Use of community genome arrays (CGAs) to assess the effects of Acacia angustissima on rumen ecology

Denis O. Krause1,2, Wendy J. M. Smith1 and Christopher S. McSweeney1

1 CSIRO Livestock Industries, St Lucia, Australia
2 Department of Animal Science, Faculty of Agricultural and Food Sciences, University of Manitoba–Winnipeg, MB, Canada R3T 2N2

Correspondence
Denis Krause
Denis_Krause{at}umanitoba.ca


   ABSTRACT
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
REFERENCES
 
This research developed a community genome array (CGA) to assess the effects of Acacia angustissima on rumen microbiology. A. angustissima produces non-protein amino acids as well as tannins, which may be toxic to animals, and CGA was used to assess the effects of this plant on the ecology of the rumen. CGAs were developed using a 7·5 cmx2·5 cm nylon membrane format that included up to 96 bacterial genomes. It was possible to separately hybridize large numbers of membranes at once using this mini-membrane format. Pair-wise cross-hybridization experiments were conducted to determine the degree of cross-hybridization between strains; cross-hybridization occurred between strains of the same species, but little cross-reactivity was observed among different species. CGAs were successfully used to survey the microbial communities of animals consuming an A. angustissima containing diet but quantification was not precise. To properly quantify and validate the CGA, Fibrobacter and Ruminococcus populations were independently assessed using 16S rDNA probes to extracted rRNA. The CGA detected an increase in these populations as acacia increased in the diet, which was confirmed by rRNA analysis. There was a great deal of variation among strains of the same species in how they responded to A. angustissima. However, in general Selenomonas strains tended to be resistant to the tannins in the acacia while Butyrivibrio fibrisolvens was sensitive. On the other hand some species, like streptococci, varied. Streptococcus bovis-like strains were sensitive to an increase in acacia in the diet while Streptococcus gallolyticus-like strains were resistant. Strep. gallolyticus has independently been shown to be resistant to tannins. It is concluded that there is significant variation in tannin resistance between strains of the same species. This implies that there are specific molecular mechanisms at play that are independent of the phylogenetic position of the organism.


Abbreviations: CGA, community genome array; VFA, volatile fatty acid


   INTRODUCTION
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
REFERENCES
 
Rumen microbiology has been limited by the inability to unambiguously identify and quantify microbial populations, but molecular techniques that do not rely on cultivation have changed this (Hugenholtz, 2002; Blaut et al., 2002). Nucleic acid probes can be used to identify specific catabolic functions (Sayler et al., 2001), and resolve population differences using small-subunit (SSU) rRNA (Ziemer et al., 2002; Forster et al., 1998). There is, however, a need for detection systems that can accommodate large numbers of rumen samples, but are quick, cheap and robust, yet not bedevilled by the PCR bias so prevalent with environmental or rumen samples (Bonnet et al., 2002).

Acacia spp. are regarded as multi-purpose trees, and can be used as fodder plants (Hove et al., 2001) to supplement nitrogen in grazing ruminants (El-Hassan et al., 2000). However, acacias contain at least two classes of toxic compounds: tannins (Smith et al., 2001) and non-protein amino acids (McSweeney et al., 2000; Odenyo et al., 1997). Tannins are polyphenolic compounds that inhibit many ruminal bacteria and the amino acids may induce systemic toxicity in animals (Khanbabaee & van Ree, 2001). The effects of multi-purpose trees on the ecology of the rumen are of interest, because of the potential of micro-organisms to metabolize these compounds and reduce their toxic effects.

Microarray technology is becoming popular because up to 10 000 probes can be immobilized on a glass support, making high-throughput processing possible when a large number of targets or samples are being evaluated (Churchill, 2002; Wilson et al., 2002). Arrays have principally been used to determine microbial species identity (Cho & Tiedje, 2002), and to profile gene expression by micro-organisms (Cho et al., 2002; Firoved & Deretic, 2003).

In this paper, we describe the development and application of a community genome array (CGA) approach based on the immobilization of whole rumen bacterial chromosomes (the probes) on a nylon support. This approach is similar to the reverse genome probing methodology previously developed (Voordouw et al., 1991, 1993) but modifications to experimental approach, quantification methodology and statistical analysis have been made (Fig. 1). CGAs were applied to rumen samples obtained from sheep that had been consuming tannin-containing diets consisting of Acacia angustissima to assess the effects of this plant on rumen microbial populations. These studies were also validated with Northern analysis of key microbial species. We have also incorporated extensive mixed-model statistical analyses to account for multiple random and fixed effects. In this paper, we will refer to DNA bound to the membrane as the probe, and to the rumen sample as the target, or sample.



View larger version (34K):
[in this window]
[in a new window]
 
Fig. 1. Diagrammatic representation of the experimental approach developed.

 

   METHODS
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
REFERENCES
 
Animal experiments.
Animal trials were conducted with rumen-cannulated Merino wethers (27–34 kg live weight) that were housed in individual metabolism crates, fed once a day, at 8 a.m., to 90 % of ad libitum intake, and had unrestricted access to water. Animals were fed a basal diet that consisted of a 50 : 50 blend of 350 g Rhodes grass (Chloris gayana) and 350 g lucerne (Medicago sativa). A. angustissima was substituted for lucerne at a rate of 50 g per day as the concentration of Acacia in the ration increased. Animals were supplemented with 50 g per day of a mineral-vitamin supplement (Siromin; Ridleys).

