Introduction

Unsaturated fatty acids, particularly α-linolenic acid (LNA; cis-9,cis-12,cis-15–18:3) and linoleic acid (LA; cis-9,cis-12–18:2), are abundant in grass and other ruminant feedstuffs, yet are present at low concentrations in meat and milk. Furthermore, tissue lipids of ruminants have been known for a long time to be more saturated than those of non-ruminants (Banks and Hilditch 1931). As the consumption of dairy products and ruminant meats is often associated with an increased incidence of coronary heart disease in man (Menotti et al. 1999), the transformation of unsaturated fatty acids to saturated fatty acids, or biohydrogenation, in ruminants represents a major human health issue. The biohydrogenation process has long been known to occur in the rumen as the result of microbial metabolic activity (Shorland et al. 1955; Viviani 1970). Thus, if ruminal biohydrogenation of unsaturated fatty acids can be controlled, it may be possible to improve the healthiness of ruminant meats and milk by increasing their unsaturated fatty acids composition in general and conjugated linoleic acid (CLA) and the n-3 fatty acids in particular (Scollan et al. 2001).

A recent preliminary study concluded that the ruminal bacteria that form stearate were also highly sensitive to the toxic effects of LA (Chaudhary et al. 2004). Differential toxicity could be envisaged as a strategy for arresting biohydrogenation of vaccenic acid (trans-11–18:1), which would lead to ruminant products with a healthier fatty acid profile (Griinari et al. 2000). Although a few studies have examined the toxicity of saturated fatty acids, oleic acid and other lipids to ruminal bacteria (Henderson 1973; Marounek et al. 2002), none has related PUFA toxicity and biohydrogenation. The present study was undertaken to investigate the effects of LA and other PUFA on individual members of the ruminal microflora.

Materials and methods

Microorganisms and growth conditions

Bacteria and fungi were from the collection held at the Rowett Research Institute. The provenance of most of the bacteria has been given previously (Wallace and McKain 1991; Avguštin et al. 1997; Edwards et al. 2005). Butyrivibrio hungatei JK611 was a gift from J. Kopečný and is described by (2003). Neocallimastix frontalis RE1 was described by Stewart et al. (1987). Piromyces communis P was a gift from C.G. Orpin (Babraham, UK).

All transfers and incubations were carried out under O2-free CO2 and at 39°C in Bellco tubes (Bellco Biotechnology, Vineland, NJ 08380, US). Inoculum volumes were 5% (v/v) of a fresh culture. The media used in these experiments were the liquid form of M2 medium (Hobson 1969), which contains 70 mM sodium lactate, or the same medium with sodium lactate omitted (M2-L) Fatty acids were prepared as a separate solution, sonicated for 4 min in water and added to the medium before autoclaving.

In order to determine if bacteria metabolised LA (cis-9, cis-12–18:2), they were inoculated firstly into the liquid form of Bryant and Robinson (1961) medium containing 50 μg LA ml−1 and incubated for 48 h. Cultures were then extracted and analysed for free fatty acid concentrations. Where cultures failed to grow but grew on medium with no added LA, the culture was inoculated into medium containing 20 μg LA ml−1 and the process was repeated. Fermentation products were determined by derivatization and capillary gas-liquid chromatography (Richardson et al. 1989) of supernatants from cultures grown in the liquid form of M2 medium.

The metabolism of PUFA was determined using medium M2-L containing 50 μg PUFA ml−1 for all species except Anaerovibrio lipolytica5S and Veillonella parvula L59, which required the addition of 10 g l−1 glycerol or sodium pyruvate respectively to support growth. Growth of bacteria was measured from the increase in optical density (OD) at 650 nm of the control tubes, in triplicate, using a Novaspec II spectrophotometer (Amersham Biosciences, UK). Growth was followed turbidimetrically at 12-h intervals to determine the length of the lag period and the cell density at stationary phase. Similar experiments were carried out using M2 medium, except that only the effects of LA, at 200 and 50 μg ml−1, were tested and the growth measurement intervals were shorter. Where cultures failed to grow but grew on medium with no added LA, the culture was inoculated into medium containing 20 or 5 μg LA ml−1 and the process was repeated.

Ruminal fungi were grown in M2-L medium to which 0.05% agar had been added. Growth was assessed visually, and fatty acid concentration was determined on the total freeze-dried culture.

