Use of real time PCR to determine population profiles of individual species of lactic acid bacteria in alfalfa silage and stored corn stover

Appl Microbiol Biotechnol (2006) 71: 329–338DOI 10.1007/s00253-005-0170-z

APPLIED MICROBIAL AND CELL PHYSIOLOGY

David M. Stevenson . Richard E. Muck .Kevin J. Shinners . Paul J. Weimer

Use of real time PCR to determine population profilesof individual species of lactic acid bacteria in alfalfa silageand stored corn stover

Received: 24 June 2005 / Revised: 26 August 2005 / Accepted: 2 September 2005 / Published online: 5 October 2005# Springer-Verlag 2005

Abstract Real-time polymerase chain reaction (RT-PCR)was used to quantify seven species of lactic acid bacteria(LAB) in alfalfa silage prepared in the presence or absenceof four commercial inoculants and in uninoculated cornstover harvested and stored under a variety of field con-ditions. Species-specific PCR primers were designed basedon recA gene sequences. Commercial inoculants improvedthe quality of alfalfa silage, but species corresponding tothose in the inoculants displayed variations in persistenceover the next 96 h. Lactobacillus brevis was the mostabundant LAB (12 to 32% of total sample DNA) in all ofthe alfalfa silages by 96 h. Modest populations (up to 10%)of Lactobacillus plantarum were also observed in in-oculated silages. Pediococcus pentosaceus populationsincreased over time but did not exceed 2% of the total.Small populations (0.1 to 1%) of Lactobacillus buchneriand Lactococcus lactis were observed in all silages, whileLactobacillus pentosus and Enterococcus faecium were

near or below detection limits. Corn stover generallydisplayed higher populations of L. plantarum and L. brevisand lower populations of other LAB species. The dataillustrate the utility of RT-PCR for quantifying individualspecies of LAB in conserved forages prepared under a widevariety of conditions.

Introduction

Ensiling is a forage preservation process by which moistplant material is subjected to an anaerobic fermentation ofits soluble sugars, primarily by lactic acid bacteria (LAB)(Muck 1991; McDonald et al. 1991). The resulting silage isstabilized against further anaerobic degradation whileretaining most of the original plant nutrients for subsequentfeeding to ruminant animals. Effective silage fermentationis characterized by (1) a rapid and substantial pH decreaseresulting from bacterial production of relatively largeamounts of lactic acid (pKa∼3.7) and minimal amounts ofweaker organic acids (e.g., acetic acid, pKa∼4.8) and un-desired nonacidic products such as ethanol, glycerol, and2,3-butanediol and (2) minimal degradation of plant cellwall carbohydrates and plant proteins. Such characteristicsarise from proliferation of homofermentative LAB at theexpense of heterofermentative LAB, Clostridium species,yeasts, and fungi (Muck 1991), particularly in the earlystages of fermentation when plant sugars are most abun-dant. Although silage production is a natural process ini-tiated by an epiphytic microbial population present on theplant at the time of harvest, commercial silage inoculants,consisting of pure or mixed cultures of homolactic LAB, areoften added by farmers at the start of ensiling to favor thehomolactic fermentation and insure a high-quality silage(Bolsen et al. 1996).

Silage fermentation displays a succession of microbialpopulations, with LAB and Enterobacteriaceae typicallydominating [>108 colony forming units (CFU) per gram]over the first few days, after which the latter populationdeclines precipitously (Muck 1991). Populations of yeastsand Clostridium are substantially lower (<104 CFU/g) and

Disclaimer: Mention of products is for informational purposes onlyand does not imply a recommendation or warranty by USDA overother products that may also be suitable

D. M. Stevenson . R. E. Muck . K. J. ShinnersDepartment of Biological Systems Engineering,University of Wisconsin-Madison,Madison, WI 53706, USA

R. E. Muck . P. J. WeimerUnited States Dairy Forage Research Center,Agricultural Research Service, US Department of Agriculture,Madison, WI 53706, USA

P. J. WeimerDepartment of Bacteriology, University of Wisconsin-Madison,Madison, WI 53706, USA

R. E. Muck (*)USDA-ARS-USDFRC,1925 Linden Drive West,Madison, WI 53706, USAe-mail: [email protected].: +1-608-890-0067Fax: +1-608-890-0076

generally remain low once sufficient acid production hasoccurred. Within the LAB, substantial changes in thepopulations of individual species have also been reported(Lin et al. 1992). However, our knowledge of the details ofmicrobial succession during ensiling has been hampered bya reliance on culture-dependent methods that distinguishamong individual microbial species only with difficulty—often requiring physiological or biochemical testing ofmany individual colony isolates—and which are subject toinherent biases of culturing methods. Culture-independentmethods have not been widely employed in silage fermen-tation studies, perhaps because there is minimal variationamong many LAB species with respect to sequences ofsmall-subunit rRNA or ribosomal intergenic spacer regions,the most commonly-used tool for phylogenetic analysis(Berthier and Ehrlich 1998; Collins et al. 1991). Recently,recA gene sequences have been reported to be useful for thetaxonomy of LAB and for differentiation of species inmixed culture (Felis et al. 2001; Torriani et al. 2001). Thepresent study was undertaken to develop an effective,culture-independent method for quantifying individualspecies of LAB in silages and conserved forages preservedunder highly divergent conditions, using a real-timepolymerase chain reaction (RT-PCR) based on species-specific primers to the recA gene.