Feeding trials were conducted longitudinally, with three sheep fed four diets consecutively. Diets were as follows: (1) establishment of baseline, 50 % Rhodes grass plus 50 % lucerne for a period of 30 days (RL1); (2) step-wise adaptation (13 days) to A. angustissima by substituting 50 g per day of lucerne for Acacia up to 37 % (RLA); (3) RLA plus 60 g per day polyethylene glycol (PEG; mol. mass 4000) (RLAP) for 10 days; and (4) back onto basal diet of 50 % Rhodes grass plus 50 % lucerne for 10 days (RL2). PEG was added because it tends to reverse the effects of tannins. Rumen samples were taken on three consecutive days at the end of each feeding period (total of 12 samplings, from each of four periods). Samples were taken 2 h after each morning feed and stored at –70 °C. Stepwise adaptation to the Acacia was followed because of reports of its toxicity in the literature (Odenyo et al., 1997). The Acacia plants used are not native to Australia, had to be grown under quarantine conditions, were not allowed to flower, and all residues had to be incinerated after the experiment. There was therefore a lack of material for follow-up experiments and there was not sufficient material to feed sheep for more than 10 days in each period, except for the adaptation period when animals were fed for 13 days.

Microbial strains.
All microbial strains were from our own laboratory collection.

Medium.
Bacterial growth medium composition was (per litre): 5 g cellobiose, 5 g casitone, 150 ml mineral solution A (contents per 100 ml: 3 g K2HPO4.3H2O), 150 ml minerals solution B (contents per 100 ml: 3 g KH2PO4, 6 g (NH4)2SO4, 6 g NaCl, 1·23 g MgSO4.7H2O, 1·58 g CaCl2.H2O), 2 ml trace mineral salts (contents per 100 ml: 0·5 mg ZnSO4.7H2O, 0·15 mg MnCl2.4H2O, 1·5 mg H3BO3, 1·0 mg CoCl2.6H2O, 0·1 mg NiCl2.H20, 0·15 mg Na2MO4.2H2O, 7·5 mg FeCl2.4H2O), 3·1 ml volatile fatty acid (VFA) solution (contents per 100 ml: 0·68 ml acetic acid, 0·3 ml propionic acid, 0·18 ml butyric acid, 0·05 ml isobutyric acid, 0·06 ml methylbutyric acid, 0·06 ml valeric acid, 0·06 ml isovaleric acid, 0·1 g phenylacetic acid), 2 g yeast extract, 1 g L-cysteine.HCl, and 0·01 % resazurin. Media were prepared anaerobically according to published methods (Bryant, 1972). The anaerobic gas was a 95 % CO2/5 % H2 mix, and 4 g Na2CO3 l–1 was included to buffer the medium at pH 6·7. Aliquots (9 ml) of anaerobically prepared medium were dispensed into 25 ml Balch tubes (18 mmx250 mm) inside an anaerobic cabinet (Coy Laboratories), stoppered and autoclaved for 15 min at 100 kPa.

DNA and RNA extraction.
DNA and RNA were extracted using a mechanical lysis method adapted to rumen samples which lyses more than 95 % of cells (Krause et al., 2001a). The main difference between the extraction of DNA and RNA was the composition and pH of the buffers used. In brief, DNA/RNA extraction was as follows: (a) a rumen sample was extracted three separate times to minimize extraction bias; (b) quality and quantity of nucleic acids were determined by electrophoresis (RNA with a formamide gel, DNA an agarose gel), and pooled if of sufficient quality and quantity; (c) each pooled sample was arrayed in triplicate on three separate membranes (see below). If the RNA, or DNA, was not of sufficient quality another aliquot was extracted. Cells from bacterial cultures were lysed using an enzymic method, instead of mechanical lysis (Krause et al., 2001a).

Fabrication of microarrays.
Chromosomal DNA from each bacterial strain, or rumen sample, was diluted 50 : 50 with sterile distilled water containing 0·0002 % bromophenol blue, so that once arrayed, the DNA could easily be located on the membrane. The concentration of DNA was 1 µg µl–1 so that once diluted the final concentration was 500 ng µl–1. Using this concentration, it was possible to array 1 µl droplets of DNA in a 96 (16x6) gridded format on a positively charged nylon membrane (Roche Diagnostics), the size of a standard microscope slide (7·5 cmx2·5 cm). To speed up manufacture of membranes, DNA was aliquoted into 384-well plates and arrayed onto the nylon using a 384-solid-pin multi-block replicator (V&P Scientific). The small size of the membrane allowed up to 44 individual membranes to be processed at once in a rotary hybridization oven (Thermo Hybaid).

Hybridization conditions for arrays.
The DNA-containing membranes were baked at 120 °C for 30 min to cross-link the DNA to the nylon. A prehybridization solution [25 % formamide, 5x SSC, 2 % blocking reagent (Roche Diagnostics), 2 % SDS, 0·1 % N-lauroylsarcosine] was incubated with the membranes for at least 2 h before the addition of the digoxigenin-labelled probe (10 ng ml–1). Probes were allowed to hybridize overnight and membranes were then washed twice, for 1 h, at 68 °C with 5x SSC. Membranes were subsequently processed according to the manufacturer's instructions (Roche Diagnostics). The amount of DNA on the membrane was determined by quantifying the 16S rRNA genes using a universal oligonucleotide probe and a standard curve of Escherichia coli DH5{alpha} DNA.

Hybridization conditions for Northern blots.
Prehybridization for 16S rRNA was done under the same conditions as described above for DNA arrays. Probes used, and the experimental conditions, have previously been described (Krause et al., 1999). Each sample was blotted in triplicate on each of three membranes. A dilution series of the reference organism was included on each membrane with the samples, or alternatively, the dilution series was carried out on a membrane with only selected samples and then cross-referenced to the master membrane.

Cross-hybridization assays.
The bacteria included in the array are underlined in Fig. 2. With this hybridization approach it is important to determine the degree of cross-hybridization between the genomes of bacteria that are to make up the master array. Each candidate organism to be included on the array was cross-hybridized to all other candidates. The degree of cross-hybridization was determined by measurement of the hybridization signal obtained with non-target organisms and expressed as a percentage of self-hybridization (100 %) (Table 1). The diagonal gives the minimum amount of DNA detected under assay conditions.



