal variations, which could explain inconsistent data across experiments.
| Protozoa | Concentrate percentage | Added fat source | Effect (%) | Reference |
|---|---|---|---|---|
| Total protozoa | 66 | LSO (4·3) | −98 | Ikwuegbu and Sutton (1982) |
| 35 | LSO (3) | NS | Ueda et al. (2003) | |
| 65 | LSO (3) | −84 | Ueda et al. (2003) | |
| 40 | SBO (4) | −23 | Yang et al. (2009) | |
| 40 | LSO (4) | −37 | Yang et al. (2009) | |
| 50 | LSO (4) | NS | Benchaar et al. (2012) | |
| 50 | Linseeds (2·3–5·6) | −18 to −89 | Martin et al. (2016) | |
| Dasytricha | 35–65 | LSO (3–4) | NS | Ueda et al. (2003); Benchaar et al. (2012) |
| Entodinium | 35 | LSO (3) | NS | Ueda et al. (2003) |
| 65 | LSO (3) | −86 | Ueda et al. (2003) | |
| 50 | LSO (4) | NS | Benchaar et al. (2012) | |
| Epidinium | 35 or 65 | LSO (3) | −74 | Ueda et al. (2003) |
| 50 | LSO (4) | NS | Benchaar et al. (2012) | |
| Eudiplodinium | 50 | LSO (4) | NS | Benchaar et al. (2012) |
| Isotricha | 35–65 | LSO (3–4) | NS | Ueda et al. (2003); Benchaar et al. (2012) |
| Ostracodinium | 50 | LSO (4) | NS | Benchaar et al. (2012) |
- Values within parenthesis are percentage of fat addition in dietary dry matter.
- LSO, linseed oil; SBO, soybean oil; NS, nonsignificant.
Studies with pure strains of archaea showed an inhibition of methane production by Methanobrevibacter ruminantium (formerly Methanobacterium ruminantium), the most abundant species of methanogens in the rumen (Henderson et al. 2015), when OA or saturated FA were added to the culture medium (Henderson 1973). Lillis et al. (2011) showed that soybean oil addition in vivo altered the abundance but not the diversity of methanogens, and hypothesized that these changes could be a consequence of the altered VFA profile (less acetate and butyrate whose production makes H2, and more propionate whose production consumes H2) due to changes in the bacterial community. Additionally, Hristov et al. (2012) hypothesized that changes in the archaeal community could be a consequence of the decreased abundance of protozoa with fat-enriched diets. Decreased methane production due to linseed, a high ALA seed, coconut oil that carries 8–14-carbon FA or fish oil that carries 20- and 22-carbon PUFA is not clearly linked to methanogenic gene abundance or changes in the archaeal community (Patra and Yu 2013; Martin et al. 2016).
In pure cultures, growth of the fungus Neocallimastix frontalis can be affected by LA (Maia et al. 2007). Boots et al. (2012) confirmed in vivo the negative effect of added LA on the Neocallimastigales order, whose richness and diversity are decreased by soybean oil addition.
Effects on bacteria growth in cultures
The inhibitory effect of oils on bacterial growth has been extensively studied on pure cultures of rumen strains (Maczulak et al. 1981; Maia et al. 2007), focusing on bacteria known to play a role in fibrolysis, amylolysis and FA metabolism (Table 2). The early study of Maczulak et al. (1981) showed that only Prevotella ruminicola (formerly Bacteroides ruminicola) and some strains of Butyrivibrio fibrisolvens are negatively affected by palmitic (C16:0) and stearic acids. Oleic acid is far more inhibitory than palmitic and stearic acids on the growth of most fibrolytic bacteria but stimulates the growth of Selenomonas ruminantium and P. ruminicola, whereas vaccenic acid (VA; trans-11-C18:1) is far less inhibitory than OLA on Ruminococcus species. These results were consistent with previous data indicating that propionate-producing bacteria (Anaerovibrio lipolyticus, S. ruminantium, Megasphaera elsdenii and P. ruminicola) are not negatively affected by OA, whereas Ruminococcus and B. fibrisolvens, which are mainly acetate and butyrate producers, are negatively affected by both OA and saturated FA (Henderson 1973). As a whole, these early studies on saturated and monounsaturated FA emphasized that the effects of FA on rumen bacteria depend on bacterial metabolism, FA unsaturation and the geometric configuration of double-bonds.