Materials and methods

Alfalfa silage

Alfalfa (Medicago sativa L.) was grown at the University ofWisconsin West Madison Agricultural Research Station,Madison, WI (43.083°N, 89.517°W). Whole herbage wasmown, wilted to 390 g dry matter/kg alfalfa, and choppedwith standard field equipment. Immediately after chopping,the alfalfa was transported to the laboratory and ensiled in500-ml glass Weck canning jars (Glashaus Inc., CrystalLake, IL). In addition to the uninoculated control thatcontained an epiphytic LAB population, four commercialsilage inoculants (various LAB species) were used: Sil-All4×4 (containing Lactobacillus plantarum, Pediococcusacidilactici, and Enterococcus faecium; Alltech, Lexington,KY), Biomax MP (L. plantarum and Pediococcus pento-saceus; Chr. Hansen, Milwaukee, WI), Pioneer 1174 (L.plantarum and E. faecium; Pioneer Hi-Bred Int’l, Johnston,IA), and PowerStart (L. plantarum and Lactococcus lactis;Genus Breeding Ltd., Nantwich, Cheshire, UK). Inoculantswere applied at the rates specified by the manufacturer (i.e.,gram inoculant per milligram crop). To insure even dis-tribution of the inoculum for small-scale experiments,products were diluted with distilled water so that all wereapplied at 1 g cell suspension/100 g crop (fresh weight);controls received 1 g water/100 g crop. The actual LABapplication rates (CFU per gram alfalfa) for the fourinoculants in the experiment as measured were 1.21×105

(Sil-All 4×4), 2.75×105 (BioMax MP), 1.25×105 (Pioneer1174), and 1.38×105 (PowerStart). For each silo, 250 galfalfa was sprayed with inoculant, mixed, and then packed

by hand into the silo. After inoculation, silos were stored atroom temperature (∼21–23°C). Duplicate minisilos wereopened and analyzed after 24, 48, and 96 h of incubation.

Corn stover

Corn hybrids were grown at the University of WisconsinArlington Agricultural Experiment Station (43.3269°N,89.3631°W) and harvested in October 2003. Within severalhours of grain harvesting, the remaining corn stover (cob,leaf, husk, and stalk) was shredded and windrowed using a6-m wide Hiniker model 5600 flail shredder set at 15 cmheight. The shredder formed a windrow on the far right-hand side of the machine, and another windrow was placedimmediately adjacent to the first on the subsequent pass inthe opposite direction to form a double windrow. Choppedstover was then collected with a forage harvester andensiled in plastic silo bags (2.4 m diameter×∼30 m length;Up North Plastics, Cottage Grove, MN). Baled stover washarvested with a large round baler and ensiled by wrappingthe cylindrical bales (1.5×1.8 m) in plastic film. Furtherdetails of wet stover harvest, baling and ensiling at differentinitial moisture contents are described in Shinners et al.(unpublished data). Individual chopped loads and balesdiffered in moisture content due to field variation anddifferences in time between grain and stover harvest. Sam-ples collected from the bag silos or bales were targeted toachieve moisture and fermentation variation. No bacterialinoculants were added prior to storage. Stover was storedfor 242 to 284 days, depending on the individual exper-iment. Each of eight bales (initial dry matter content =24.8to 51.0%) were sampled by removing a 50-mm diameterbore sample to a depth of roughly 500 mm from oppositesides of each bale. In either case, the initial section (outerrind consisting of roughly 200 mm) of the bore sample wasdiscarded, and random material from the remaining boresample was packed tightly into a sterile 50-ml plasticcentrifuge tube and transported to the laboratory on ice.Samples from two silo bags (initial moisture content ofstover =41 or 51%) were collected after bag opening andmechanical removal of a portion of the silage (0.5 to 3 m) atone end of the bag; two random grab samples, collectedwith a latex-gloved hand from the exposed face of thesilage, were packed into sterile 50-ml plastic centrifugetubes and transported to the laboratory on ice.

DNA preparation

Different methods were used in DNA purification frombacterial cultures and environmental samples. Corn stoversamples were found to contain high levels of contaminants,especially humic acids and so were subjected to a moreextensive procedure. Stover (∼25 g) was first mixed with25 ml TE buffer (100 mM Tris/HCl, 10 mM EDTA, pH8.0) in a Waring blender fitted with a small stainless steelcup, then blended twice (10 s each) at full speed. Themixture was then squeezed through four layers of cheese-

330

cloth, and the filtrate (∼10–15 ml) was centrifuged(4,000×g, 10 min). The bacterial cell pellet was resus-pended in 700 μl TE buffer in a screw-cap microfuge tubecontaining 0.5 g glass beads (0.1 mm diameter), 50 μl of20% sodium dodecyl sulfate (SDS), and 700 μl ofequilibrated phenol (pH 8.0). After bead-beating (4°C,2 min) to lyse the cells, tubes were placed in a waterbath (60°C, 10 min) and then bead-beaten again. Theaqueous and phenol phases were separated by centrifu-gation (12,000×g, 5 min), and the aqueous phase wassequentially extracted with 500 μl phenol (pH 8.0), twicemore with 500 μl phenol/chloroform, pH 8.0, and twicewith 500 μl chloroform. Small amounts of buffer wereoccasionally added to keep the aqueous volume at ∼700 μl.The final supernatant (volume generally reduced to∼450 μl or less) was combined with 0.1 vol of 3.0 M Naacetate and 1,000 μl of 100% ethanol, and the mixturecooled to −20°C overnight. The mixture was centrifuged(12,000×g, 10 min), and the alcohol decanted off. Theprecipitated DNA pellet was then washed twice with1,000 μl of 70% ethanol and centrifuged (5 min). After afinal centrifugation, the remaining liquid was removed, andthe pellet left at room temperature for ∼20 min. The drypellet was resuspended in 80 μl TE, and 20 μl loading dyewas added prior to electrophoresis in a 6% agarose gel(90 V, 15–30 min) to remove the large amount of humicacids present in the samples. Upon staining with ethidiumbromide, the brown humic acids were seen to run ahead ofthe genomic DNA. The smear of genomic DNA observedabove any RNA present was excised and put into a gelextraction spin column (Ambion, Austin, TX). The gelfragments were crushed and covered with approximately100 μl TE, and after overnight incubation, the suspension