Measurement of cell integrity using propidium iodide

The methods used to assess the influence of LA on cell integrity of different species were based on the work of Ben Amor et al. (2002). One ml of overnight M2 culture was inoculated into 10 ml of M2 medium and incubated at 39°C until it reached mid-exponential phase (OD650 = 0.4, approx. 4 h). The bacterial culture was centrifuged (3,000 g, 10 min, 4°C) and the pellet was washed twice with anaerobic potassium phosphate buffer (100 mM; pH 7.0) containing 1 mM dithiothreitol (DTT). Anaerobic conditions were maintained by carrying out transfers in an anaerobic chamber, with a gas phase of 80% N2, 10% CO2 and 10% H2. Cells were resuspended to an OD650 of 0.4 in the same buffer, then 0.05 ml of 1 mg ml−1 LA solution, also in the same buffer and prepared by sonicating for 3 min, was added to 0.95 ml of cell suspension. The suspension was incubated at 39°C for 15 min. Cells resuspended in buffer alone and heat-treated cells (100°C for 15 min) served as control samples.

Propidium iodide (PI) working solution (1.5 mM) was prepared in distilled water and stored at 4°C in the dark. Fifty μl of each sample were added to 149 μl of anaerobic potassium phosphate buffer (100 mM, pH 7.0, containing 1 mM DTT) in the presence of 1 μl of PI solution. The mixtures were incubated for 15 min at 39°C in the anaerobic chamber. Samples were kept in the dark on ice and used within 45 min for fluorescence analysis. Fluorimetry measurements were made using a Spectramax GeminiXS spectrofluorimeter (Molecular Devices, Wokingham, Berks, UK) set at λ EX = 530 nm and λ EM = 620 nm.

Influence of LA and sodium lactate on growth of C. proteoclasticum P-18:

Clostridium proteoclasticum P-18 was grown overnight in medium M2-L. Fresh tubes of the following media were then inoculated (5%, v/v): M2; M2-L; M2+50 μg LA ml−1; M2-L+50 μg LA ml−1. Similar experiments were done with vaccenic acid (trans-11–18:1). Growth was followed turbidimetrically and samples for fatty acid analysis were prepared as before.

Fatty acid extraction and analysis

Extraction, derivatization of fatty acids and GC analysis of methyl esters were carried out using procedures described by Wąsowska et al. (2006).

Results

Metabolism of linoleic acid by individual species of ruminal bacteria

The metabolism of LA and its influence on the growth of 26 culture-collection strains was determined in Bryant and Robinson (1961) complete medium containing 50 μg ml−1 LA (Table 1). Eleven strains metabolised LA to an extent easily detectable by GC. Most of these were Butyrivibrio and related strains, which showed an accumulation of vaccenic acid (VA; trans-11–18:1), but Clostridium aminophilum and Mitsuokella multiacidus formed oleic acid (cis-9–18:1), Fibrobacter succinogenes appeared to form some C16:0, and Streptococcus bovis produced a C-18 hydroxy fatty acid product. Only C. proteoclasticum P-18 formed a CLA, cis-9,trans-11–18:2, as one of its end products. C. proteoclasticum B316Tand P-18 were the only bacteria to form stearate. The products of LA metabolism by Lachnospira multipara and Peptostreptococcus anaerobius were not determined; LA was used incompletely by these bacteria. No trans-10,cis-12–18:2 was produced by any culture-collection strain, including the two Megasphaera elsdenii. Five of the 26 bacteria suffered growth inhibition by 50 μg LA ml−1. Of these, three were butyrate producers, viz. B. hungatei, Clostridium sticklandii and Eubacterium ruminantium, and the other two were cellulolytic Ruminococcus spp.

