Taxon‐specific associations between protozoal and methanogen populations in the rumen and a model rumen system

Methanogen populations in the rumen and in model rumen systems (operated over a 240-h period) were studied using the small subunit (SSU) rRNA phylogenetic framework for group-specific enumerations. Representatives of the family Methanobacteriaceae were the most abundant methanogen population in the rumen, accounting for 89.3% (± 1.02%) of total archaea in the rumen fluid and 99.2% (± 1.8%) in a protozoal fraction of rumen fluid. Their percentage of archaea in the model rumen systems declined from 84% (± 8.5%) to 54% (± 7.8%) after 48 h of operation, correlated with loss of protozoa from these systems. The Methanomicrobiales, encompassed by the families Methanomicrobiaceae, Methanocorpusculaceae, and Methanospirillaceae were the second most abundant population and accounted for 12.1% (± 2.15%) of total SSU rRNA in rumen fluid. Additionally this group was shown to be essentially free living, since only a negligible hybridization signal was detected with the ruminal protozoal fraction. This group constituted a more significant proportion of total archaea in whole rumen fluid, 12.1% (± 2.1%) and model rumen fluid containing no protozoa (26.3 ± 7.7%). In contrast, the Methanosarcinales, generally considered the second most abundant population of rumen methanogens, accounted for only 2.8% (± 0.3%) of total archaeal SSU rRNA in rumen fluid.


Introduction
Rumen methanogens participate in the terminal steps of the degradation of plant organic matter in the rumen primarily by the removal of hydrogen generated by fermentation.
Hydrogen removal promotes the more complete oxidation of fermented substrates and greater energy recovery by the fer-* Corresponding author. Tel.: +1 (847)  are therefore thought to be integral components of the microbial food chain in the rumen. The predominant species of methanogens so far identified by cultural enumeration are Methanobrevibacter ruminantim and Methanosarcina barkerii [2,3]. While techniques are available for culture-based enumeration of methanogens, a complication in some environments is their frequent and intimate association with protozoa. The methanogen-protozoa association is a common feature of many anaerobic systems, observed between hydrogenotrophic methanogens and bovine ruminal ciliates [4], ovine ruminal ciliates [5], termite hindgut flagellated protozoa [6], cockroach gut ciliate Nycutherus ovalis [7], termite gut protozoon Reticzrlitermes speratus [8], marine ciliate Metopus contortus [9], lakewater ciliate Metopus pch~ftivmis [lo] and Tri-rn~~emu sp. [l 11. This association is proposed to be of selective value to both organisms since it provides the capacity for interspecies hydrogen transfer [12]. Rumen protozoa depend upon a hydrogen evolving fermentation that provides substrate for the methanogens. The protozoa, in turn, benefit from hydrogen removal, since hydrogen is inhibitory to their metabolism if not removed.
We earlier developed a set of small subunit ribosomal RNA (SSU rRNA) probes for the major groups of methanogens [13,14] and have used them to study the population structure of natural and engineered communities [15]. We are currently using these and other group-specific probes to study the community structure of artificial rumens, using the Hoover dual flow continuous culture fermenters as a model system [ 16,171. A key feature of these systems is the loss of protozoa during long term system operation. Since the protozoa are hosts to many rumen methanogens, their loss from the model systems was hypothesized to alter the methanogen population structure. To evaluate this hypothesis, taxon specific probes for methanogens were used to quantify shifts in methanogen populations that accompanied loss of protozoa from the model rumens. This revealed a specific association between rumen protozoa and certain taxonomic groups of methanogens.

Collection of' rumen jluid
Rumen fluid was removed from the rumen of a multiparous Holstein cow fed a total mixed diet [Alfalfa hay (600 g), dry shelled corn (329 g), soybean meal (58 g), dicalcium phosphate (8 g), trace mineral salt (4 g), vitamin A, D and E mix (1 g) kgg'] twice daily for 14 days before sampling. The fluid was transferred to a prewarmed (38°C) container. filled to the top and covered to exclude air for transport to the laboratory. 7 7 Opewtion oj' twdel ~'~uw~~.s __1_ The original model rumen system (dual flow continuous culture system) and the modified version have been previously described [16.17]. This system was designed to simulate differential solid-liquid phase removal rates that occur in the rumen. Each vessel was supplied with feed (11 g) every 3 h [alfalfa hay (600 g), dry shelled corn (329 g), soybean meal (58 g), dicalcium phosphate (8 g), trace mineral salt (4 g). vitamin A, D and E mix (1 g) kg-'] by an automatic feeding device. The pH was maintained at 6.25 by a pH sensor and infusion pumps delivering sodium hydroxide or hydrochloric acid. The temperature of the system was kept at 38°C using a thermocoupled probe to monitor temperature. Four model systems were operated for 240 h. Samples were removed for nucleic acid extraction at 0, 6, I?. 24, 48, 72, 96. 120, 168 and 240 h after inoculation by graduated pipette, placed in ?-ml screw-cap polypropylene tubes (Sarstedt. Inc., Newton. NC) containing 500 mg of zirconium beads, and immediately frozen on dry ice.
One liter of rumen fluid was transferred to a prewarmed (38°C) separative funnel previously flushed with nitrogen gas to maintain anaerobiosis. After 15 min of incubation, the protozoa had settled to form a milky layer at the bottom of the funnel. Excellent fractionation was confirmed by microscopic inspection and by hybridization (below). Aliquots of the protozoa1 fraction (0.5 ml) were sampled and immediately frozen in 2.0-ml screw-cap tubes containing 500 mg of zirconium beads on dry ice. Samples of rumen fluid were similarly frozen for subsequent analysis.