was centrifuged (5 min), reextracted with 100 μl TE, andcentrifuged again. The combined supernatants were thenextracted twice with an equal volume of butanol to re-move residual ethidium bromide. To remove contaminatingRNA, the solution was incubated with RNAse One(Promega, Madison, WI) following manufacturer’s direc-tions. Finally, the DNA solution was purified with aPromega Wizard DNA Clean-up system (Promega) fol-lowing manufacturer’s directions; this generally requiredtwo columns per sample. The resultant DNA solution wasquantified by spectrophotometry (ε260=0.020 l mg−1 cm−1

for double-stranded DNA) and stored at 4°C for short-termuse, or at −80°C for the long term.

For the alfalfa minisilos, 20 g of material was removedfrom the silos into a clean, tared blender jar and dilutedwith distilled water to a total of 200 g. The suspension wasblended at high speed (30 sec), then filtered throughcheesecloth. Filtrates (5 ml) from duplicate minisilos werecombined and used for DNA extraction. For pure cultures,2 ml of culture was used for DNA extraction. The sampleswere then centrifuged to pellet the bacterial cells from theminisilos (4,000×g, 10 min) or the cultures (12,000×g,10 min). After this, the samples were treated identically,using the procedure described above for corn stover, exceptthat an electrophoresis step was not necessary to removehumic acid materials.

Primer design

Although PCR primers specific for individual species ofclosely-related bacteria (Matsuki et al. 2004a,b) includingLAB (Furet et al. 2002; Settanni et al. 2005; Klocke et al.

Table 1 Percent nucleotide identity between a 307-bp region of recA common to all strains tested

Bacillussubtilis

Enterococcusfaecium

Lactobacillusbrevis

Lactobacillusbuchneri

Lactococcuslactis

Lactobacilluspentosus

Lactobacillusplantarum

Bacillus subtilis(2)

100

Enterococcusfaecium (1)

74.9 –

Lactobacillusbrevis (1)

70.4 73.3 –

Lactobacillusbuchneri (3)

70.4–70.7 73.6–74.3 75.2–75.9 98.0–99.3

Lactococcuslactis (2)

71.0–71.3 72.6–73.0 68.7–71.3 70.4–71.7 91.5

Lactobacilluspentosus (2)

72.3–73.0 77.2–77.5 76.9–78.2 77.5–79.5 69.4–70.0 98.4

Lactobacillusplantarum (3)

73.3–73.6 74.3–74.6 79.8–80.1 75.2–76.9 70.7–71.3 87.3–88.3 99.7–100

Pediococcuspentosaceus(1)

70.4 75.2 75.6 73.9–74.3 71.7–72.0 72.3–72.6 73.6–73.9

This region was either de novo sequenced or found in the available databases. Because several sequences were incomplete, only this 307-bpregion provided overlapping sequences sufficient to permit a meaningful comparison. Numbers in parenthesis indicate the number ofsequences from different stains examined

331

2005) have been successfully designed based on 16S rRNAgenes, we observed nearly complete homology in therRNA genes of several LAB commonly found in plantmaterial. Consequently, we elected to use the more di-vergent recA gene as a basis for the design of species-specific primers.

Primer pairs for quantitative analysis (Table 1) weredesigned using the Applied Biosystems (Foster City, CA)Primer Express software, which generated a list of primerpairs selected with Tm values between 58 and 60°C, a totalamplicon size between 60 and 80 bp, and no more than twoG’s and/or C’s in the last five bases. From this list, primerpairs were selected to cover areas of recA that wereconserved within a species and were as different as possiblefrom other closely related species. Because recA sequencesfor Lactobacillus brevis and Lactobacillus buchneri werenot available in databases, a single degenerate primer pairwas used to clone a fragment of recA from these two spe-cies. Several recA sequences [using the PILEUP programwithin the GCG Wisconsin software package (Accelrys,San Diego, CA)] were aligned to determine a region of recAsequence conservation among Lactobacilli (Table 1). De-generate primers were then designed within this regionutilizing as little degeneracy as possible (Table 2). Thisprimer pair was then used to clone fragments of recAfrom L. brevis ATCC 14869 (Genbank accession numberDQ080023), L. buchneri P (DQ080024), L. buchneri TY16(DQ080025), and Weissella confusa ATCC 14434 (for-merly L. brevis, see below; DQ080026). The determinedsequences were then used to generate quantitative analysisprimer pairs as described above (LacBuc2 and LacBre3 inTable 2).

Cloning

The universal degenerate recA primer in Table 2 was used toamplify recA fragments from L. brevis and L. buchneri

DNA. Taq polymerase and other reagents were obtainedfrom Promega and used in a 100 μl reaction mix containing6.0 μl of 25 mM MgCl2, 10.0 μl of 10× reaction buffer(without MgCl2), 2.0 μl of PCR nucleotide mix (10 mMeach dNTP), 2.5μl of each primer (UrecA3F and UrecA3R,each 20 μM), 0.5 μl Taq DNA polymerase (5 U/μl), 2.0 μltemplate DNA (approximately 1–20 ng/μl genomic DNAisolated from target species as described above), and 24.5μlnuclease-free water. This mixture was then loaded (50 μleach) into two 0.2-ml thinwall microfuge tubes and sub-jected to the following thermocycles: 1 cycle of one step of95°C for 120 s, followed by 40 cycles of four steps of 95°Cfor 60 s, 55°C for 90 s, 57°C for 90 s, and 72°C for 120 s.