Table 1 Fermentation products and ability of predominant species of ruminal bacteria to metabolize LA (50 μg ml−1)
Full size table

Influence of different PUFA on growth of predominant species of ruminal bacteria

Most species of ruminal bacteria grew in the presence of 50 μg ml−1 PUFA in M2-L medium (Table 2). In about half these species, there was no discernible lag phase induced by the LA, with a slight decrease in cell density being the only sign of toxicity. The most resistant species were members of the genera Prevotella, Megasphaera, Selenomonas, Veillonella and Anaerovibrio. The two cellulolytic Ruminococcus species were unable to grow in the presence of any PUFA in this medium, while the other cellulolytic species, F. succinogenes, grew to a much lower final cell density in LA and DHA and was unable to grow in medium containing LNA or EPA. The other most sensitive bacteria were B. hungatei, E. ruminantium and C. proteoclasticum. The relative sensitivity of the different species to other PUFA was generally similar to their sensitivity to LA. LNA was more toxic than LA, while the fish fatty acids, EPA and DHA, were more toxic than LNA. There was a tendency for EPA to be more toxic than DHA. LNA was metabolised by the same strains that had been found to metabolise LA. Significant metabolism of EPA and DHA was not observed in any culture.

Table 2 Influence of polyunsaturated fatty acids (50 μg ml−1) on growth of predominant species of ruminal bacteria in M2-L medium
Full size table

A third set of experiments was done with LA using M2 medium, which contains 70 mM sodium lactate (Table 3). Once again, members of the genera Prevotella, Megasphaera, Selenomonas, Veillonella and Anaerovibrio were insensitive to the fatty acids, and the most sensitive bacteria were the cellulolytic species, and B. hungatei and C. proteoclasticum. Ruminococcus flavefaciens remained unable to grow at 20 μg LA ml−1, but the other sensitive strains grew, albeit, especially with B. hungatei, after a long lag phase. B. hungatei was not inhibited by 5 μg LA ml−1 (data not shown). Comparing Tables 2 and 3, the presence of sodium lactate in the medium influenced the sensitivity of several species to LA. E. ruminantium, which was insensitive to LA in the presence of sodium lactate (Table 3), became highly sensitive in its absence (Table 2).

Table 3 Influence of linoleic acid on growth of predominant species of ruminal bacteria in M2 medium (containing 70 mM sodium lactate)
Full size table

Influence of PUFA on growth of ruminal anaerobic fungi

Two species of ruminal fungi were grown in M2-L medium to which 0.05% agar and 50 μg ml−1 PUFA had been added. At 96 h following a 5% inoculum, the only culture to have grown was N. frontalis in the LA-containing medium; no growth was observed with P. communis in the presence of fatty acids 1 week after a 10% inoculum had been added to the 50 μg ml−1 LA medium, both species grew. N. frontalis had metabolised about half of the LA in the medium, forming cis-9,trans-11–18:2. P. communis grew but did not metabolise LA.

Influence of LA on cell integrity of different species of ruminal bacteria

Fluorescence of PI was used to assess how LA affected the cell integrity of some of the most common species of ruminal bacteria (Fig. 1). PI fluoresces when in contact with DNA, but is normally unable to permeate across the cell membrane. Thus, only cells which have lost their cell membrane integrity fluoresce with PI. Fluorescence readings from treated cell suspensions were therefore compared with values obtained from the same culture which had been boiled for 10 min. F. succinogenes and B. fibrisolvens A38T had a high permeability in the absence of LA, while others gave 20% or less of the values obtained with boiled cells. LA increased the fluorescence of all bacteria. The increase differed with different species and even among the same species or closely related bacteria, such as the Prevotella spp., sensitivity to LA varied considerably. The least sensitive species was M. elsdenii. Correlation analysis was carried out by grouping the species in five groups according to sensitivity to LA in Table 2 or Table 3, then comparing with the cell integrity data from Fig. 1. Correlation coefficients of 0.52 and 0.36 were found between the effects of LA on cell integrity (Fig. 1) and effects of LA on growth in the absence (Table 2) or presence (Table 3) of sodium lactate, respectively.

Fig. 1
figure 1

Influence of LA (50 μg ml−1) on membrane integrity of different strains of ruminal bacteria. Loss of cell integrity was determined by fluorescence in the presence of propidium iodide. 100% loss of cell integrity was taken as the fluorescence of the same culture which had been incubated at 100 °C for 15 min. No addition \((\square)\), LA (■). Results are means and SD from 3 cultures, each of which was subject to 8 replicate measurements (n =  24)