| Bacterial species | Added fatty acid | Effect on growth | Reference |
|---|---|---|---|
| Anaerovibrio lipolyticus | OLA (50–100) | No effect | Henderson (1973) |
| LA or ALA (50) | No effect | Maia et al. (2007) | |
| Butyrivibrio fibrisolvens | PMA (10–200) | No or decreasea | Maczulak et al. (1981) |
| STA (10–100) | No or decreasea | Maczulak et al. (1981) | |
| OA (1–10) | No or decreasea | Maczulak et al. (1981) | |
| VA (100) | No effect | Maczulak et al. (1981) | |
| LA or ALA (50) | No or increaseb | Maia et al. (2007) | |
| Butyrivibrio hungatei | LA or ALA (50) | No growth | Maia et al. (2007) |
| Butyrivibrio proteoclasticus | LA (50) | Strong decrease | Maia et al. (2007) |
| ALA (50) | No growth | Maia et al. (2007) | |
| Fibrobacter succinogenes | PMA (10–200) | No effect | Maczulak et al. (1981) |
| STA (10–100) | No effect | Maczulak et al. (1981) | |
| LA (50) | Strong decrease | Maia et al. (2007) | |
| ALA (50) | No growth | Maia et al. (2007) | |
| Megasphaera elsdenii | LA or ALA (50) | No effect | Maia et al. (2007) |
| Ruminococcus albus | PMA (10–200) | No effect | Maczulak et al. (1981) |
| STA (10–100) | No effect | Maczulak et al. (1981) | |
| OA (1–10) | Strong decrease or no growth | Maczulak et al. (1981) | |
| VA (100) | Decrease | Maczulak et al. (1981) | |
| LA or ALA (50) | No growth | Maia et al. (2007) | |
| Ruminococcus flavefaciens | PMA (10–200) | No effect | Maczulak et al. (1981) |
| STA (10–100) | No effect | Maczulak et al. (1981) | |
| OA (1–10) | Strong decrease or no growth | Maczulak et al. (1981) | |
| VA (100) | Decrease | Maczulak et al. (1981) | |
| LA or ALA (50) | No growth | Maia et al. (2007) | |
| Selenomonas ruminantium | PMA (10–200) | No effect | Maczulak et al. (1981) |
| STA (10–100) | No effect | Maczulak et al. (1981) | |
| OA (1–10) | Increase | Maczulak et al. (1981) | |
| VA (100) | No effect | Maczulak et al. (1981) | |
| LA or ALA (50) | No effect | Maia et al. (2007) | |
| Streptococcus bovis | LA or ALA (50) | No effect | Maia et al. (2007) |
- Values within parenthesis are the concentration (mg l−1).
- OA, cis-9-C18:1 (oleic acid); PMA, C16:0 (palmitic acid), STA, C18:0 (stearic acid); VA, trans-11 C18:1 (vaccenic acid).
- a Depending on the strain.
- b Depending on lactate addition in the culture medium.
Further studies on pure strains addressed the effects of PUFA and showed that LA or conjugated linoleic acids (CLA) prevent the growth of B. fibrisolvens A38 at very low concentrations (4 mg l−1) (Kim et al. 2000). Using a different culture medium, Maia et al. (2007) showed that growth of the JW11 strain is not inhibited at 50 mg l−1. Other species of the Butyrivibrio genus, including Butyrivibrio hungatei and stearic-producing strains close to Butyrivibrio proteclasticus (formerly Clostridium proteoclasticum), are inhibited from 5 mg of LAl−1 (Paillard et al. 2007a). This latter species cannot grow in the presence of 50 mg l−1 of cis-9,trans-11 CLA or trans-10,cis-12 CLA (McKain et al. 2010).
Negative effects of FA on B. fibrisolvens are stronger for ALA than LA, and even stronger for long-chain PUFA eicosapentaenoic (EPA; cis-5,cis-8,cis-11,cis-14,cis-17-C20:5) and docosahexaenoic (DHA; cis-4,cis-7,cis-10,cis-13,cis-16,cis-19-C22:6) acids due to a longer lag phase in growth cultures (Fukuda et al. 2009; Maia et al. 2010). Similarly, ALA strongly increases the lag phase and lowers the growth rate of Propionibacterium acnes, a lactate utilizer in the rumen (Maia et al. 2016). On the contrary, bacteria belonging to the genera Prevotella, Megasphaera, Selenomonas, Veillonella, Anaerovibrio and Streptococcus are poorly or not affected by LA and ALA (Maia et al. 2007).
Effects on bacteria abundance in vivo
In addition to these studies that focused on pure strains of rumen bacteria, the effects of fat supplements have been investigated in vivo (Table 3), assigning bacteria at the species level using quantitative PCR (Yang et al. 2009; Huws et al. 2010; Liu et al. 2011; Shingfield et al. 2012; Martin et al. 2016; Vargas-Bello-Pérez et al. 2016) or at the genus level using 16S rDNA pyrosequencing (Zened et al. 2013a; Huws et al. 2014; Li et al. 2015). These latter studies reported significant effects of fat on as yet uncultured or unclassified bacteria. These experiments were mainly based on oil addition, as opposed to most studies on pure strain cultures that used free FA. As a whole, effects were lower than observed with pure cultures, which could be due to the type of added fat or to the fact that the effects on the lag phase cannot be seen in vivo. Changes in rumen microbiota due to an increased concentrate proportion are much higher than effects due to fat addition, and some genera were differently affected by oil addition in low and high concentrate diets, especially Acetitomaculum, Lachnospira and Prevotella (Zened et al. 2013a).
