Differential Real-Time PCR Assay for Enumeration of Lactic Acid Bacteria in Wine

Abstract

Oenococcus oeni is often employed to perform the malolactic fermentation in wine production, while nonoenococcal lactic acid bacteria often contribute to wine spoilage. Two real-time PCR assays were developed to enumerate the total, and nonoenococcal, lactic acid bacterial populations in wine. Used together, these assays can assess the spoilage risk of juice or wine from lactic acid bacteria.

The lactic acid bacteria (LAB) found in wine are comprised primarily of species in the genera Lactobacillus, Pediococcus, and Oenococcus (formerly Leuconostoc oenos) (3, 12, 17). LAB are a constant concern to winemakers as they can impact the quality of wine negatively, by acting as spoilage organisms, or positively, by conducting malolactic fermentation (2). While the LAB population in wines with pH values below 3.5 tends to be mostly Oenococcus oeni, higher-pH wines often harbor species of Lactobacillus and Pediococcus during and after fermentation (4, 6, 10). Nonoenococcal LAB in wine often contribute undesirable aromas and flavors (12). Conversely, O. oeni is the inoculum of choice for performing the malolactic fermentation due to its greater tolerance to wine conditions (a pH of <3.5 and >10% ethanol) and because it is less prone to off-flavor production than other wine-related LAB (18).

The traditional methods for enumerating LAB in wine involve time-consuming plating techniques (9). Real-time or quantitative PCR (QPCR) methods have been developed to enumerate several species of LAB, including those found in wine (5, 8, 16); however, no QPCR methods that quantify wine-related LAB as a group have been developed. In an earlier work, we examined various primer sets for analysis of LAB populations in wine by using PCR-denaturing gradient gel electrophoresis (13, 20). Previous database analysis of the WLAB1-WLAB2 (WLAB1-2) (13) primer set indicated that this set describes 75% of the Bacillus-Lactobacillus-Streptococcus subgroup (including 90% of the Lactobacillus subgroup) but did not describe the Proteobacteria subgroup (including wine-related acetic acid bacteria). Walter et al. (20) previously characterized the Lac1-Lac2 (Lac1-2) primer set as group specific for Lactobacillus, Pediococcus, Leuconostoc, and Weisella species; however, when empirically tested, we noted that the primer set did not amplify O. oeni (13).

In this work, QPCR assays were generated with both the WLAB1-2 and Lac1-2 primer sets. By comparing the results of the two QPCR assays, an estimate of the percentage of nonoenococcal LAB present is obtained, thereby allowing a rapid assessment of the spoilage potential of the LAB population in juices or wines.

QPCR was performed on an Applied Biosystems Prism 7700 sequence detection system. Primer concentrations and PCR conditions were optimized for 25-μl reaction mixtures in a 96-well plate. The final reagent concentrations were 1× SYBR green master mix (Applied Biosystems, Foster City, CA), 9.5 μl of 10-fold-diluted DNA, and 600 nM for each primer. All reactions were run for 40 cycles, and the cycle threshold (C_T) value was set automatically using Sequence Detection Software (version 1.7; Applied Biosystems, Foster City, CA). QPCR thermal cycler conditions for both primer sets were 40 cycles at 95°C (melting) for 60 seconds, 63°C (annealing) for 30 seconds, and 72°C (extension) for 60 seconds. Melting curve analysis was performed to assess the specificities of the amplicons.

The target and nontarget LAB strains examined were cultivated at 30°C in MRS medium (Becton Dickinson, Sparks, MD) or, in the case of O. oeni UCD21 (PSU-1), in MRS-MF (MRS with the addition of 5 g/liter fructose and 1 g/liter malic acid). Saccharomyces cerevisiae UCD522 was cultivated in yeast malt broth (Becton Dickinson). DNA was purified from the cell pellet by using an Epicentre MasterPure gram-positive DNA purification kit (Epicentre, Madison, WI) as per the manufacturer's instructions except for the inclusion of 5 units per μl of mutanolysin (Sigma-Aldrich) with the lysozyme incubation.

As indicated in Table 1, the WLAB1-2 primer set readily amplifies Lactobacillus plantarum UCD1 and O. oeni at low cell densities but did not amplify S. cerevisiae at high cell densities. WLAB1-2 minimally amplified acetic acid bacteria (Gluconobacter oxydans UCD217 and Acetobacter aceti UCD114) at high cell densities; however, the C_T values were quite high, and melting curve analysis indicated that these amplicons were false positives. The Lac1-2 primer set was a bit more selective—there was no significant amplification of yeast or acetic acid bacterial DNA.

TABLE 1.

Specificities of WLAB1-2 and Lac1-2 primers^a

Microorganism	Plating no. of CFU/ml	CT value (40 cycles total)
		WLAB1-2	Lac1-2
Lactobacillus plantarum	5.3E+07	14.30	14.25
Lactobacillus plantarum	5.3E+06	19.15	18.35
Lactobacillus plantarum	5.3E+05	24.13	22.44
Oenococcus oeni	7.1E+05	25.27	40.00
Gluconobacter oxydans	1.5E+08	36.47	40.00
Acetobacter aceti	2.1E+08	34.20	40.00
Saccharomyces cerevisiae	6.1E+07	40.00	40.00

^aC_T and plating values are averages of results from three replicates.