View larger version (36K):
[in this window]
[in a new window]
 
Fig. 2. Phylogenetic relationships between rumen bacteria included on the array and close relatives. Relationships based on 16S rDNA sequences. Organisms included on the array are underlined; those also in bold are bacteria which have been isolated in association with tannins.

 

View this table:
[in this window]
[in a new window]
 
Table 1. Cross-hybridization and minimal detection of bacterial chromosomes on array under assay conditions

The diagonal is the minimum quantity of DNA detected (pg) under standard assay conditions. The numbers below the diagonal indicate the percentage cross-hybridization. No number means that no cross-hybridization was detected under the assay conditions.

 
DNA labelling.
Chromosomal DNA was denatured at 95 °C for 10 min and immediately placed on ice. Digoxigenin-11-dUTP was incorporated into chromosomal DNA by addition of random primers which were elongated using Klenow polymerase (DIG random primed DNA labelling kit; Roche). Oligonucleotides were labelled with terminal transferase by addition of a mixture of DIG-11-dUTP and dATP to the 3' end of the oligonucleotide (DIG oligonucleotide tailing kit; Roche). DIG-labelled probes that hybridized to the target sequence were detected with an alkaline-phosphatase-labelled anti-DIG antibody. Membranes were incubated with CDP-Star (Roche) and the resultant chemiluminescence was detected by exposure to X-ray film. The intensity of the signal was measured with a Gel Doc 2000 system (Bio-Rad).

Measurement of pH and VFAs.
Rumen fluid was obtained by filtering rumen contents through a 50 µm polyester sieve. A 15 ml portion of filtered rumen fluid was mixed with 300 µl 4 % mercuric chloride, and centrifuged at 20 000 g for 20 min. The supernatant was frozen at –20 °C until ready for analysis of pH and VFA (Bush et al., 1979).

Linear statistical modelling.
Although similar hybridization profiles within a probe/sample combination were observed, the signal intensities between replicate arrays and Northern blots varied considerably. Thus, an analysis-of-variance (ANOVA) model was developed to account for the observed variability between replicate arrays, so that we could statistically evaluate the effects of probe–sample hybridization that would also accommodate variation within arrays and within Northern blots, and between replicates and treatments. For this application, the linear (ANOVA) statistical model included fixed and random terms (Tables 2 and 3). The model is a direct reflection of the experimental design, which featured random effects such as membrane, and fixed effects such as bacterium. Table 2 lists the factors with levels and types that were considered in this study. The full mixed-effects linear model is listed in Table 3. The model was analysed by using the ANOVA capabilities in the Statistica statistical analysis software package (StatSoft).


View this table:
[in this window]
[in a new window]
 
Table 2. Experimental factors affecting array signal intensities

 

View this table:
[in this window]
[in a new window]
 
Table 3. Full ANOVA mixed model containing fixed and random effects

 

   RESULTS
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
REFERENCES
 
Fabrication of microarrays
Initial array fabrication experiments were conducted using a slot-blotter (Hoefer) that produced a membrane (11·5 cmx8 cm) with 48 slots. It was thought that significant separation between slots would be necessary to enable accurate imaging of the autoradiograph resulting from this membrane. Subsequent experimentation demonstrated that a 1 µl DNA-containing droplet could be arrayed onto a membrane with approximately 5 mm separation, yet still produce an autoradiographic signal that was distinct. An advantage of using the small (7·5 cmx2·5 cm) membrane was that the amount of background hybridization was very small; the less the non-specific background, the higher the quality of the signal quantification.

Cross-hybridization experiments
Each bacterial chromosome was labelled and cross-hybridized to all other bacterial genomes potentially to be included on the array (Table 1). The extent of hybridization was quantified with a reference series of labelled DNA and cross-hybridization was expressed as a percentage of self-hybridization (100 %). The minimum amount of DNA detectable is also given in Table 1; it was within an order of magnitude for most bacterial strains, and indicates that low levels of cross-hybridization due to low levels of DNA detectability were not a confounding factor. It should be noted that the amount of detectable DNA reported was that under the experimental conditions used; smaller quantities of DNA could be detected under less stringent conditions, or longer exposure to X-ray, but this led to lower specificity, or more background signal. As expected, there was no cross-hybridization between genomes of separate species, but there was cross-hybridization between strains of the same species such as Butyrivibrio fibrisolvens and Selenomonas ruminantium. The cross-hybridization matrix could essentially be predicted from the degree of sequence similarity between 16S rRNA genes (Fig. 2).

We concluded that the sensitivity of the cross-hybridization methodology was robust because initial experiments revealed cross-hybridization between phylogenetically distantly related strains (e.g. Streptococcus bovis JB1 and Prevotella bryantii B14) which were the result of contamination. P. bryantii B14 was restreaked and picking of discrete colonies revealed that this culture had been contaminated with Strep. bovis JB1. This occurred with several other cultures and purification proved to be extremely time-consuming. However, once the chromosomal DNA had been prepared it could be stored for long periods (up to 2 years) without deterioration. We assessed deterioration of the DNA by evaluating the extent of fragmentation (smearing) on an agarose gel. Additionally, we were able to prepare up to 100 membranes at a time using the small-membrane format and these could be stored at –20 °C for up to 2 years.

Quantification of array experiments
Each membrane included the chromosomal DNA of the sample to which it was hybridized to provide a self-hybridization control; using our definition, rumen-extracted DNA was both a sample and a probe. Self-hybridization was the benchmark to which individual strains were compared, and quantitative results were expressed as a proportion thereof. The variation for each probe/sample combination was compared across the membrane triplicate as well as triplicate membranes (nine spots per probe). If the variation was greater than 5 %, the experiment was repeated. We found this to be an essential element in controlling experimental variation and ensuring data quality. Our approach to controlling experimental variation is borne out by the results of the mixed-model ANOVA analysis because membrane and extraction random effects were not significant (Table 3).