Ertraction oj' nucleic acid
Total nucleic acids were recovered from the samples using a slight modification of a published procedure [18]. Mechanical disruption on a reciprocating shaker (Beadbeater, Biospec Products, Bartelsville. OK) with zirconium beads (0.1 mm) was employed to extract total nucleic acid from about 1 g of sample. Approximately 1 g beads was used with each sample, together with 50 cl1 of 20% SDS (w/v). 500 l.tl of phenol equilibrated with 50 mM sodium acetate-l0 mM EDTA buffer (pH 5.1). The samples were reciprocated for 3 min at room temperature and transferred to a 60°C waterbath for 10 min before a further 3-min period of reciprocation. Samples were extracted again with bufferequilibrated phenol before extracting twice with phenol/chloroform (4: 1) and twice with chloroform. Total nucleic acid was precipitated with ammonium acetate (2 M final concentration) and two volumes of ethanol. Concentrations of the recovered nucleic acid were measured spectrophotometrically, assuming that 1 mg of RNA per ml was equal to 20 optical density units at a wavelength of 260 nm.

Oligonucleotide probe hybridization
The RNAs extracted from samples and reference organisms were diluted and applied in triplicate to Magna Charge Nylon membranes as previously described [ 151. Established hybridization conditions were used for three group-specific probes for methanogens, a universal probe (S-Univ-1392-a-A-15) and probes for each of the three domains, Archaea (

Results and discussion
The quantification of natural microbial populations is dependent on representative sampling and unbiased enumeration.
Of these two requirements. unbiased enumeration is the more fundamental since representative sampling cannot be established without it. Although the utility of culture based techniques for enumeration is well established, they are not generally feasible for studies encompassing a wide variety of physiological types of differing abundance and including a large number of samples [18]. For example, in the absence of a selective medium, culture-based enumeration of low abundance populations is difficult [18]. Thus, a fundamental utility of molecular methods is the ability to establish general conditions for enumerating the full spectrum of microbial diversity. In this regard, key issues of nucleic acid extraction efficiency from different populations, and the stringency and precision of hybridization, have been established in previous studies [l, 191. Also, the phylogenetic framework provides an additional basis for evaluation of these data. A key point of validation is consistency between quantification using general and specific probes. For example, in this study domain probe summation (the total hybridization response from Archaea, Eucarya and Bacteria) is generally within 10% of total SSU rRNA measured using the universal probe. This internal consistency provides essential support for data interpretation.