The resulting PCR product was purified by agarose gelelectrophoresis using standard methods (Sambrook et al.1989) followed by gel excision, isolation, and purificationutilizing the Wizard SV Gel and PCR Clean-up system.This purified PCR product was cloned into the pGEM-T-Easy vector using the pGEM-T and pGEM-T Easy VectorSystem (Promega) following manufacturer’s directions.The resulting clones were screened by blue-white selectionusing standard methods, and DNA sequenced by theUniversity of Wisconsin Biotechnology Center. Clones thatvery strongly matched recA sequences from Lactobacilluswhen BLAST-searched against GenBank were selected forfurther sequencing and utilization for primer design.

Real-time polymerase chain reaction

A master mix for each primer set was prepared such thateach well contained the following: 2× SYBR Green MasterMix (Applied Biosystems) that contained all the nucleo-tides, polymerase, reaction buffer, and SYBR Green dye;forward and reverse primers at concentrations of 50, 300,or 900 nM depending on previously determined optimalconcentrations; and nuclease-free water to a total of 23 μlper well. To this, 2.0 μl of each sample or standard was

Table 2 Primers designed to target species-specific regions of the recA gene of lactic acid bacteria

Target species Primer name Forward/reverse Primer sequence Amplicon length (bp)

E. faecium EfRecA1 F GACGGCGAAATGGGTGACT 73R CAGAGAGTTTACGCAATGCTTGA

L. brevis LacBre3 F GCAGTTGCCGAGGTCCAA 64R CCAACGCATTTTCAGCATCA

L. buchneri LacBuc2 F GGACCAATGCAGCAACTGAA 72R AGATTACTGACGCATTGGTTACCA

L. lactis LcoLac1 F TGTCACAAGCCATGCGTAAAC 78R CACGCAATTGGTTGATGAAAA

L. pentosus LacPen1 F CAAGCCCGGTTAATGTCACA 70R GTGGGATGGTCTTTGTCTTGTTC

L. plantarum LacPla1 F AGGCGCGGCTGATGTCA 68R CGCGATTGTCTTGGTTTTGTT

P. pentosaceus PedPen3 F CTATTGACTTGGTCGTTATTGATTCC 72R CCCCCATCTCTCCATCAATTT

Universal Lactobacillus UrecA3 F TTTAYGCGGAACAYYTRGGKGT ∼300–450R CCAAACATCACVCCRACTT

332

added, and the plate was briefly centrifuged and placed inthe thermocycler for analysis.

Species-specific real-time quantitative PCR was per-formed using the Applied Biosystems Prism 7000 se-quence detection system, with fluorescence detection ofSYBR Green dye. Amplification consisted of an initialhold for 10 min followed by 40 to 50 cycles of 95°C for15–25 s and 58–60°C for 60–90 s. Standards were run onthe same plate in triplicate, and the amount of DNA of thetarget species was calculated as a percentage of the totalDNA added to the wells (determined spectrophotometri-cally). In general, approximately 2–20 ng of target DNAwas loaded in each well, and standards serially diluted in1:5 increments. A typical set of RT-PCR reactionsincorporating a range of standards from pure culture(viz., P. pentosaceus) and from conserved forage samplesof varying levels of this species is shown in Fig. 1.

Statistical analysis

Corn stover data were analyzed by the General LinearModel of the SAS Statistical Software Package 6.12, withpercentages of each LAB species as the response variableand with form (bale or bag), sample location within the baleor bag (inside, middle, or outside), and percent moisturecontent as main effects and as interaction terms. Data werealso analyzed separately for bales and bags, with locationand moisture content as main effects and interaction terms.Mean separations were performed using Duncan’s multiplerange test at p<0.05.

Results

Species-specific detection of LAB

Because there were no recA sequences in the databases forL. buchneri and L. brevis, fragments of the recA gene fromtwo strains of each species (L. buchneri strains TY16 and Pand L. brevis ATCC 14869 and ATCC 14434) were clonedusing the universal Lactobacillus primer. The resultingsequences for the two L. buchneri strains and L. brevisATCC 14434 (subsequently reclassified as W. confusa;Collins et al. 1993) were nearly identical, while that of L.brevis ATCC 14869 was considerably different. Moreover,preliminary work using primers designed to amplify re-gions of the internally transcribed spacer region betweenthe 16S and 23S rRNA coding regions showed the strainsclustering in a manner identical to the cloned recA (data notshown). For our quantitative studies, we used the L. brevisprimer designed on the basis of the recA sequence of L.brevis ATCC 14869.

Table 3 shows that the primer sequences used foramplification of the recA genes in the various LAB werespecies-specific within the LAB group. The E. faecium, L.lactis, and P. pentosaceus primers did not amplify DNAfrom the nontarget species tested, including that from thefour Lactobacillus species. Among the Lactobacillusprimers, only LacBuc2, designed for L. buchneri, showedcross-reaction in the form of amplifying DNA from W.confusa strain 14434.