Full size image

Influence of LA, VA and sodium lactate on growth of C. proteoclasticum P-18

Growth of C. proteoclasticum was more sensitive to the addition of PUFA than most other bacteria (Tables 1–3). The presence of sodium lactate in M2 medium had no influence on the growth of C. proteoclasticum, however it tripled the lag phase in the presence of 50 μg LA ml−1 (Fig. 2a). The cause appeared to be a prolonged accumulation of CLA (Fig. 2b). Any influence of sodium lactate on LA isomerisation during early lag phase was slight (Fig. 2c). Growth was also inhibited slightly by VA (Fig. 3a), but the result was a slight lowering of the specific growth rate rather than an obvious effect on the lag phase. Sodium lactate slowed the specific growth rate further with little influence on the lag phase (Fig. 3a) or the rate of metabolism of VA (Fig. 3b). Sodium chloride added to the medium at the same concentration as sodium lactate had no effect on growth (not shown).

Fig. 2
figure 2

Influence of sodium lactate on growth of C. proteoclasticum P-18 in the presence of 50 μg ml−1 LA. Medium M2 contains 70 mM sodium lactate, medium M2-L contains no added lactate. (a) Growth in M2-L (○), M2-L + LA \((\vartriangle)\), M2 (●), M2 + LA (▴); (b) fatty acid concentrations in M2-L medium (open symbols) and M2 medium (closed symbols) - LA (○, ●), cis-9,trans-11–18:2 (\(\square\), ■), trans-11–18:1 ( \(\vartriangle\), ▴), 18:0 ( \(\lozenge\), \(\blacklozenge\)); (c) fatty acid concentrations during initial phase in M2-L medium (open symbols) and M2 medium (closed symbols)—LA (○, ●), cis-9,trans-11–18:2 ( \(\square\), ■). Results are means and standard deviations from three cultures

Full size image
Fig. 3
figure 3

Influence of sodium lactate on growth of C. proteoclasticum P-18 in the presence of 50 μg ml−1 trans-11–18:1. (a) Growth in M2-L (○), M2-L + trans-11–18:1 ( \(\vartriangle\)), M2 + trans-11–18:1 (▴); (b) fatty acid concentrations: trans-11–18:1 (○,●), 18:0 ( \(\square\), ■), open symbols—M2-L medium, closed symbols—M2 medium. Results are means and standard deviations from three cultures

Full size image

Discussion

There has been much recent interest in PUFA, particularly n-3 PUFA, and intermediates of the biohydrogenation pathway from LA as possible health-enhancing metabolites in man (Pariza 2004). Thus, it is timely to revisit the microbial basis of this pathway in ruminal bacteria. Previous microbiological studies (Polan et al. 1964; Kemp et al. 1975; Hazlewood et al. 1976) were carried out before the significance of n-3 PUFA and CLA were known. They also used radiolabelled fatty acids at low concentrations (2 μg ml−1) that make comparing the activity and quantitative significance of different bacteria difficult. Here, higher concentrations of fatty acids were used with bacteria representative of the wide range of ruminal bacteria (Edwards et al. 2004), revealing patterns not only of metabolism but toxicity that are relevant to microbial ecology and nutrition. The concentrations used (up to 200 μg ml−1) were chosen for practical reasons, mainly to provide optically clear suspensions when sonicated. These were nevertheless lower than might occur transiently in ruminal digesta in cows grazing fresh grass, where the concentration might reach a peak of 3 mg ml−1 (Wąsowska et al. 2006). Other issues, such as fatty acid solubility and compartmentation of bacterial populations, mean that it is very difficult to predict the true concentration of PUFA that bacteria might actually encounter in vivo.

Several species metabolised LA, by at least four possible routes. The most common product was vaccenic acid (VA; trans-11–18:1), which was produced by three Butyrivibrio strains and two strains of C. proteoclasticum. Only C. proteoclasticum P-18 produced the CLA, cis-9,trans-11–18:2 in a 48-h incubation, along with VA and stearic acid. However, as VA is formed via cis-9,trans-11–18:2 as an intermediate (Harfoot and Hazlewood 1997), the bacteria forming VA might be considered CLA producers that had completed its conversion to VA. Thus, as C. proteoclasticum is related closely to Butyrivibrio (Kopečný et al. 2003; van de Vossenberg and Joblin 2003), these results are in agreement with the original results of Polan et al. (1964), who investigated more than 20 strains of ruminal bacteria. The most abundant CLA isomer usually found in milk is cis-9,trans-11–18:2 (Offer et al. 2001; Shingfield et al. 2003). The cis-9,trans-11 isomer is generally considered to be the main health-promoting CLA for human consumption (Pariza 2004). VA is also desirable as a product flowing from the rumen because VA acts as a substrate for the formation of cis-9,trans-11–18:2 in the animal’s own tissues (Griinari et al. 2000). Thus, the VA producers can be considered beneficial for producing ruminant meat and milk with a healthy fatty acid profile. The finding that N. frontalis produces CLA is the first published report of a fungal contribution to CLA formation; however, a preliminary report to this effect was presented by I.S. Nam and P.C. Garnsworthy at the 26th World Congress of the International Society for Fat Research, Prague, 25–28 September 2005. The activity of N. frontalis in forming CLA from LA was very small in comparison with B. fibrisolvens: about half a 50 μg ml−1 concentration of LA was converted to CLA after 96 h. A similar conversion by a culture of B. fibrisolvens takes a few minutes.