The response of individual strains of bacteria varied widely depending on diet (Fig. 3). In some cases the bacterial strains were sensitive to the presence of Acacia in the diet (e.g. B. fibrisolvens strains), while others appeared to be quite resistant (e.g. S. ruminantium strains). Individual strains within the same species (e.g. B. fibrisolvens and S. ruminantium) reacted differently to the change in diet, indicating within-species variation in phenotype with respect to the diets. The arrays also included Ruminococcus albus AR67, Ruminococcus flavefaciens Y1 and Fibrobacter succinogenes S85 (Fig. 3). These organisms increased in abundance as Acacia increased in the diet, which we found surprising, because our previous work with Calliandra callothyrsus, another tannin-containing plant, resulted in a decline in these species (McSweeney et al., 2001).



View larger version (33K):
[in this window]
[in a new window]
 
Fig. 3. Relative abundances of bacteria in the rumens of sheep on the different diets, determined by array analysis. Abundance is the percentage of a strain within a rumen sample. Treatments within bacterial strain are significantly different (P<0·05) if bars do not have common superscripts. Succ. dextrinosolvens, Succinivibrio dextrinosolvens; see Fig. 2 for full names of other species.

 
16S rRNA quantification of cellulolytic populations
As a means of validating the data from the arrays the unexpected increase in abundance of Ruminococcus and Fibrobacter was further investigated. Validation was accomplished by extracting rRNA from ruminal samples and hybridization with 16S rRNA-directed oligonucleotide probes to Fibrobacter and Ruminococcus. These hybridization experiments indicated that Ruminococcus (Fig. 4a) increased from 4·4 to 9·5 % (P<0·05), and F. succinogenes (Fig. 4b) from 1·7 to 5·6 % (P<0·05) when expressed as a proportion of the total bacterial population in the presence of Acacia. To test the hypothesis that tannins result in a decrease in the total bacterial population, thus resulting in an apparent increase in Ruminococcus and F. succinogenes, we expressed these species as a proportion of the total microbial population. Using this measure Ruminococcus increased from 1·3 % to 5·1 % (P<0·05) (Fig. 4a) of the total microbial population and F. succinogenes rose from 0·8 to 3·0 % (P<0·05) of the total microbial population (Fig. 4b). However, Ruminococcus declined (P<0·05) as a proportion of total microbial population when PEG was added to the diet (Fig. 4a). Expressing the bacterial population as a proportion of the total microbial population (Fig. 5) indicated that bacteria declined in the presence of tannins, but this was not significant (P>0·05).



View larger version (19K):
[in this window]
[in a new window]
 
Fig. 4. Relative abundance of Ruminococcus (a) and Fibrobacter (b) expressed as a percentage of total bacterial 16S rRNA (filled bars), or as a percentage of total community rRNA (open bars). Dietary treatments were: RL1, Rhodes grass+lucerne; RLA, Rhodes grass+lucerne+37 % Acacia; RLAP, RLA+PEG; and RL2, Rhodes grass+lucerne. Treatments are significantly different (P<0·05) if bars do not have common superscripts.

 


View larger version (12K):
[in this window]
[in a new window]
 
Fig. 5. Relative abundance of bacterial 16S rRNA expressed as a proportion of the total 16/18S rRNA. Dietary treatments were: RL1, Rhodes grass+lucerne; RLA, Rhodes grass+lucerne+37% Acacia; RLAP, RLA+PEG; RL2, Rhodes grass+lucerne.

 
Measurement of fermentation parameters
There was a decrease (P<0·05) in the acetate : propionate ratio when Acacia was added to the diet, which was reversed in the presence of PEG (Fig. 6a). There was also a statistically significant (P<0·05) decline in the branched-chain to short-chain VFA ratio when Acacia was included in the diet; this response disappeared when PEG was included (Fig. 6b). There was little change in the pH across treatments except when PEG was added to the diet, which resulted in a significant (P<0·05) decline in ruminal pH (Fig. 6c).



View larger version (21K):
[in this window]
[in a new window]
 
Fig. 6. Measurement of ruminal acetate : propionate ratio (a), branched-chain to straight-chain VFA ratio (b) and pH. Dietary treatments were: RL1, Rhodes grass+lucerne; RLA, Rhodes grass+lucerne+37 % Acacia; RLAP, RLA+PEG; RL2, Rhodes grass+lucerne. Means are indicated by the small squares, standard errors by the rectangles and standard deviations by the error bars.

 
Animal experiments and toxicity
In the animal experiment, it was essential to monitor the sheep very carefully, given the reports of toxicity in ruminants with this plant (Odenyo et al., 1997). Therefore, during the adaptation period in which lucerne hay was substituted by Acacia, very careful observations were made of feed intake and behaviour. Indications of an onset of intoxication would have been a reduction in feed intake, hyperactivity or lethargy, none of which occurred. We also observed no evidence of feed selection against Acacia and the whole meal was consumed by the animals. In an ideally designed experiment, all the feeding periods would have been equal but as the amount of material that could be grown was restricted, shorter feeding periods had to be used.


   DISCUSSION
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
REFERENCES
 
Microarray approaches in ecological microbiology are usually based on the rRNA genes (Wilson et al., 2002; Rudi et al., 2002) but rRNA is too conserved to provide good resolution at the species and subspecies level (Fox et al., 1992; Krause et al., 1999; Stackebrandt & Goebel, 1994). The relationship between 16S rRNA gene similarity and percentage DNA–DNA reassociation is logarithmic, and sequence similarity within a species (>70 % DNA relatedness) is normally greater than 97·5 % (Devereux et al., 1990). Studies comparing 16S rDNA sequence similarity and DNA–DNA reassociation of Methanococcus species found that the standard error was 19 % at 98 % sequence similarity, and 8 % at 90 % sequence similarity (Keswani & Whitman, 2001). This being the case, it is not possible to reliably differentiate between bacteria at the subspecies level, and in many instances, differentiation at the species level is questionable. In contrast, DNA–DNA hybridization provides more resolution at the species level and the 70 % criterion has formed a cornerstone in bacterial systematics (Stackebrandt & Goebel, 1994).