Alfalfa silage fermentation

The utility of the primers in quantifying LAB was tested inlaboratory-scale “minisilos” containing alfalfa. The timecourse of the silage fermentation is shown in Table 4. Allfour commercial inoculants significantly (p<0.05) im-proved the speed and course of fermentation relative tothe untreated control. The most rapid pH declines occurredin the Biomax and Pioneer inoculants. However, the pHvalues among the four inoculated treatments were notsignificantly different at 96 h. The inoculated treatmentshad elevated lactic acid concentrations and in mostinstances reduced acetic acid and ethanol concentrationsrelative to the control, indicating that the inoculants

20

25

30

35

40

15-3 -2 -1 0 1 2-4-5

11

-10123456789

10

1 5 10 15 20 25 30 35 40 45

A

B

X

Y

Z

X

Y

LOG ng DNA

Ct

Cycle

Fluo

resc

ence

8.36

1.672

0.3344

0.06688 0.0133760.002675

Fig. 1 Representative PCR reactions. a Net fluorescence detectionfrom varying amounts of P. pentosaceus DNA (indicated innanograms) run on the same PCR plate with DNA extracted fromthree environmental samples [X (1.84 ng), Y (5.0 ng), and Z(5.0 ng); amounts determined spectrophotometrically]. b Standardcurve of the data in a along with the calculated DNA amount forsamples X and Y. By calculations from the standard curve, theunknowns X, Y, and Z contained 0.1465, 0.0001236, and 0.0 ng ofP. pentosaceus DNA, respectively, and thus represented approxi-mately 7.97 and 0.0025% and an undetectable level of the totalDNAwithin the sample, respectively. Sample identities: X, minisilosample inoculated with BioMax additive sampled at time zero; Y,sample from bagged corn stover put up at 51% moisture samplednear the outside of the bag; Z, second of the above corn stover at adifferent location within the bag

333

produced a more homofermentative lactic acid fermenta-tion than did the epiphytic population in the controls.

RT-PCR reactions conducted with samples ensiled for 0,24, 48 and 96 h revealed that control silage contained asmall population of LAB at zero time, predominantly in theform of L. brevis and L. buchneri, with only traces of L.plantarum and E. faecium (Table 5). By 96 h, the controlsilage contained L. brevis as the dominant LAB, compris-ing 27% of the bacterial population; modest populations(∼1% of sample DNA) of L. plantarum were also evident.

While it appears that all four inoculated silages displayeddramatic reductions in the populations of the inoculumspecies within the first 48 h of incubation (E. faecium in Sil4×4, P. pentosaceus in BioMax MP, L. buchneri in 1174, L.lactis in PowerStart, and L. plantarum in all four silages),these zero-time values are expressed on the basis of theinoculum solution only and do not include the substantialepiphytic LAB and non-LAB populations present in theoriginal plant material. During the 96 h after inoculation,the populations of some species increased (e.g., L. planta-

Table 3 Reaction of PCR primers with test strains of LAB

Test strains Primers and target speciesa

EfRecA1 E.faecium

LacBre3 L.brevis

LacBuc2 L.buchneri

LcoLac1 L.lactis

LacPen1 L.pentosus

LacPla1 L.plantarum

PedPen3 P.pentosaceus

E. faecium SF68b + − − − − − −E. faecium WNb + − − − − − −L. brevis ATCC 14869c − + − − − − −L. buchneri TY16d − − + − − − −L. buchneri Pe NT − + NT NT NT NTL. lactis Gf NT − − + NT NT NTL. pentosus AK-127b − − − − + − −L. plantarum 5M-1Rb − − − − − + −L. plantarum NCIMB40027g

− − − − − + −

P. pentosaceus NCIMB30078g

− − − − − − +

P. pentosaceus 5M-1Pb − − − − − − +W. confusa ATCC14434c,h

NT +/− + NT NT NT NT

a+, positive reaction; −, at least six PCR-cycle difference in CT relative to primary target strain; NT not testedbAgri-King (Fulton, IL)cAmerican Type Culture Collection (Manassas, VA)dLab isolate from alfalfa silageeLab isolate from Pioneer 1174 inoculant (Pioneer)fLab isolate from PowerStart inoculant (Genus)gNational Collection of Industrial, Food and Marine Bacteria (University of Aberdeen, Aberdeen, UK)hFormerly classified as L. brevis (Collins et al. 1993)

Table 4 Fermentation characteristics of alfalfa silage from the minisilos

Characteristic Hour Control Sil-All 4×4 Biomax MP Pioneer 1174 PowerStart

pH 0 6.32±0.08 6.32±0.08 6.32±0.08 6.32±0.08 6.32±0.0824 6.17±0.14 6.18±0.04 6.05±0.05 6.15±0.07 6.17±0.0748 5.72±0.01 5.46±0.02 5.20±0.01 5.29±0.01 5.40±0.0496 5.34±0.09 4.82±0.10 4.70±0.05 4.76±0.01 4.86±0.16

Lactic acid, (% DM) 24 1.26±0.19 1.39±0.06 1.43±0.02 1.24±0.11 1.50±0.1148 1.98±0.52 3.24±0.28 3.68±0.23 3.99±0.54 3.06±0.3196 3.99±0.11 6.06±0.39 6.47±0.25 6.09±0.20 5.67±0.07

Acetic acid, (% DM) 24 0.77±0.08 0.70±0.10 0.67±0.06 0.50±0.11 0.81±0.0948 0.84±0.22 0.88±0.02 0.84±0.01 0.90±0.09 0.97±0.0296 1.36±0.01 1.24±0.00 1.05±0.01 1.19±0.03 1.52±0.35

Ethanol, (% DM) 24 0.16±0.00 0.19±0.03 0.14±0.01 0.09±0.01 0.15±0.0648 0.11±0.01 0.08±0.00 0.14±0.01 0.08±0.00 0.14±0.0496 0.24±0.21 0.15±0.05 0.10±0.01 0.11±0.01 0.15±0.00

Results are the means±standard errors of duplicate minisilos with the exception of pH at 0 h (mean of four samples)% DM, percent by weight (dry matter basis)

334

rum), while some populations declined (e.g., L. buchneri).In particular, L. brevis became a substantial component ofthe microbial population by 96 h, at which time itrepresented 12 to 32% of the bacterial DNA in all silages.