| Bacteria | Main forage and concentrate percentage | Added fat source | Percentage variation | Reference |
|---|---|---|---|---|
| Anaerovibrio lipolyticus | GS 0 | FO (1–4) | NS | Huws et al. (2010) |
| GS+CS 60 | SBO (2·7) | NS | Vargas-Bello-Pérez et al. (2016) | |
| GS+CS 60 | HPMO (2·7) | NS | Vargas-Bello-Pérez et al. (2016) | |
| Genus Butyrivibrio | GS 40 | LSO (3) | +52 | Huws et al. (2014) |
| CS 17 | SFO (5) | NS | Zened et al. (2013a) | |
| CS 65 | SFO (5) | NS | Zened et al. (2013a) | |
| Straw 80 | LSO (3·2) | −40 | Li et al. (2015) | |
| Butyrivibrio fibrisolvens | CS 65 | SFO (5) | NS | Zened et al. (2013a) |
| Pseudobutyrivibrio | GS 40 | LSO (3) | +57 | Huws et al. (2014) |
| Butyrivibrio hungatei or Butyrivibrio proteoclasticus | CS 65 | SFO (5) | NS | Zened et al. (2013a) |
| GS+CS 60 | SBO (2·7) | NS | Vargas-Bello-Pérez et al. (2016) | |
| GS+CS 60 | HPMO (2·7) | NS | Vargas-Bello-Pérez et al. (2016) | |
| Butyrivibrio hungatei | GS 52 | FO (0·4–1·6) | NS | Shingfield et al. (2012) |
| Butyrivibrio proteoclasticus | GS + 0 | FO (1–4) | NS | Huws et al. (2010) |
| GS 52 | FO (0·4–1·6) | NS | Shingfield et al. (2012) | |
| Genus Fibrobacter | GS 40 | LSO (3) | −68 | Huws et al. (2014) |
| Fibrobacter succinogenes | AH+CS 40 | SBO (4) | −39 | Yang et al. (2009) |
| AH+CS 40 | LSO (4) | −94 | Yang et al. (2009) | |
| GS 0 | FO (1–4) |
NS in LAB −17 in SAB |
Huws et al. (2010) | |
| Mixed 40 | LA (2·7) | NS | Liu et al. (2011) | |
| Mixed 40 | DHA (0·5) | −42 | Liu et al. (2011) | |
| GS+CS 60 | SBO (2·7) | NS | Vargas-Bello-Pérez et al. (2016) | |
| GS+CS 60 | HPMO (2·7) | NS | Vargas-Bello-Pérez et al. (2016) | |
| CS 40 | Linseeds (2·3–5·6) | NS | Martin et al. (2016) | |
| Genus Lachnospira | CS 17 | SFO (5) | −79 | Zened et al. (2013a) |
| CS 65 | SFO (5) | +1500 | Zened et al. (2013a) | |
| Lachnospiraceae incertae sedis | CS 17 | SFO (5) | −17 | Zened et al. (2013a) |
| CS 65 | SFO (5) | +450 | Zened et al. (2013a) | |
| Megasphaera elsdenii | GS 0 | FO (1–4) | ND | Huws et al. (2010) |
| Mixed 40 | LA (2·7) | NS | Liu et al. (2011) | |
| Mixed 40 | DHA (0·5) | NS | Liu et al. (2011) | |
| Genus Prevotella | GS 40 | LSO (3) | −68 | Huws et al. (2014) |
| CS 17 | SFO (5) | −17 | Zened et al. (2013a) | |
| CS 65 | SFO (5) | +119 | Zened et al. (2013a) | |
| Straw 80 | LSO (3·2) | +100 | Li et al. (2015) | |
| GS 0 | FO (1–4) | NS | Huws et al. (2010) | |
| RC9-gut-group | CS 17 | SFO (5) | −38 | Zened et al. (2013a) |
| CS 65 | SFO (5) | −84 | Zened et al. (2013a) | |
| Straw 80 | LSO (3·2) | +24 | Li et al. (2015) | |
| Genus Ruminococcus | GS 40 | LSO (3) | −62 | Huws et al. (2014) |
| CS 17 | SFO (5) | NS | Zened et al. (2013a) | |
| CS 65 | SFO (5) | NS | Zened et al. (2013a) | |
| Straw 80 | LSO (3·2) | +17 | Li et al. (2015) | |
| Ruminococcus albus | GS 0 | FO (1–4) | NS | Huws et al. (2010) |
| CS 40 | Linseeds (2·3–5·6) | NS | Martin et al. (2016) | |
| Ruminococcus flavefaciens | AH+CS 40 | SBO (4) | −4 | Yang et al. (2009) |
| AH+CS 40 | LSO (4) | −28 | Yang et al. (2009) | |
| GS 0 | FO (1–4) | NS | Huws et al. (2010) | |
| Mixed 40 | LA (2·7) | NS | Liu et al. (2011) | |
| Mixed 40 | DHA (0·5) | NS | Liu et al. (2011) | |
| CS 40 | Linseeds (2·3–5·6) | NS | Martin et al. (2016) | |
| Genus Selenomonas | CS 17 | SFO (5) | NS | Zened et al. (2013a) |
| CS 65 | SFO (5) | NS | Zened et al. (2013a) | |
| Streptococcus bovis | GS 0 | FO (1–4) | ND | Huws et al. (2010) |
| GS 52 | FO (0·4–1·6) | NS | Shingfield et al. (2012) |
- AH, alfalfa hay; CS, corn silage; GS, grass silage; DHA, docosahexaenoic acid; FO, fish oil; HPMO, hydrogenated palm oil; LA, linoleic acid; LSO, linseed oil; SBO, soybean oil; SFO, sunflower oil. LAB, liquid-associated bacteria; SAB, solid-associated bacteria; NS, non significant.
- Values within parenthesis are percentage of fat addition in dietary dry matter.