Greene & Voordouw (2003) discussed the relationship between hybridization intensity (Ix), the fraction of community DNA (fx) containing genome x, and Cx, the concentration of x on the membrane, so that Ix=k·fx·Cx. They also used an efficiency constant (k). This means that the intensity (Ix), or hybridization signal, increases as the abundance of a species in the sample (fx) increases, or the concentration of the probe on the membrane (Cx) increases. To increase sensitivity the concentration of the probe (Cx) can be increased, but at higher concentrations, the hybridization intensity (Ix) is non-linear. A decrease in sensitivity also occurs because cross-hybridization increases as Cx increases, and if cross-hybridization is greater than fx, then only populations at high abundance can be detected with confidence. As discussed below we formed a compromise, by increasing Cx (from 50 ng to 500 ng; Voordouw et al., 1991), and increasing the stringency of the post-hybridization washes (from 6x SSC to 5x SSC; Voordouw et al., 1991), so that cross-hybridization error was kept to a minimum. Lastly, by increasing Cx, we typically worked in the non-linear range for Ix and we thus used CGA primarily as a tool to survey rumen populations for candidates. Candidate populations could then be carefully quantified against a standard curve using a robust method such as rRNA Northern-blot analysis. Our logic was that since CGA is not robust enough to give accurate quantitative data, it was better to try increasing sensitivity, decreasing cross-hybridization, and dealing with quantification separately (Fig. 1).

In general, we found little cross-hybridization between bacterial strains except for strains within a species (Table 1). For example, we observed cross-hybridization among stains of B. fibrisolvens, Strep. bovis and S. ruminantium species (Table 1). This was expected because the 16S rRNA sequence similarity among these organisms is greater than 98 % (Maidak et al., 2001). At 65 °C wash temperature there was also some cross-hybridization among different species (e.g. B. fibrisolvens spp. and Eubacterium cellulosolvens) but this could be eliminated by increasing the hybridization temperature to 68 °C. Cross-hybridization among strains of the same species (e.g. B. fibrisolvens and Streptococcus spp.) was almost entirely abolished by increasing the wash temperature to 72 °C but at this temperature almost no hybridization signal was obtained from rumen samples. Two reasons could account for this result: (a) the hybridization conditions are so stringent that only homologous chromosomes will produce a signal, and (b) the abundance of the specific homologous DNA is at low levels in the rumen. For these reasons we decided to use a wash temperature of 68 °C. We did not attempt to correct data for cross-hybridization because this approach is questionable if the ‘rumen metagenome’ has not been characterized. In contrast, eliminating cross-hybridization between strains of closely related species with 16S rRNA-based approaches is far more difficult because the lack of sequence variation limits experimental options (Koizumi et al., 2002; Peplies et al., 2003).

Voordouw et al. (1991) included lambda DNA as an external marker for quantification. The assumption is that lambda DNA and the environmental DNA will be labelled with the same efficiency. We found this not to be the case: the lambda DNA was more efficiently labelled than the sample DNA. This was particularly problematic with the tannin-containing Acacia samples, which are rich in phenolics that are known to inhibit enzymic reactions (Haslam, 1989). Greene & Voordouw (2003) indicated that differential labelling could result in an fx of greater than one. We obtained good results by measuring the abundance of rDNA with a universal oligonucleotide probe and quantified the signal by comparison to a standard curve.

Strains within a species tended to respond to dietary changes in a similar manner. For example, B. fibrisolvens strains were sensitive to tannins, while S. ruminantium was tolerant (Fig. 3). Closely related species such as Strep. bovis and Strep. gallolyticus responded differently: Strep. bovis was sensitive while Strep. gallolyticus was tolerant, results which have been confirmed in vitro (Brooker & O'Donovan, 2001) and in vivo (Brooker et al., 2002). Actinomyces spp. LP1283 and Clostridium botulinum LP1284 were tolerant to tannins and both these strains were previously isolated in association with tannins (McSweeney et al., 1999). Ruminococcus spp. and F. succinogenes increased in the presence of Acacia (Fig. 4). This was a surprising result because it had previously been shown that Ruminococcus and Fibrobacter were sensitive to tannins both in vivo (McSweeney et al., 2001) and in vitro (Odenyo et al., 1997).

More accurate assessment of the changes in populations of Ruminococcus spp. and F. succinogenes was performed by Northern blot analysis. In the case of both Ruminococcus and Fibrobacter, the increase in abundance when Acacia was added to the diet was significant (P<0·05) when expressed as a proportion of the bacterial 16S rRNA population (4·3 to 9·5 % for Ruminococcus and 1·7 to 5·6 % for Fibrobacter) (Fig. 4) and there was little change (P>0·05) in these two populations when PEG was added to the diet. In contrast the array data (Fig. 3) demonstrated a significant (P<0·05) decline in Ruminococcus, but not in F. succinogenes, when PEG was added to the diet.

We hypothesized that Ruminococcus and F. succinogenes were indeed sensitive, but were less so than most other bacterial species, leading to an increase in their apparent relative abundance. When Ruminococcus and F. succinogenes are expressed as a proportion of the total bacterial, or microbial population, there was a significant increase in their abundance (Fig. 4) (P<0·05). There was also a decline in bacteria as a proportion of the total microbial population, but this was not significant (P>0·05) (Fig. 5). However, Ruminococcus declined as a percentage of the total microbial population (P<0·05), but not as a proportion of the total bacterial population, when PEG was added to the diet (Fig. 4). Expressing the Ruminococcus and F. succinogenes populations as a proportion of total microbial population more accurately reflects the array data because both are proportions of the total microbial population and not only the bacterial population.