P. pentosaceus was a substantial component of theBioMax inoculant, but its relative population size declinedconsiderably during the first 24 h of ensiling. However, inall of the silages, the populations of this species increasedconsiderably over the next 72 h to reach values up to 1.8%of the total population. Small populations (0.1 to 1% ofsample DNA) of L. buchneri and L. lactis were observed inall silages, while Lactobacillus pentosus and E. faeciumwere near or below detection limits.

Corn stover fermentation

A further test of the primers was conducted at field scale,using corn stover harvested using different equipment fromdifferent fields, then stored at scales of >1 ton in largeround bales or plastic bags for 8–9 months (Shinners et al.,unpublished data). The stover contained an epiphyticmicrobial population, but no bacterial inoculant was addedprior to storage. Despite the differences in source andpreparation of the stovers, nearly all of the stoversdeveloped dominant populations of L. plantarum, andsome developed substantial or even dominating popula-tions of L. brevis (Table 6). L. buchneri was widely

Table 5 Percentage of total DNA detected as target species DNA in “minisilos” containing alfalfa plus commercial inoculants or containingan uninoculated alfalfa control having only the epiphytic population

Time (h) Sil 4×4 BioMax 1174 PowerStart Control

E faecium0a 65.9 0.0462 1.31 0.0003 0.0000524 0.0006 ND 0.0006 0.0001 ND48 0.0043 0.00009 0.0001 0.0011 0.000496 0.0147 0.0002 0.0030 0.0002 0.0027L. brevis0a 0.0668 0.0982 0.0466 0.0179 1.7224 3.28 1.44 3.13 3.94 2.3148 3.84 5.87 8.98 3.32 4.8096 16.9 12.0 31.1 32.2 27.2L. buchneri0a 0.670 1.85 4.95 0.165 0.22424 1.073 0.245 0.336 0.406 0.17948 0.215 0.225 0.120 0.280 0.30596 0.299 0.0939 0.140 0.198 0.164L. pentosus0a 0.00089 1.24 0.133 ND ND24 ND ND ND ND ND48 ND 0.0009 0.0012 ND ND96 0.00014 0.0528 0.0001 ND 0.0007L. plantarum0a 56.7 13.5 85.0 14.6 0.000224 0.322 0.440 0.278 0.0624 0.037648 4.12 4.04 4.62 3.10 0.11896 0.133 9.67 0.181 9.40 1.07L. lactis0a ND ND ND 37.4 ND24 0.482 0.141 0.131 1.05 0.21348 0.153 0.174 0.194 0.438 0.27296 0.144 ND 0.0925 0.799 0.0113P. pentosaceus0 0.00082 7.97 0.0138 0.0022 0.0000724 0.125 0.360 0.0917 0.0577 0.069948 0.261 0.598 1.07 0.218 0.52496 0.522 0.712 0.287 0.389 1.78

Minisilos were sampled after the indicated incubation timesND Not detectedaFor inoculated silages, values do not represent true time-zero values of DNA extracted from minisilos, but instead are values for cellsuspensions used to inoculate minisilos. See text for discussion

335

detected, but rarely exceeded 1% of the population, and itsmean value of 3.62% of the population was skewed by asingle sample (mean value declined threefold to 1.23%upon deletion of the anomalous value). L. pentosus wasalso widely detected, but in even lower amounts (<0.016%of the population; data not shown). P. pentosaceus was

widely detected as well, but was present in significantamounts (0.05 to 0.36% of the population) in only a fewsamples. E. faecium was barely detected in baled stoversamples and was not detected at all in stover stored insealed bags (data not shown). L. lactis was not detected inany of the stover samples. Regression analysis revealed

Table 6 Percentage of total DNA detected as target species DNA in samples removed from bales or bags of corn stover stored at differentinitial moisture contents for ∼9 months