Among bacteria genera or species that were studied in several experiments, Fibrobacter and Ruminococcus were negatively affected in most cases, but the effects on Butyrivibrio and Prevotella were highly variable. These latter genera comprise many species with somewhat different functions, different metabolic pathways (Hackmann and Firkins 2015) and different sensitivities to LA in cultures (Maia et al. 2007). Moreover, a decreased abundance in vivo of a bacterial genus following a dietary change cannot be unequivocally interpreted as a direct effect of a dietary component, but could also reflect a more global change in nutrient and function partitioning among different rumen micro-organisms. As expected from culture studies, high ALA oils are more inhibitory than high LA oils (Yang et al. 2009), but Table 3 shows no clear difference in the effect of fat supplements according to type of silage (corn vs grass) and concentrate percentage in the diet.
Mechanism of FA effect on rumen micro-organisms
The mechanism of protozoa inhibition by medium-chain FA or UFA has not been established but could be due to the incorporation of dietary FA or BH products into membranes (Reveneau et al. 2012; Diaz et al. 2014) or to alterations in chemotaxis and substrate acquisition (Diaz et al. 2014).
Several hypotheses have been proposed to explain the inhibitory mechanism of FA on bacterial growth. Most lipids are associated with dietary particles in the rumen, and the coating of dietary particles could impair adhesion, decreasing fibre degradation in the rumen (Devendra and Lewis 1974). This could explain why fibre addition decreases the negative effects of fat addition on bacterial growth (Maczulak et al. 1981). However, this hypothesis is not consistent with the increased number of bacteria that adhere to particles in the solid phase of the rumen when fat is added to the diet (Bauchart et al. 2007). The effect of fibre addition could also rely on a more efficient UFA BH process, thus lowering rumen concentrations of FAs with an inhibiting effect (Yang et al. 2009).
Devendra and Lewis (1974) also proposed that dietary fat could decrease the availability of cations to bacteria due to salt formation, which is consistent with the protective effects of alfalfa ash, rich in calcium, on cellulose degradation when oil was added to the diet (Brooks et al. 1954). The addition of calcium to the diet reversed the decrease in ionized calcium concentration observed in the rumen after fat addition and the effects of fat on fibre degradation (Palmquist et al. 1986). However, this interaction of dietary fat with cations cannot explain all of the negative effects of dietary fat. Indeed, saturated FA salts have a higher ruminal stability than UFA salts (Jenkins and Palmquist 1982). Therefore, in diets without calcium addition, saturated FA should trap more calcium than UFA and, consequently, more strongly inhibit bacteria, which is the opposite of the hierarchy observed.
Finally, Devendra and Lewis (1974) also hypothesized that FA could exert a direct toxicity on rumen bacteria, which is consistent with the incorporation of UFA into bacteria: PUFA represent up to 20% of the total FA in solid-adherent bacteria (Bauchart et al. 1990). This toxicity could be due to an impediment in the nutrient passage due to FA adhering to the cell wall (Henderson 1973). As opposed to trans double-bonds, if the cis stereochemical configuration of FA, which is the common configuration in dietary lipids, is incorporated into the bacteria cell membrane, it can result in a high fluidity due to the 30° angle of the acyl chain. A specific effect of the cis double-bonds could explain why most known biohydrogenating bacteria produce trans-C18:1, but do not further reduce it to stearic acid (see below). Hackmann and Firkins (2015) reported that B. fibrisolvens could be less sensitive to LA than B. hungatei and Butyrivibrio proteoclasticus because its membrane has a lower fluidity due to more palmitic acid and less branched-chain FA.
Linoleic acid can disrupt cell integrity, but there is no relationship between this disruption and the level of growth inhibition across different strains of rumen bacteria, including B. fibrisolvens (Maia et al. 2007, 2010). Knowing that B. fibrisolvens, which produces butyrate via a butyryl CoA transferase, is less sensitive to LA than B. hungatei and B. proteoclasticus, which produce butyrate using a butyrate kinase (Paillard et al. 2007a), Maia et al. (2010) concluded that PUFA toxicity in butyrate producers is probably mediated via a metabolic effect involving butyrate production. Beyond butyrate metabolism, other Acyl CoA and ATP pools are affected by exposure to PUFA, suggesting a metabolic interruption (Firkins and Yu 2015).
Rumen BH: a microbiota response to dietary UFAs
Most dietary FA are glyceride esters: mainly triacylglycerols in concentrates and galactolipids and phospholipids in forage, except in silage where free FA are released by plant lipases (van Ranst et al. 2011).
BH and rumen lipolysis pathways
The first step of ruminal metabolism of acylglycerols is lipolysis, resulting in free FA release. During ruminal lipolysis of triacylglycerols, the concentration of partial acylglycerols remains low (Noble et al. 1974), suggesting that lipolysis of di- and monoacylglycerols is quicker than that of triacylglycerols. FAs released by rumen lipolysis remain adsorbed on feed particles and are in part incorporated into solid-adherent bacteria.
Most UFAs released by lipolysis undergo BH. Figure 1 shows the main BH pathways of OLA, LA and ALA. It is mainly a one-step reduction for OLA, and a succession of reactions for PUFAs. The first one is an isomerization that produces CLA or conjugated linolenic acids (CLnA) through the displacement of a double-bond and its transformation from a cis to a trans geometric configuration. According to the double-bond involved, several isomers can be produced, the dominant being trans-11 FAs. In most conditions, this first reaction results in the disappearance of >70% of OA, >80% of LA and about 90% of ALA (Enjalbert and Troegeler-Meynadier 2009). The subsequent reactions are reductions, first affecting cis double-bonds. The last reduction hydrogenates the trans double-bond created during the isomerization. It is slower than the previous reductions so that trans-C18:1 FAs accumulate in the rumen and flow into the small intestine in much larger amounts than CLA and CLnA.