The decline observed with Ruminococcus in the presence of PEG would suggest that a subset of the bacterial population adapt to the tannins. The differences in the Acacia and Calliandra results observed previously (McSweeney et al., 2001) can also be explained by the differences in the neutral detergent fibre (NDF)-bound tannins and the NDF digestibility (Hove et al., 2001). The NDF-bound tannins are 50 % less than in Calliandra and the NDF digestibility of Calliandra is extremely low. Ruminococcus and F. succinogenes typically adhere to the fibre surface (Krause et al., 2003a) and if the local concentration of tannins is lower in Acacia then high numbers of Ruminococcus and F. succinogenes might be expected. It was also not possible that the discrepancies in bacterial populations were the result of differential extraction of DNA or rRNA, because we used mechanical shearing which lyses at least 95 % of the microbial population.

From the above experiments, we conclude that there were no inherent deficiencies in our experimental method and thus decided to do a survey of the literature to ascertain to what extent variation in microbial population responses occurred. Close examination of the literature demonstrates that the responses of a bacterium can vary significantly depending on the type of plant and type of assay (in vitro or in vivo). Table 4 summarizes a range of studies that illustrate this point. Studies with Prevotella bryantii B14 have indicated that this strain is tolerant in vitro to Onobryachis viciifolia (Jones et al., 1994) and Yucca shidigera (Wallace et al., 1994) tannins, but sensitive in vitro to Lotus corniculatus and Lotus pedunculatus (Molan et al., 2001). Strep. bovis and Strep. gallolyticus are very closely related at the 16S rDNA sequence level, but Strep. bovis tends to be sensitive and Strep. gallolyticus tolerant (Brooker & O'Donovan, 2001). It is also clear that the response in vivo is far smaller than it is in vitro (McSweeney et al., 2001; Min et al., 2002; Molan et al., 2001) but small changes in the 16S rRNA-described population can often be far smaller than the functional changes in the system (Hashsham et al., 2000).


View this table:
[in this window]
[in a new window]
 
Table 4. Rumen bacteria and their ability to grow in the presence of various tannin-containing plants

 
In this paper, we have demonstrated that CGA is a robust method for surveying the microbial populations of the rumen, but that a post-CGA step should be contemplated if accurate quantitative data on population abundance are required. One of the drawbacks of CGA as described here, as compared with 16S rRNA-based methodology, is that the array is developed based on cultured species. It is well known that the majority of micro-organisms have not been cultured and in the rumen this proportion is 90–95 % (Krause & Russell, 1996). One solution to this problem is the construction of the rumen metagenome: cloning of megabase-size environmentally derived DNA into bacterial artificial chromosomes (BACs) (Rondon et al., 2000). Given that the fragments cloned are very large, there is a good possibility that an rRNA gene will be on the same fragment of DNA as a BAC clone, providing phylogenetic identity. CGAs constructed with BAC clones would greatly enhance the use of this technology to assess the phylogeny and physiology of the uncultured portion of the rumen (Krause & McSweeney, 2002; Krause et al., 2003b).


   ACKNOWLEDGEMENTS
 
This work was partially funded by the Australian Centre for International Agricultural Research. We would also like to thank Professor Gary Crow for his advice in mixed-model statistical analysis and its application to array data.


   REFERENCES
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
REFERENCES
 
Blaut, M., Collins, M. D., Welling, G. W., Dore, J., van Loo, J. & de Vos, W. (2002). Molecular biological methods for studying the gut microbiota: the EU human gut flora project. Br J Nutr 87, Suppl 2, S203–S211.[CrossRef]

Bonnet, R., Suau, A., Dore, J., Gibson, G. R. & Collins, M. D. (2002). Differences in rDNA libraries of faecal bacteria derived from 10- and 25-cycle PCRs. Int J Syst Evol Microbiol 52, 757–763.[Abstract/Free Full Text]

Brooker, J. D. & O'Donovan, L. (2001). Effect of hydrolysable and condensed tannins on growth, morphology and metabolism of Streptococcus gallolyticus (S. caprinus) and Streptococcus bovis. Microbiology 147, 1025–1033.[Medline]

Brooker, J. D., O'Donovan, L. A., Skene, I., Clarke, K., Blackall, L. & Muslera, P. (1994). Streptococcus caprinus sp. nov., a tannin-resistant ruminal bacterium from feral goats. Lett Appl Microbiol 18, 313–318.

Brooker, J. D., Rakhamani, S., O'Donovan, L., McSweeney, C. S. & Krause, D. O. (2002). Effect of tannins and other secondary plant products on microbial populations and gut function. In Beyond Antimicrobials: the Future of Gut Microbiology, 12–15 June, 2002. Edited by R. J. Wallace. Aberdeen: Rowett Research Institute.

Bryant, M. P. (1972). Commentary on the Hungate technique for culture of anaerobic bacteria. J Clin Nutr 25, 1324–1328.

Bush, K. J., Russell, R. W. & Young, J. W. (1979). Quantitative separation of volatile fatty acids by high-pressure liquid chromatography. J Liq Chromatogr 2, 1367–1374.