Sample location Moisture (%) L. brevis L. buchneri L. plantarum P. pentosaceus

Bale Near center 28.4 0.0272 0.0739 2.02 ND0.0429 1.66 1.80 0.0002

28.8 0.0251 0.0145 0.0720 ND0.0258 0.0476 2.52 ND

38.1 64.8 1.19 58.8 0.00221.34 1.87 6.24 0.0002

38.8 0.118 0.240 41.5 ND1.06 0.177 4.05 0.0003

45.2 6.87 0.475 7.10 0.00216.27 0.369 14.2 0.0060

45.6 0.140 0.138 6.02 0.00100.381 0.349 8.80 0.0100

51.4 1.73 0.0967 31.8 0.00010.392 0.219 6.42 0.0012

51.5 10.2 0.0524 3.12 0.3600.550 0.184 8.00 0.0312

Near outside 28.4 0.250 0.131 2.22 0.000128.8 0.0795 0.272 1.89 0.0445

0.0831 0.240 1.70 0.000138.1 7.76 0.183 18.8 ND

37.7 0.282 22.1 0.000238.8 7.21 11.5 10.0 0.0017

39.9 0.191 32.8 0.000245.2 12.7 0.268 12.2 0.0027

0.463 0.257 24.3 0.000245.6 0.115 0.154 34.7 0.0249

0.0440 0.0708 21.9 0.004951.4 0.0499 0.138 8.59 0.0024

2.41 0.196 65.0 0.000251.5 3.20 0.102 61.3 ND

0.0297 0.126 19.1 0.0003Bag Near center 41 4.85 106.3 4.14 0.0002

4.71 12.0 4.49 0.000851 8.63 0.154 10.2 0.0002

15.3 0.163 8.01 0.0006Midway 41 6.79 1.18 9.82 0.0015

0.112 0.601 3.86 0.000751 13.6 0.0742 4.82 0.0001

0.265 0.319 2.19 0.0001Near outside 41 0.160 0.105 1.58 0.0001

1.49 0.277 3.35 0.000851 23.7 0.120 30.4 0.0025

110.6 15.0 1.52 NDMean values all treatments 9.21bc 3.62cd 14.50b 0.012d

bcdWithin a row, values not sharing a superscript differ (p<0.05)ND Not detected

336

that none of the populations of individual speciescorrelated with one another (r2<0.12).

While a general hierarchy of population sizes amongLAB species was apparent, substantial variation amongreplicates made it difficult to identify treatment conditionsthat affected the relative population sizes of individualLAB species. Analysis of variance indicated that popula-tions of most LAB species were not affected (p>0.05) byform of storage (bale vs bag), by the sampling locationwithin the bale or bag, or by moisture content. Separateanalysis of data from bales indicated a trend (p=0.065)toward an effect of moisture content on LAB populations.More specifically, LAB populations (particularly L. plan-tarum and L. brevis), expressed as a percentage of species-specific DNA relative to total DNA in the sample, weresignificantly lower in bales put up at low (28.4 and 28.8%)moisture content than at higher moisture contents. Mois-ture effects were not observed in silage bags, probablybecause they retained moisture above levels required forLAB growth. L. brevis was the dominant species detectedin stover put up in silage bags at high (51%) moisturecontent, but was less abundant in bales put up at thismoisture level. P. pentosaceus populations were generallyvery low, but in several of the high-moisture stovers, thepopulations of this species became significant.

Discussion

RT-PCR has been used to quantify several different speciesof fermentative bacteria in several natural and engineeredenvironments (Coeuret et al. 2004; Grattepanche et al.2004; Lin et al. 1992; Tajima et al. 2001). The datapresented here represent the first report of the use of RT-PCR to systematically quantify different specific LABpopulations in conserved forages, although Klocke et al.(2005) have recently used RT-PCR based on 16S rDNA-directed primers to quantify L. plantarum in grass silage.Primers directed toward the recA gene exhibited acceptablelevels of species specificity for use in two divergent typesof stored forage, alfalfa silage prepared conventionallyunder high moisture conditions, and corn stover stored inbales or silage bags at a wide variety of moisture levels (28to 51%). The specificity of recA-directed primers has alsobeen recently demonstrated in multiplex PCR assays fordifferentiation and detection (but not quantitation) of 16different Lactobacillus species in sourdough bread, usingconventional PCR methods (Settanni et al. 2005). Thepresent work extends the use of recA-based primers todifferent LAB species in a different habitat and using themore quantitative technique of RT-PCR.

Experiments with alfalfa in minisilos indicated that eachof the four commercial silage inoculants provided higherquality silage (lower pH and higher lactic acid content)than the control silage prepared with only the epiphyticmicrobial population. Persistence beyond the initial inoc-ulation varied with inoculant, in accord with earlier work

that showed a lack of competitiveness by, or persistence of,silage inoculant strains depending on both the inoculantand environmental conditions (Rooke and Kafilzadeh1994). It appears that inoculants fulfill their role as startercultures that strongly influence the earliest stage of silagefermentation, after which their persistence may or may notbe necessary. Both inoculated and control silages displayedrelatively low LAB populations during the first 24 h, butsubstantial increases thereafter, with L. brevis beginningto displace L. plantarum within 48 h of inoculation. Theresults are in accord with previous population studiesperformed by culture-dependent methods that indicate thatL. brevis and other heterofermentative lactobacilli tend toappear somewhat later in the fermentation than do someother LAB species (Lin et al. 1992; Lindgren et al. 1985;Seale et al. 1986).

Experiments with uninoculated corn stover, conservedeither as bags or bales and at various initial moisturecontents, indicate that the epiphytic population is domi-nated by L. plantarum and L. brevis. This pattern wasobserved in stover harvested using different farm imple-ments (each of which can contribute to the epiphyticmicroflora by contacting the stover with cutting blades andother surfaces), suggesting that expected variations in theepiphytic populations and in subsequent storage conditionsnevertheless translate to an enrichment of these two LABspecies. The most similar studies are of mature whole-plantcorn silage examined by culture-dependent methods. Inone study (Dellagio and Torriani 1986), species in 60-daysilage were about evenly split among homofermentativelactobacilli, pediococci, and heterofermentative lactobacil-li. The latter were comprised of L. buchneri, L. brevis, andLactobacillus fermentum. In a survey of commercial cornand alfalfa silages (Grazia and Suzzi 1984), L. plantarumwas the dominant homofermentative LAB, and L. buchneriand L. brevis were the dominant heterofermentative LABisolated.

Although additional research is necessary to relate LABpopulations, and their changes over time, to harvesting andstorage conditions, the present results indicate that RT-PCRcan effectively quantify populations of LAB in conservedforages such as alfalfa silage and corn stover. Previously,species quantification in these materials relied on inher-ently biased culture-dependent methods and extensivetesting of many individual isolates. RT-PCR avoids theseissues and further provides the advantage of exquisitesensitivity and the potential for simultaneously assayingmany different environmental samples for desired taxa—advantages that counterbalance the substantial effortrequired for purifying DNA from these samples.