Beside the major pathways involving trans-11 BH intermediates, the trans-10 pathway can become dominant, especially when ruminants are fed high LA and high-concentrate diets (Zened et al. 2013b). Although ALA can result in trans-10 FA as BH intermediates, it is a minor precursor of trans-10 FA, even when ALA is the predominant dietary FA (Lee and Jenkins 2011; Zened et al. 2011; Alves and Bessa 2014). A large number of other positional or geometrical isomers of CLnA, CLA or trans-C18:1 are found in rumen contents, with double-bond positions ranging from carbons 7 to 15 (Honkanen et al. 2016), 7 to 14 (Jenkins et al. 2008) and 4 to 16 (Loor et al. 2002) for CLnA, CLA and trans-C18:1 respectively. Different isomers of CLnA or CLA could result from alternative isomerization pathways in the rumen (McKain et al. 2010; Honkanen et al. 2016) or from interconversions for C18:1 FA (Proell et al. 2002; Laverroux et al. 2011). Geometrical isomerization of CLA isomers can also be an analytical artefact (Kramer et al. 1997).
Contrary to BH, OLA and LA can undergo hydration, resulting in 10-hydroxystearic and 10-ketostearic acids (Hudson et al. 1995; Jenkins et al. 2006) or 10-hydroxy,cis-12-C18:1 and 13-hydroxy,cis-9-C18:1 (Hudson et al. 1998) respectively.
Lipolytic and biohydrogenating micro-organisms and enzymes
Lipolytic micro-organisms and enzymes
Protozoal lipolytic activity has not been extensively studied, and the possibility that ingested bacteria still exhibit a lipolytic activity makes it difficult to assess the specific activity of protozoa (Lourenço et al. 2010).
Phospholipids and galactolipids can be hydrolysed by some strains of B. fibrisolvens (Hazlewood and Dawson 1979). Triacylglycerols are also hydrolysed by different species of the Butyrivibrio group (Latham et al. 1972; Paillard et al. 2007a), but A. lipolyticus is the best-known triacylglycerol-hydrolysing bacteria. Its relative 16S rRNA gene abundance is around 0·05% in the rumen (Minuti et al. 2015). Its lipase was first studied by Henderson (1971) and its genome contains three genes coding for lipases (Privé et al. 2013). The characteristics of A. lipolyticus lipases are summarized in Table 4. The three enzymes were more active against laurate and myristate than palmitate and stearate, whereas dietary FAs mainly contain 16- and 18-carbon FAs. These studies did not investigate the lipolytic activity against PUFA, which are dominant in most ruminant diets.
| Enzyme | Origin | Localization | Optimal pH | Amino-acids | Molecular weight (kDa) | Reference |
|---|---|---|---|---|---|---|
| Lipases | Anaerovibrio lipolyticus | Extracellular | 7·4 | 248–492 | 28–56 | Henderson (1971); Privé et al. (2013) |
| Lipase | Pseudomonas aeruginosa | – | 8·0 | – | 29 | Unni et al. (2016) |
| Lipases | Rumen metagenome | – | 7·5 | – | – | Liu et al. (2009); Privé et al. (2015) |
| Δ12 isomerase | Butyrivibrio fibrisolvens | Membrane | 7·0–7·2 | – | – | Kepler and Tove (1967) |
| Δ9 isomerase | Propionibacterium acnes | Cytosol | 7·2–7·5 | 424 | 48 | Deng et al. (2007); Farmani et al. (2010) |
| cis-9,trans-11 CLA reductase | Butyrivibrio fibrisolvens | Membrane | 7·2–8·2 | – | 53 | Hughes et al. (1982); Fukuda et al. (2007) |
More recently, lipolytic activity was reported for other rumen bacteria belonging to the Clostridium, Propionibacterium, Staphylococcus and Selenomonas genera (Edwards et al. 2013), and Unni et al. (2016) purified the lipase of a rumen Pseudomonas aeruginosa strain (Table 4). The effectiveness of an antibody against a commercial purified Pseudomonas lipase to decrease lipolytic activity in cultures of A. lipolyticus, B. fibrisolvens, Propionibacterium avidum and P. acnes strongly suggests genetic similarity between rumen bacteria lipases (Edwards et al. 2017).
Liu et al. (2009) characterized two lipases from a metagenomics library of cow rumens, with a high affinity for 16- and 18-carbon FAs. In the same way, Privé et al. (2015) isolated 14 novel lipases from a bovine rumen metagenome, mainly active on short- and medium-chain FA esters. Such studies cannot indicate which bacteria produce these lipases that could originate from unknown or uncultured bacteria. In the whole rumen content, Moate et al. (2008) reported a Km of 0·8 × 10−3 M for the lipolysis of trilinolein, considering that Michaelis–Menten kinetics was applicable for this reaction. This value is in the range of triglyceride concentrations in the rumen of a cow after a meal. For example, a 5-kg dry matter meal containing 2% triglycerides would lead to an initial concentration slightly over 1 mmol l−1, rapidly decreasing over time.