Cho, J. C. & Tiedje, J. M. (2002). Quantitative detection of microbial genes by using DNA microarrays. Appl Environ Microbiol 68, 1425–1430.[Abstract/Free Full Text]

Cho, J. C., Tiedje, J. M., Bina, J., Zhu, J., Dziejman, M., Faruque, S., Calderwood, S. & Mekalanos, J. (2002). Quantitative detection of microbial genes by using DNA microarrays: ToxR regulon of Vibrio cholerae and its expression in vibrios shed by cholera patients. Appl Environ Microbiol 68, 1425–1430.[Abstract/Free Full Text]

Churchill, G. A. (2002). Fundamentals of experimental design for cDNA microarrays. Nat Genet 32, Suppl, 490–495.[CrossRef]

Devereux, R., He, S. H., Doyle, C. L., Orkland, S., Stahl, D. A., LeGall, J. & Whitman, W. B. (1990). Diversity and origin of Desulfovibrio species: phylogenetic definition of a family. J Bacteriol 172, 3609–3619.[Medline]

El-Hassan, S. M., Kassi, A. L., Newbold, C. J. & Wallace, R. J. (2000). Chemical composition and degradation characteristics of foliage of some African multipurpose trees. Anim Feed Sci Technol 86, 27–37.[CrossRef]

Firoved, A. M. & Deretic, V. (2003). Microarray analysis of global gene expression in mucoid Pseudomonas aeruginosa. J Bacteriol 185, 1071–1081.[Abstract/Free Full Text]

Forster, R. J., Whitford, M. F., Teather, R. M. & Krause, D. O. (1998). Investigations into rumen microbial diversity using molecular cloning and probing techniques. In Genetics, Biochemistry, and Ecology of Cellulose Degradation, pp. 16–24. Japan: Sukuka.

Fox, G. F., Wisotzkey, J. D. & Jurtshuk, J. P. (1992). How close is close: 16S rRNA sequence identity may not be sufficient to guarantee species identity. Int J Syst Bacteriol 42, 166–170.[Abstract]

Greene, E. A. & Voordouw, G. (2003). Analysis of environmental microbial communities by reverse sample genome probing. J Microbiol Methods 53, 211–219.[CrossRef][Medline]

Hashsham, S. A., Fernandez, A. S., Dollhopf, S. L., Dazzo, F. B., Hickey, R. F., Tiedje, J. M. & Criddle, C. S. (2000). Parallel processing of substrate correlates with greater functional stability in methanogenic bioreactor communities perturbed by glucose. Appl Environ Microbiol 66, 4050–4057.[Abstract/Free Full Text]

Haslam, E. (1989). Enzyme inhibition. In Plant Polyphenols. Vegetable Tannins Revisited, pp. 172–176. Melbourne: Cambridge University Press.

Hove, L., Topps, J. H., Sibanda, S. & Ndlovu, L. R. (2001). Nutrient intake and utilization by goats fed dried leaves of the shrub legumes Acacia angustissima, Calliandra calothyrsus and Leucaena leucocephala as supplements to native pasture hay. Anim Feed Sci Technol 91, 95–106.[CrossRef]

Hugenholtz, P. (2002). Exploring prokaryotic diversity in the genomic era. Genome Biol 3, 1–8.

Jones, G. A., McAllister, T. A., Muir, A. D. & Cheng, K.-J. (1994). Effects of sainfoin (Onobrychis viciifolia Scop.) condensed tannins on growth and proteolysis by four strains of ruminal bacteria. Appl Environ Microbiol 60, 1374–1378.[Abstract]

Keswani, J. & Whitman, W. B. (2001). Relationship of 16S rRNA sequence similarity to DNA hybridization in prokaryotes. Int J Syst Evol Microbiol 51, 667–678.[Abstract]

Khanbabaee, K. & van Ree, T. (2001). Tannins: classification and definition. Nat Prod Rep 18, 641–649.[CrossRef][Medline]

Koizumi, Y., Kelly, J. J., Nakagawa, T., Urakawa, H., El-Fantroussi, S., Al-Muzaini, S., Fukui, M., Urushigawa, Y. & Stahl, D. A. (2002). Parallel characterization of anaerobic toluene- and ethylbenzene-degrading microbial consortia by PCR-denaturing gradient gel electrophoresis, RNA-DNA membrane hybridization, and DNA microarray technology. Appl Environ Microbiol 68, 3215–3219.[Abstract/Free Full Text]

Krause, D. O. & McSweeney, C. S. (2002). Construction of microbial metagenomes from uncultured rumen microorganisms. In CSIRO Biotechnology Conference, Australia, 17 to 19 April. Noosa, Australia. Canberra, CSIRO Publishing.

Krause, D. O. & Russell, J. B. (1996). How many ruminal bacteria are there? J Dairy Sci 79, 1467–1475.[Abstract/Free Full Text]

Krause, D. O., Dalrymple, B. P., Smith, W. J., Mackie, R. I. & McSweeney, C. S. (1999). 16S rDNA sequencing of Ruminococcus albus and Ruminococcus flavefaciens: design of a signature probe and its application in adult sheep. Microbiology 145, 1797–1807.[Medline]

Krause, D. O., Smith, W. J. & McSweeney, C. S. (2001a). Extraction of microbial DNA from rumen contents containing plant tannins. Biotechniques 31, 294–298.[Medline]

Krause, D. O., Bunch, R. J., Conlan, L. L., Kennedy, P. M., Smith, W. J., Mackie, R. I. & McSweeney, C. S. (2001b). Repeated ruminal dosing of Ruminococcus spp. does not result in persistence, but changes in other microbial populations occur that can be measured with quantitative 16S-rRNA-based probes. Microbiology 147, 1719–1729.[Medline]

Krause, D. O., Denman, S. E., Mackie, R. I., Morrison, M., Rae, A. L., Attwood, G. T. & McSweeney, C. S. (2003a). Opportunities to improve fiber degradation in the rumen: microbiology, ecology, and genomics. FEMS Microbiol Rev 27, 663–693.[CrossRef][Medline]

Krause, D. O., Pamanadbha, J., Trebbin, A., Al-Jassim, R. & McSweeney, C. S. (2003b). Construction of BAC libraries from the rumen microbial ecosystem. In CSIRO Horizons Conference, Surfers Paradise, Australia 25–28 May. Canberra: CSIRO Publishing.