LA was metabolised in different ways by other bacteria, reflecting a diversity in the way in which different species deal with LA. Oleic acid (cis-9–18:1) was formed from LA by C. aminophilum and M. multiacidus. Ruminal oleic acid producers were not identified previously (Harfoot and Hazlewood 1997). F. succinogenes formed a product that appeared to be 16:0, though this was not confirmed by mass spectrometry. Presumptive hydroxy acids were formed by S. bovis, which would be consistent with the conclusions of Hudson et al. (2000).

Many isomers of C18:2 other than cis-9,trans-11–18:2 appear in digesta and milk too (Shingfield et al. 2003). The source of these is not known, either biochemically or microbiologically, nor are their likely effects on health. For example trans-11,cis-13–18:2, trans-7,cis-9–18:2 and trans-11,trans-13–18:2 were the next most abundant CLA isomers found in milk in the Shingfield et al. (2003) study. None of the cultures tested here formed trans-11,cis-13–18:2, trans-7,cis-9-18:2 or trans-11,trans-13–18:2; it is possible they are formed from cis-9,trans-11-CLA by species other than Butyrivibrio. The trans-10,cis-12 isomer is usually present at low concentrations in digesta and milk (Offer et al. 2001; Shingfield et al. 2003), yet it, or its derived trans-10–18:1 biohydrogenation product, can accumulate under certain dietary circumstances, causing milk fat depression, which can be an important problem in the dairy industry (Griinari and Bauman 1999). No isomerisation of LA to the trans-10,cis-12-CLA isomer was found in any of the bacteria investigated here. Bifidobacterium, Propionibacterium, Lactococcus, Streptococcus and Lactobacillus isolates from other habitats have been shown to form trans-10,cis-12-CLA (Jiang et al. 1998; Ando et al. 2004; Coakley et al. 2003). In the rumen, Kim et al. (2002) found that two strains of ruminal M. elsdenii produced thetrans-10,cis-12 isomer from LA. In contrast, neither of the M. elsdenii strains analysed here, which included the T81 strain tested by Kim et al. (2002), formed trans-10,cis-12-CLA. Our initial purchase of M. elsdenii T81 from ATCC produced trans-10,cis-12-CLA, but it was contaminated with Streptococcus gordonii; the negative test reported is from a pure strain supplied subsequently. Thus, the bacteria responsible for trans-10,cis-12-CLA formation in the rumen must be considered unknown at this time.

The present results therefore appear to confirm what was previously thought by others to be the dominant role of Butyrivibrio and related bacteria in fatty acid biohydrogenation, although some questions remain. In addition to the examples of CLA production by unknown organisms just mentioned, DHA and EPA are metabolized in ruminal digesta (AbuGhazaleh and Jenkins 2004; Wąsowska et al. 2006), yet none of the microorganisms tested here had any effect on their concentration. There are many species of bacteria that have been isolated from the rumen (Stewart et al. 1997) which were not screened here. Furthermore, those bacteria whose 16S rRNA genes have been recovered from ruminal contents but which have not yet been cultivated (Whitford et al. 1998; Tajima et al. 1999; Edwards et al. 2004) may be of significance in this respect. Studies are under way in our laboratories to try to correlate ruminal bacterial populations, particularly the Butyrivibrio group, with biohydrogenating activity of the mixed population. The protozoa have also been considered to be candidates for biohydrogenation because of their high content of CLA and VA, although no truly protozoal (i.e., not due to associated bacteria) biohydrogenation activity has been demonstrated (Devillard et al. 2006).