Biohydrogenating micro-organisms and enzymes
The involvement of protozoa in rumen BH remains unclear. Devillard et al. (2006) showed that LA does not disappear when incubated with protozoa alone, and defaunation did not clearly modify rumen LA metabolism (Yáñez-Ruiz et al. 2006). Protozoa can engulf bacteria, and bacterial BH could take place inside protozoa (Jenkins et al. 2008) and explain their high concentrations of BH intermediates (Devillard et al. 2006). Rumen fungi also have a limited ability to biohydrogenate LA (Maia et al. 2007; Jenkins et al. 2008).
Pioneer data showed that B. fibrisolvens reduces LA to trans C18:1 but not to stearic acid (Kepler et al. 1966) and that it also hydrogenates ALA to trans intermediates (Kepler and Tove 1967). It cannot metabolize EPA and DHA (Maia et al. 2007) as opposed to B. proteoclasticus (Jeyanathan et al. 2016). Strains of B. hungatei also produce VA from LA (Maia et al. 2007). Paillard et al. (2007a) and Hussain et al. (2016) showed differences in LA metabolism among several tens of Butyrivibrio isolates, a large number being able to metabolize LA to VA. Slight differences in BH pathways between strains have also been found for ALA metabolism, MDT-5, A38 and MDT-10 strains producing trans-11,cis-13-CLA, trans-11,cis-15-C18:2 or VA, respectively, when incubated with ALA (Fukuda et al. 2009). The Butyrivibrio genus has a relative 16S rRNA gene abundance averaging 3·4% in the rumen across different species, including 0·25% of B. fibrisolvens (Henderson et al. 2015). It is established in the rumen as early as 2 days after birth (Rey et al. 2013).
The first BH step by B. fibrisolvens is mainly a Δ12 isomerization, leading to trans-11 FA. Kepler and Tove (1967) revealed that B. fibrisolvens Δ12 isomerase is probably localized in membranes or tightly bound to membranes of the bacteria (Table 4). Its substrates are FA with a free carboxyl function and double-bounds on cis-9 and cis-12 carbons (Kepler et al. 1970). Moate et al. (2008) modelled this isomerization using Michaelis–Menten kinetics and obtained a Km = 1·6 × 10−3 mol l−1. The optimal 7·0 to 7·2 pH of B. fibrisolvens linoleate isomerase (Kepler and Tove 1967) explains its inhibition by low pH (Troegeler-Meynadier et al. 2006). It is also inhibited by LA and ALA (Kepler and Tove 1967) due to a saturation of the isomerase by an excess of substrate (Troegeler-Meynadier et al. 2006) or by an excess of product due to a lack of CLA release by the isomerase, preventing it from recycling like other enzymes and consequently catalysing more substrate (Kim et al. 2000). Wallace et al. (2007) suggested that B. fibrisolvens linoleate isomerase acts as a radical intermediate enzyme. This could explain the inhibition of Δ12 isomerase by 13-hydroperoxy-cis-9,trans-11-C18:2 and 9-hydroperoxy-trans-10,cis-12-C18:2 (Kaleem et al. 2013), possibly chemically close to some LA BH intermediates.
Other bacteria have been shown to be able to isomerize LA to cis-9,trans-11 CLA. They were isolated from rumen (Unni et al. 2016) or other digestive tracts (Verhulst et al. 1985; Devillard et al. 2007) and mainly belong to the Clostridium, Pseudobutyrivibrio, Lactobacillus, Propionibacterium, Bifidobacterium, Eubacterium, Roseburia, Enterococcus and Pediococcus genera. Lactic bacteria produce CLA through a hydration–dehydration process with a hydroxy FA as an intermediate (Ogawa et al. 2005). Hydration of UFA in the rumen is mainly due to Streptococcus bovis. Several strains of Streptococcus, Staphylococcus, Lactobacillus, Enterococcus and Pediococcus can also catalyse this reaction (Hudson et al. 2000).
Other rumen species, including Ruminococcus albus F2/6 (Kemp et al. 1975), can convert LA and ALA to C18:1, but their relative activity in the rumen remains unknown (Jenkins et al. 2008). As opposed to B. fibrisolvens that produces trans-11 double-bonds, R. albus F2/6 produces large proportions of trans-10-C18:1 (Kemp et al. 1975). Kim et al. (2002) isolated a bacteria identified as M. elsdenii YJ-4 producing trans-10,cis-12-CLA from LA from the rumen of a cow receiving a high starch grain diet and found that the T81 strain also produces this isomer. However, Maia et al. (2007) found that only a contaminated strain of Megasphaera esldenii T81 and not a pure strain produced trans-10,cis-12-CLA, and Shingfield et al. (2012) emphasized that M. elsdenii is often below the detection limit in the rumen. Wallace et al. (2007) demonstrated the production of trans-10,cis-12-CLA by a P. acnes strain isolated from ovine rumen. This bacterium does not further reduce trans-10,cis-12-CLA (McKain et al. 2010) and isomerizes ALA to several C18:3 intermediates without further reduction to C18:2 or C18:1 FAs, and without producing trans-10,cis-12,cis-15-CLnA (Maia et al. 2016).
Deng et al. (2007) purified and characterized the linoleate isomerase of P. acnes ATCC 6919 (Table 4), which requires FAD as a cofactor and is not sensitive to an excess of substrate (Farmani et al. 2010). Liavonchanka et al. (2006) postulated that rumen Δ9 isomerization is an ionic reaction, with a LA carbocation as the intermediate.