Maidak, B. L., Cole, J. R., Lilburn, T. G. & 7 other authors (2001). The RDP-II (Ribosomal Database Project). Nucleic Acids Res 29, 173–174.[Abstract/Free Full Text]

McSweeney, C. S., Palmer, B., Bunch, R. & Krause, D. O. (1999). Isolation and characterization of proteolytic ruminal bacteria from sheep and goats fed the tannin-containing shrub legume Calliandra calothyrsus. Appl Environ Microbiol 65, 3075–3083.[Abstract/Free Full Text]

McSweeney, C. S., Conlan, L. L., Hegarty, M., Krause, D. O., Gough, J. & Orr, P. (2000). HPLC identification of toxic non-protein amino acids in the tropical leguminous tree Acacia angustissima. In Aust Soc Anim Prod, Sydney, Australia.

McSweeney, C. S., Palmer, B., Bunch, R. & Krause, D. O. (2001). Effect of the tropical forage calliandra on microbial protein synthesis and ecology in the rumen. J Appl Microbiol 90, 78–88.[CrossRef][Medline]

Min, B. R., Attwood, G. T., Reilly, K., Sun, W., Peters, J. S., Barry, T. N. & McNabb, W. C. (2002). Lotus corniculatus condensed tannins decrease in vivo populations of proteolytic bacteria and affect nitrogen metabolism in the rumen of sheep. Can J Microbiol 48, 911–921.[CrossRef][Medline]

Molan, A. L., Attwood, G. T., Min, B. R. & McNabb, W. C. (2001). The effect of condensed tannins from Lotus pedunculatus and Lotus corniculatus on the growth of proteolytic rumen bacteria in vitro and their possible mode of action. Can J Microbiol 47, 626–633.[CrossRef][Medline]

Odenyo, A. A. & Osuji, P. O. (1998). Tannin-tolerant ruminal bacteria from East African ruminants. Can J Microbiol 44, 905–909.[CrossRef][Medline]

Odenyo, A. A., Osuji, P. O. & Adinew, K. K. (1997). Microbiological evaluation of Acacia angustissima as a protein supplement for sheep. Anim Feed Sci Technol 65, 99–112.[CrossRef]

Peplies, J., Glockner, F. O. & Amann, R. (2003). Optimization strategies for DNA microarray-based detection of bacteria with 16S rRNA-targeting oligonucleotide probes. Appl Environ Microbiol 69, 1397–1407.[Abstract/Free Full Text]

Rondon, M. R., August, P. R., Bettermann, A. D. & 13 other authors (2000). Cloning the soil metagenome: a strategy for accessing the genetic and functional diversity of uncultured microorganisms. Appl Environ Microbiol 66, 2541–2547.[Abstract/Free Full Text]

Rudi, K., Flateland, S. L., Hanssen, J. F., Bengtsson, G. & Nissen, H. (2002). Development and evaluation of a 16S ribosomal DNA array-based approach for describing complex microbial communities in ready-to-eat vegetable salads packed in a modified atmosphere. Appl Environ Microbiol 68, 1146–1156.[Abstract/Free Full Text]

Sayler, G. S., Fleming, J. T. & Nivens, D. E. (2001). Gene expression monitoring in soils by mRNA analysis and gene lux fusions. Curr Opin Biotechnol 12, 455–460.[CrossRef][Medline]

Smith, A. H., Odenyo, A. A., Osuji, P. O., Wallig, M. A., Kandil, F. E., Seigler, D. S. & Mackie, R. I. (2001). Evaluation of toxicity of Acacia angustissima in a rat bioassay. Anim Feed Sci Technol 91, 41–57.[CrossRef]

Stackebrandt, E. & Goebel, B. M. (1994). Taxonomic note: a place for DNA-DNA reassociation and 16S rRNA sequence analysis in the present species definition in bacteriology. Int J Syst Bacteriol 44, 846–849.[Abstract]

Voordouw, G., Voordouw, J. K., Karkhoff-Schweizer, R. R., Fedorak, P. M. & Westlake, D. W. S. (1991). Reverse sample genome probing, a new technique for identification of bacteria in environmental samples by DNA hybridization, and its application to the identification of sulfate-reducing bacteria in oil field samples. Appl Environ Microbiol 57, 3070–3078.

Voordouw, G., Shen, Y., Harrington, C. S., Telang, A. J., Jack, T. R. & Westlake, D. W. S. (1993). Quantitative reverse sample genome probing of microbial communities and its application to oil field production waters. Appl Environ Microbiol 59, 4101–4114.[Abstract]

Wallace, R. J., Arthaud, L. & Newbold, C. J. (1994). Influence of Yucca schidigera extract on ruminal ammonia concentrations and ruminal microorganisms. Appl Environ Microbiol 60, 1762–1767.[Abstract]

Wilson, K. H., Wilson, W. J., Radosevich, J. L., DeSantis, T. Z., Viswanathan, V. S., Kuczmarski, T. A. & Andersen, G. L. (2002). High-density microarray of small-subunit ribosomal DNA probes. Appl Environ Microbiol 68, 2535–2541.[Abstract/Free Full Text]

Ziemer, C. J., Sharp, R., Stern, M. D., Cotta, M. A., Whitehead, T. R. & Stahl, D. A. (2002). Persistence and functional impact of a microbial inoculant on native microbial community structure, nutrient digestion and fermentation characteristics in a rumen model. Syst Appl Microbiol 25, 416–422.[Medline]

Received 1 December 2003; revised 28 April 2004; accepted 11 May 2004.



This Article
Abstract
Full Text (PDF)
Alert me when this article is cited
Alert me if a correction is posted
Citation Map
Services
Email this article to a friend
Similar articles in this journal
Similar articles in PubMed
Alert me to new issues of the journal
Download to citation manager
Google Scholar
Articles by Krause, D. O.
Articles by McSweeney, C. S.
Articles citing this Article
PubMed
PubMed Citation
Articles by Krause, D. O.
Articles by McSweeney, C. S.
Agricola
Articles by Krause, D. O.
Articles by McSweeney, C. S.


HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
INT J SYST EVOL MICROBIOL MICROBIOLOGY J GEN VIROL
J MED MICROBIOL ALL SGM JOURNALS
Copyright © 2004 Society for General Microbiology.