Acknowledgements This research was supported by the Agricul-tural Research Service, US Department of Agriculture, CRIS projects3655-41000-004-00D and 3655-31000-018-00D. We thank DaveSpangler (Agri-King, Inc.) for providing cultures. We also thankAlltech, Inc., Chr. Hansen Biosystems, Genus Breeding Ltd., andPioneer Hi-Bred International for providing bacterial inoculantproducts.

337

References

Berthier F, Ehrlich SD (1998) Rapid species identification withintwo groups of closely related lactobacilli using PCR primersthat target the 16S/23S rRNA spacer region. FEMS MicrobiolLett 161:97–106

Bolsen KK, Ashbell G, Weinberg ZG (1996) Silage fermentationand silage additives: review. Asian-Austr J Anim Sci 9:483–493

Coeuret V, Gueguen M, Vernoux JP (2004) Numbers and strains oflactobacilli in some probiotic products. Int J Food Microbiol97:147–156

Collins MD, Rodrigues UM, Ash C, Aguirre M, Farrow JAE,Martinez-Muchia A, Phillips BA, Williams AM, Wallbanks S(1991) Phylogenetic analysis of the genus Lactobacillus andrelated lactic acid bacteria as determined by reverse transcrip-tase sequencing of 16S rDNA. FEMS Microbiol Lett 77:5–12

Collins MD, Samelis J, Metazopoulos J, Wallbanks S (1993)Taxonomic studies on some leuconostoc-like organisms fromfermented sausages: description of a new genus Weissella forthe Leuconostoc paramesenteroides group of species. J ApplBacteriol 75:595–603

Dellagio F, Torriani S (1986) DNA–DNA homology, physiologicalcharacteristics and distribution of lactic acid bacteria isolatedfrom maize silage. J Appl Bacteriol 60:83–92

Felis GE, Dellaglio R, Mizzi L, Torriani S (2001) Comparativesequence analysis of a recA gene fragment brings new evidencefor a change in the taxonomy of the Lactobacillus casei group.Int J Syst Evol Microbiol 51:2113–2117

Furet JP, Quenee P, Talliez P (2002) Quantification de bacterieslactiques par PCR quantitative. Sci Aliments 22:33–44

Grattepanche F, Lacroix C, Audet P, Lapointe G (2004) Quantifica-tion by real-time PCR of Lactococcus lactis subsp. cremoris inmilk fermented by a mixed culture. Appl Microbiol Biotechnol66:414–421

Grazia L, Suzzi G (1984) A survey of lactic acid bacteria in Italiansilage. J Appl Bacteriol 56:373–379

Klocke M, Mundt K, Idler, C, McEniry J, O’Kiely P, Barth S (2005)Monitoring Lactobacillus plantarum in grass silages with theaid of 16S rDNA-based quantitative real-time PCR assays. SystAppl Microbiol (in press; DOI:10.1016/j.syapm.2005.06.001)

Lin C, Bolsen KK, Brent BE, Dung DYC (1992) Epiphytic lacticacid bacteria succession during the pre-ensiling and ensilingperiods of alfalfa and maize. J Appl Bacteriol 73:375–387

Lindgren S, Pettersson K, Jonsson A, Lingvall P, Kaspersson A(1985) Silage inoculation: selected strains, temperature, wiltingand practical application. Swed J Agric Res 15:9–18

Matsuki T, Watanabe K, Fujimoto J, Kado Y, Takada T, MatsumotoK, Tanaka R (2004a) Quantitative PCR with 16S rRNA-gene-targeted species-specific primers for analysis of human intes-tinal bifidobacteria. Appl Environ Microbiol 70:167–173

Matsuki T, Watanabe K, Fujimoto J, Takada T, Tanaka R (2004b)Use of 16S rRNA-gene-targeted group-specific primers for realtime PCR analysis of predominant bacteria in human feces.Appl Environ Microbiol 70:7220–7228

McDonald P, Henderson N, Heron S (1991) The biochemistry ofsilage. Chalcombe, Marlow

Muck RE (1991) Silage fermentation. In: Zeikus JG, Johnson EA(eds) Mixed cultures in biotechnology. McGraw-Hill, NewYork, pp 198–204

Rooke JA, Kafilzadeh F (1994) The effect upon fermentation andnutritive value of silages produced after treatment by threedifferent inoculants of lactic acid bacteria applied alone or incombination. Grass Forage Sci 49:324–333

Sambrook J, Fritsch EF, Maniatis T (1989) Molecular cloning: alaboratory manual, 2nd edn. Cold Spring Harbor Press, NewYork

Seale DR, Henderson AR, Pettersson KO, Lowe JF (1986) Theeffect of addition of sugar and inoculation with two commercialinoculants on the fermentation of lucerne silage in laboratorysilos. Grass Forage Sci 41:61–70

Settanni L, van Sindern D, Rossi J, Corsetti A (2005) Rapiddifferentiation and in situ detection of 16 sourdough Lactoba-cillus species by multiplex PCR. Appl Environ Microbiol71:3049–3059

Tajima K, Aminov RI, Nagamine T, Matsui H, Nakamura M, BennoY (2001) Diet-dependent shifts in the bacterial population ofthe rumen revealed with real-time PCR. Appl Environ Micro-biol 67:2766–2774

Torriani S, Felis GE, Dellaglio F (2001) Differentiation ofLactobacillus plantarum, L. pentosus, and L. paraplantarumby recA gene sequence analysis and multiplex PCR assay withrecA gene-derived primers. Appl Environ Microbiol 67:3450–3454

338