The cis-9,trans-11 CLA reductase of B. fibrisolvens was studied by Hughes et al. (1982) and Fukuda et al. (2007) (Table 4). It is associated with a phosphatidylethanolamine and represents 0·5% of the cell protein. It requires iron, and α-tocopherolquinol and NADH are also involved in the reduction system, which does not contribute to ATP synthesis (Hackmann and Firkins 2015). This enzyme recognizes conjugated double-bounds, and 18-carbon UFAs increase its expression at the transcriptional level. It not only reduces cis-9,trans-11 CLA but trans-10,cis-12-CLA and cis-9,trans-11,cis-15-CLnA as well, but not trans-11,cis-15-C18:2 (Fukuda et al. 2007). Moate et al. (2008) modelled the reduction of cis-9,trans-11-CLA to VA by a Michaelis–Menten-type process involving competitive inhibition by VA: Km = 3·6 × 10−6 mol l−1. High CLA and VA amounts in the media decrease its efficiency by competitive inhibition (Troegeler-Meynadier et al. 2006).
Two bacteria belonging to the Fusocillus genus were shown to reduce C18:1 FA to stearic acid (Harfoot 1978). van de Vossenberg and Joblin (2003) isolated a strain of Butyrivibrio (identified as B. hungatei Su6), phenotypically similar to Fusocillus and able to complete the BH of both LA and ALA to stearic acid. Based on the studies of Paillard et al. (2007a), Wallace (2008) concluded that the stearic acid-forming isolates previously identified as Fusocillus sp. or B. hungatei occupy a specific branch of the Butyrivibrio tree, which includes B. proteoclasticus. The B316 and P-18 strains of this species are stearate producers (Wallace et al. 2006; Maia et al. 2007). Butyrivibrio proteoclasticus has been reported to represent a major part of the Butyrivibrio group in cows and goats (Paillard et al. 2007b), but recent data report only a 0·04% relative 16S rRNA gene abundance across several ruminant species (Henderson et al. 2015). Other bacteria belonging to the Butyrivibrio group are able to produce stearic acid from LA, and uncultured bacteria also participate in VA reduction (Li et al. 2012).
The VA reductase is different from the CLA reductase since many bacteria that convert CLA to VA do not reduce VA to stearic acid. Moreover, in B. proteoclasticus, which reduces both CLA and VA, the two reductases may be different because they were not similarly affected by lactic acid (Maia et al. 2007). Moate et al. (2008) showed that reduction of VA to stearic acid could be modelled by a quasi-first-order process (k = 0·533/h) with an inhibition of this reaction by VA itself, as reported by Troegeler-Meynadier et al. (2006) as well who also showed that this enzyme is inhibited by a low pH.
Beyond studies based on the culture of selected isolates, attempts have been made to assess in vivo or in vitro the relationship between rumen bacteria and BH by adding bacteria and measuring BH products, or by adding dietary supplements that are known to affect BH and measuring bacteria abundance. Inoculating B. fibrisolvens into the rumen of goats fed a diet enriched with a high LA oil increased VA and total CLA concentrations in the rumen fluid, which confirms that this bacterium is involved in BH in vivo (Shivani et al. 2016). Long-chain PUFAs from fish oil or marine algae are known to lower the reduction from VA to stearic acid, and because B. proteoclasticus is the only well-characterized stearate producer, it can be hypothesized that these supplements affect B. proteoclasticus abundance. Indeed, AbuGhazaleh and Ishlak (2014) showed that fish oil decreases the abundance of different bacteria belonging to the Butyrivibrio group, including B. proteoclasticus, in continuous cultures. On the contrary, Kim et al. (2008) and Huws et al. (2010) in steers and Shingfield et al. (2012) in lactating dairy cows found no clear change in B. proteoclasticus abundance upon addition of fish oil to the diet and concluded that this species plays a minor role in stearic acid production. Similarly, BH inhibition by algae was not associated with any change in relative abundance of butyrivibrios (Zhu et al. 2016).
Experiments based on molecular fingerprints or next generation sequencing also revealed the lack of a positive relationship between B. proteoclasticus species or the Butyrivibrio genus and their expected products (Huws et al. 2011; Toral et al. 2012; Petri et al. 2014). On the contrary, these workers showed correlations between BH products and diverse bacteria including uncultured Prevotella, Lachnospiraceae incertae sedis, and unclassified Bacteroidales, Clostridiales and Ruminococcaceae (Huws et al. 2011), uncultured Lachnospiraceae or Quinella-related bacteria (Toral et al. 2012), Flavobacterium sp., Anerophaga, Fibrobacter, Guggenheimella, Paludibacter and Pseudozobellia (Petri et al. 2014), and Acetobacter (Bainbridge et al. 2016). As a whole, these correlation studies have not yet added clear conclusions to the generally accepted knowledge regarding bacteria involved in rumen BH or interacting with BH. First, from a bacterial point of view, BH is a detoxification process and not a nutritional process, so that the abundance of biohydrogenating bacteria is probably more strongly linked to their energetic substrate than to UFA that are toxic to them. Second, as outlined by Huws et al. (2011), Petri et al. (2014) and Firkins and Yu (2015), sequencing methods have several drawbacks: most studies do not unequivocally identify species, which is a problem since some known biohydrogenating bacteria are very closely related. They measure DNA concentration, not RNA, the latter being more indicative of the microbial taxa that respond to a feeding challenge. Finally, even enzyme synthesis by active bacteria would not necessarily be strongly linked to the efficiency of BH reactions due to the effects of rumen pH on enzyme activity (see Table 4).