I. Domain abundance
Domain representation is expressed as a percentage of the total SSU rRNA in each sample quantified using the universal probe (S-Univ-1392-a-A-15). As noted. we routinely determine domain summation for each sample as a key element of experimental validation. For example. we have observed that partial sample degradation results in greatly elevated domain summation as a consequence of the sensitivity of the universal probe target site to nuclease digestion [l]. In this study, the sum of the domain fraction, compared to 57.7 ? 3.2% in the rumen fluid (Fig. 1). Bacteria were not visible in this fraction using microscopic examination. suggesting that much of the bacterial signal represented bacteria ingested by, or associated with. protozoa.
The relative abundance of eucaryotic SSU rRNA in the model rumens rapidly declined following inoculation (Fig. 2) as the eucaryotic biomass was lost from the systems (Fig. 2). Since methanogens are the only Archaea that have been isolated from either the rumen or the large intestine of non-ruminants.
we anticipated that archaeal abundance would correspond to the abundance of methanogens.
Thus. the sum of the methanogen groups quantified with the individual methanogen probes should equal the value obtained by the archaeal probe. Fig. 3    micicum. Methanobrevibacter smithii and Methwobacterium autotrophicum [7,8,25]. The Methanobacteriaceae was the only probe target group of methanogens associated with the protozoa1 fraction, suggesting a preferred association between members of this family and ruminal protozoa.
The second most abundant group of methanogens were representatives of the order Methanomicrobiales. Methanogens associated with this probe-target group (Methanomicrobiaceae, Methanocorpusculaceae, and Methanospirillaceae) accounted for 12.1 ? 2.15% of the total archaeal rRNA in rumen fluid and 0.05 ?O.Ol% in the protozoa1 fraction (Fig. 3). Metllntlomicrobiuttl mobilis represents the only methanogen so far isolated from the rumen which would fall within this group [26]. The Methanosarcinales accounted for only 2.8 ? 0.32% of archaea in the protozoa1 fraction and 2.75 ? 0.3 1% in the rumen fluid. Although Methanosarcinrr isolates similar to Ms. burkerii are generally thought to be the second most significant population of rumen methanogens, after Methanobrevibacter txminuntium [2], our findings suggest they are much less abundant than the Methanomicrobiales.
The absence of Methanomicrobiales (Methanomicrobiaceae, Methanocorpusculaceae and Methanospirillaceae) from the protozoa1 fraction was a striking observation.
Although a significant proportion was observed in rumen fluid and model rumen fluid, virtually none was protozoa1 associated. This indicates that ruminal representatives were primarily free living in this study animal.
The association of the rumen protozoa with members of the Methanobacteriaceae is different from specificities observed in some other environments. The marine ciliate Metopus contortus together with TzYrnyemz cornpresszrm and Plrrgiop?du fioututu have been shown to form intracellular associations with close relatives of Metlzurzoco~pzncttlrlrtl purvrmt and Metizurzoczdlez~s r~zazkigri, respectively, as their associated methanogens using rRNA-based approaches [9,27]. Culture-based techniques have been used to isolate Methrzoplanus endosymbionts from Metopzu contortus [28]. These species are representatives of the families Methanocorpusculaceae and Methanomicrobiaceae within the order Methanomicrobiales. In contrast, our results show that members of this order do not significantly associate with the rumen protozoa. Examples of Methanobacteriaceae association with protozoa are less numerous; ribosomal rDNA closely related to Metlzunuhczcterirrn? hrymtii has been isolated from the protozoon Metopm puluefhwzis [27].
The rumen harbors a complex protozoan community, including flagellates, chytrids, and ciliates. The community of ciliates constitute the most conspicuous component of the rumen, the entodinio- . Given this diversity it is surprising that the rumen protozoa show the same general specificity for methanogenic symbionts.
3.5. The cornposition oj rnethunogerz popuhtiorzs in tlze model mnze3z .s?~stenu The abundance of each population of methanogens expressed as a proportion of the total archaeal rRNA in individual model rumens is shown in Figs. 4-6. The composition of the methanogen populations in the inoculant (time 0) used for the model rumens was, as expected, very similar to the rumen fluid samples. However. after 6 h of operation, it was apparent that the methanogen populations had been perturbed. Over the 240-h period of operation, the relative abundance of Methanobacteriaceae declined from 84.2 k 3.6% to 54.9 t 3.95% of total archaea. This was most apparent in the first 48 h of operation. the period of most rapid loss of rumen protozoa, and was consistent with a protozoa1 association.
The relative abundance of a subset of the Methanomicrobiales increased from 9.6 + 3.5% to 26.3 It 7.7% (Fig. 5). A relative increase in other populations was not observed. Thus, this group may R. Slurp et al. I FEMS Microbiology Ecology 3 (1998) 71-78 71 have replaced those methanogens lost via protozoa1 displacement.
Methanosarcinales increased in abundance shortly after inoculation into the model systems, from 3.4 f 2.02% to 11.9 5 5.13%. However, they returned to initial abundance after 72 h. Although there was much variation between the individual model systems during the first 48 h, they were remarkably consistent at later times. We suggest that their transient increase reflects the nutritional versatility of this group. Methanosarcina barkerii isolated from the rumen contents of both cows and sheep uses a variety of substrates including methanol and methyl amines [30]. Most Methanosarcina species utilize acetate as a carbon and energy source [31].
The use of molecular methods provided by comparative rDNA sequencing has greatly facilitated the accurate identification and classification of endosymbionts and their hosts. This in turn reveals important information about the diversity and evolution of symbioses. Of the four major lineages of methanogens (Methanobacteriales, Methanomicrobiales, Methanosarcinales, and Methanococcales), characterized endosymbionts are drawn from throughout the order Methanomicrobiales and from the family Methanobacteriaceae within the order Methanobacteriales. The results obtained in these studies suggested the importance of Methanobacteriaceae as symbionts of rumen protozoa. A more specific set of phylogenetic probes for the Methanobacteriaceae will be used in future studies to more fully resolve the character of these symbioses.
Syntrophic and competitive relationships are key determinants of microbial community structure in the rumen. The techniques used for this study could be equally valuable for evaluating various of these key microbial relationships. Since the nutrition of the host animal is largely determined by the products of the overall metabolism of this community (volatile fatty acids. bacterial protein and vitamins). the metabolic interactions that define the microbial community and its collective activities also determine host nutrition and production. We anticipate that information provided by the use of molecular techniques combined with established culture based enumerations and rumen fermentation characteristics will help resolve the contribution of microbial inter-actions to overall ruminal metabolism and host production.