It is well known that fatty acids have bacteriostatic and bacteriocidal effects, indeed man has been utilising these properties for centuries. The present study compares for the first time the toxicity of PUFA across a range of the ruminal microflora, including the fungi. Henderson (1973) examined the effects of fatty acids on ruminal bacteria. A Butyrivibrio sp. was generally most sensitive to fatty acids, but only saturated and monoenoic acids were included in the study. Oleic acid was much more toxic than the saturated fatty acids. Marounek et al. (2002) found that C-12 and C-14 fatty acids were more toxic to ruminal and rabbit caecal bacteria than other chain lengths, but again the study was of saturated acids and oleic acid. In non-ruminal bacteria, LA and LNA were much more toxic than saturated or monoenoic acids (Galbraith et al. 1971). The present paper covers the effects of several PUFA on a range of ruminal bacterial and fungal species. It emerged that the most sensitive of all were cellulolytic bacteria and fungi, B. hungatei and the closely related C. proteoclasticum. Henderson (1973) also found that a Ruminococcus sp. was sensitive to fatty acids.

There are several possible reasons why unsaturated fatty acids are more toxic than saturated fatty acids. The double bonds alter the shape of the molecule, such that kinked unsaturated fatty acids disrupt the lipid bilayer structure (Keweloh and Heipeiper 1996). Whether multiple double bonds caused more toxicity because they are more sterically disruptive, or because it takes longer to biohydrogenate the whole molecule, is not clear from this study. Also it is not clear that the toxicity was necessarily a membrane effect. The extent to which cell integrity was disrupted by LA, as measured by PI fluorescence, in different species did not correlate with the inhibition of growth. An enhanced toxicity of PUFA in the presence of sodium lactate was found here in C. proteoclasticum. This was obvious from the growth curve (Fig. 2) even though it had not been apparent by comparing Tables 2 and 3 because of the timing of the observations. Similar increases in lag phases when LA and sodium lactate were combined were reported in other strains related to Butyrivibrio (McKain et al. 2004). Lactate appears to have a significant role in E. ruminantium too, effectively protecting the bacterium from the toxic effects of LA. White et al. (1970) found that one of the original ‘Fusocillus’ isolates hydrogenated oleic acid more extensively when grown in a lactate-containing medium. In common with Butyrivibrio, E. ruminantium and C. proteoclasticum are butyrate producers. Thus, there appears to be a metabolic link between lactate, butyrate metabolism and fatty acid biohydrogenation. Perhaps the different sensitivities are due to different mechanisms of butyrate production (Diez-Gonzalez et al. 1999). Lactate appeared to inhibit hydrogenation of CLA, but not VA, in C. proteoclasticum, indicating that these two reductase activities may be carried out by different enzymes and that only the former is affected by sodium lactate. It is worth noting that all the experiments were done here with growing bacteria and fungi. Stationary-phase bacteria were much less active in biohydrogenation than growing cells (Chaudhary et al. 2004).

Major advances have been made in achieving the desired changes in fatty acid content of meat and milk experimentally, via dietary manipulation in ruminants, generally by adding oils containing unsaturated fatty acids to the diet (Jenkins 1994; Offer et al. 2001; Scollan et al. 2001; Shingfield et al. 2003). Often, benefits can be attributed directly to the oils which were added to the diet escaping the rumen without being hydrogenated by ruminal microorganisms. However, other dietary changes, such as the inclusion of fish oil, seem to alter the fatty acid composition of the products via a direct effect on biohydrogenating activity in the rumen. This paper now indicates that the basis of these effects may be a differential toxicity of different PUFA to ruminal bacteria, particularly those in the Butyrivibrio group. It will be important to determine the mechanism by which PUFA exert their toxicity in C. proteoclasticum because selective suppression of this species might be expected to decrease the degree of saturation of fatty acids leaving the rumen and being incorporated into milk and meat. Furthermore, all the main species that comprise the ruminal cellulolytic flora, i.e., F. succinogenes, Ruminococcus spp. and the anaerobic fungi, appear vulnerable to inhibition by PUFA. Any strategy for manipulating biohydrogenation using PUFA would therefore have to take care that fibre digestion was not compromised.