Balance between trans-10 and trans-11 BH pathways
Milk fat depression in dairy cows is linked to a BH shift toward the trans-10 pathway that can be experimentally induced by high starch or high starch plus LA-rich oil diets (Piperova et al. 2002; Weimer et al. 2010; Rico and Harvatine 2013; Zened et al. 2013b). This could be due to a decreased capacity of the microbiota to isomerize the supplemental LA because of a negative effect of both starch and oil addition on bacteria that biohydrogenate via the trans-11 pathway, triggering a necessary adaptation of the microbiota to shift toward another BH pathway (i.e. the trans-10 BH) to avoid the toxic effects of UFA. Rumen production of trans-10 FA or milk fat depression is associated with a rumen pH under 6·0 (Enjalbert et al. 2008) and an acetate-to-propionate ratio under 2·0 (Zened et al. 2013b), which reflect changes in both the ruminal environment and the microbial activity. The bacterial community is strongly affected during the trans-10 shift, with large variability between animals (Weimer et al. 2010; Zened et al. 2013a). More specifically, Rico et al. (2015) showed that the Butyrivibrio group decreases and M. elsdenii increases during induction of milk fat depression, with reverse changes observed during recovery. Because M. elsdenii has been shown to produce trans-10,cis-12-CLA in pure cultures, Weimer et al. (2015) tested the effects of ruminal dosing of several strains of M. elsdenii but could not trigger a milk fat depression because the dosed strains did not establish itself in the rumen. Maia et al. (2007) reported that the high abundance of this species with high starch and high oil diets probably links to its resistance to both low pH and UFA, but Palmonari et al. (2010) observed that the link between milk fat depression and M. elsdenii abundance is independent on rumen pH. Altogether, these data do not provide clear evidence for the involvement of M. elsdenii in the rumen trans-10 BH pathway. These in vivo studies did not report P. acnes abundance, precluding any conclusion on the possible involvement of this species in the trans-10 BH shift. Since P. acnes is a minor species in the rumen, McKain et al. (2010) emphasized that a major implication of this species in the trans-10 BH shift is questionable.
Conclusion and perspectives
Although most ruminant diets have low fat content, fat supplementation is commonly used to improve meat or milk FA profile. The relationship between dietary lipids and rumen microbiota are dominated by toxicity of UFA on many micro-organisms, especially fibrolytic bacteria, and UFA BH by bacteria. The Butyrivibrio genus is known to be strongly involved in the detoxification process. Many recent studies suggest that biochemical pathways are more complex and the bacteria involved might be more diverse than was assumed several decades ago.
Practical applications involve both sides of this relationship. Fat addition to the diet shapes the rumen microbial community, modulating rumen function, and a great deal of research in recent years has been devoted to the mitigation of methane emissions. Fat addition effectively reduces methane, but the mode of action is not fully understood and could depend on the fat source (Patra and Yu 2013). Negative effects on feed efficiency can be observed together with lower methane production (Beauchemin et al. 2009).
Over the last decades, modulating rumen BH has mainly been addressed by treatment of fat sources. More recently, research has focused on approaches linked to microbiota action. Because most dietary FA are glycerides and need lipolysis before BH, decreasing the rumen metabolism of FA could limit rumen BH. A preliminary study using esterase inhibitors has shown promising results in vitro (Sargolzehi et al. 2015). Inhibition of ruminal lipase-producing bacteria using an antilipase antibody affected several lipase-producing bacteria, including B. fibrisolvens, and decreased BH in vitro (Edwards et al. 2017). Fukuda et al. (2006) and Fukuda et al. (2009) postulated that some strains of B. fibrisolvens could be used as a probiotic to better control FA resulting from BH in animal products, but only Shivani et al. (2016) tested this supplementation, obtaining time-dependent effects. Apás et al. (2015) showed an increased proportion of cis-9,trans-11-CLA in the milk of goats supplemented with a mixture of Lactobacillus, Bifidobacterium and Enterococcus.
Plant extracts could also modulate the activity of biohydrogenating bacteria. Essential oils decreased (Durmic et al. 2008) or increased (Ishlak et al. 2015) the abundance of B. fibrisolvens in vitro, which could explain changes in the profile of BH products in the rumen (Lourenço et al. 2010). Similarly, tannins decreased abundances of B. proteoclasticus and increased B. fibrisolvens (Ishlak et al. 2015) in vitro, which is consistent with VA accumulation as observed by Vasta et al. (2009). On the contrary, saponins strongly inhibited B. fibrisolvens growth (Wallace et al. 1994), but did not affect rumen BH (Lourenço et al. 2010).
The most appropriate choices for shaping rumen microbiota and its activity depend on many factors, including production system, economic conditions, local regulations or specifications. However, to allow the use of these different manipulations in the field, new data should be obtained in vivo in various dietary conditions with long-term studies because the resilience of the rumen microbiota or its adaptation to the degradation of plant compounds can alter effects over time (Weimer 2015). Additionally, targeting new applied research on rumen FA metabolism makes it necessary to better understand which micro-organisms, which enzymatic mechanisms, and which interactions between micro-organisms and between microbiota and host are involved. Omics, including metagenomics, metatranscriptomics, metaproteomics and metametabolomics, can be useful tools for such investigations.
Conflict of Interest
The authors declare no conflict of interest.