Significant amounts of nontarget DNA can interfere with endpoint PCR by nonspecific binding of primers and PCR reagents (1, 11, 13). While both QPCR assays did not significantly amplify nontarget DNA known to be in wine, it was necessary to determine if the accuracy of the target QPCR assay was impacted by the presence of high concentrations of nontarget DNA. To examine this, target LAB cells were serially diluted in sterilely filtered red wine (2003 Mourvedre, pH 3.56, 13.6% [vol/vol] ethanol) or the same wine containing either 10⁸ CFU per ml of S. cerevisiae or 10⁸ CFU per ml of an acetic acid bacterium (A. aceti or G. oxydans). The serially diluted target LAB cells were then plated for CFU enumeration, and DNA was purified for QPCR. DNA purification was similar to that described above except that zirconium hydroxide (∼7 g/liter) was added to facilitate pelleting of the bacteria in wine (14).

QPCR results for both assays of different target LAB species exhibited good correlations with the plating results (Fig. 1) and were not altered by the presence of the 10,000-fold-greater cell concentration of S. cerevisiae, G. oxydans, or A. aceti. The relationship between the number of CFU per ml and the C_T value changed slightly between different species of LAB (Fig. 1). This change could be due to a variation in the number of 16S rRNA gene copies between bacterial species (7, 15) or to a slight heterogeneity in the 16S rRNA gene sequences between target organisms or even among 16S rRNA gene copies within the same organism (19). Moreover, differential lysis of gram-positive cells during the DNA purification procedure may contribute to differences in DNA extraction efficiency.

FIG. 1.

Determination of the specificity of WLAB1-2 (A) and Lac1-2 (B) primer sets on the amplification of DNA purified from wine containing LAB species with or without the presence of nontarget microbes. The strains used were L. plantarum UCD1 (•), L. plantarum UCD1 plus 10⁸/ml A. aceti UCD114 (○), L. plantarum UCD1 plus 10⁸/ml O. oeni UCD21 (✠), Lactobillus brevis ATCC 14869 (▪), L. brevis plus 10⁸/ml G. oxydans (□), O. oeni UCD21 plus 10⁸/ml S. cerevisiae UCD522 (▴), and Pediococcus pentasaceus ATCC 25745 plus 10⁸/ml S. cerevisiae (Δ). Negative controls were DNA of the nontarget DNA in wine alone: G. oxydans UCD217 (+), A. aceti UCD114 (×), S. cerevisiae UCD522 (−), and O. oeni UCD21 (▴). Correlations of C_T value to numbers of CFU/ml of the plating data were an R² of 0.912 (WLAB1-2) and an R² of 0.815 (Lac1-2).

To further examine these assays, DNA was purified from 11 spoiled wines, and the LAB population was determined by plating and by QPCR using a standard curve generated with L. plantarum (Fig. 2). QPCR consistently gave larger estimations of LAB populations than the plating results (Table 2). This is likely due to the fact that QPCR will amplify DNA from all cells in wine, be they living, metabolically active but nonculturable, or dead. For the majority of wines displaying cell growth on a plate, the colony morphotypes were identical, and microscopic examination of the wine revealed the cells to be an apparent monoculture. In an effort to identify the dominant population, the end product amplicon generated in the QPCR was sequenced. The WLAB1-2 amplicon sequence identification correlated well with the QPCR prediction of whether or not O. oeni was the dominant LAB (Table 2). Moreover, the morphology of the dominant bacterial population corresponded with the genera obtained by WLAB1-2 sequence identification (i.e., lactobacilli [rods], pediococci [cocci often in tetrads], or oenococci [small cocci often in chains]). The sequences of Lac1-2 amplicons revealed the nonoenococcal LAB populations in these wines (Table 2). In some cases where the Lac1-2-derived population was less than 0.1% of that obtained with the WLAB1-2 assay, the sequence analysis indicated O. oeni. However, other samples with similar ratios revealed other LAB species, suggesting that minimal amplification of O. oeni does occur in the Lac1-2 assay, but the presence of a low number (0.1%) of a non-LAB species is still readily detected.

FIG. 2.

QPCR standard curve of L. plantarum cells serially diluted in wine. The WLAB1-2 primer set (▪) had an R² of 0.996, and the Lac1-2 primer set (□) had an R² of 0.995. C_T values are averages of results from three replicates.

TABLE 2.

Enumeration by plating and QPCR of lactic acid bacteria in spoiled wines^a

Wine varietal	Mean no. of LAB CFU/ml ± SD			% Non-O. oeni organisms	Amplicon sequence ID
	Plating	WLAB1-2	Lac1-2		WLAB1-2	Lac1-2
Cabernet sauvignon	NG	5.5E+06 ± 1.9E+06	3.3E+03 ± 2.6E+03	0.1	O. oeni	Lactobacillus sp.
Columbard	1.1E+05 ± 1.0E+04	6.8E+06 ± 1.8E+06	4.6E+04 ± 2.4E+04	0.7	O. oeni	Mixed
Grenache noir	NG	2.8E+06 ± 1.0E+06	4.5E+03 ± 3.9E+03	0.2	O. oeni	Pediococcus parvulus
Merlot	NG	1.5E+07 ± 3.3E+06	8.1E+03 ± 9.7E+03	0.1	O. oeni	O. oeni
Sangiovese	NG	7.8E+03 ± 2.3E+03	3.1E+03 ± 5.9E+02
Syrah-1	2.3E+05 ± 1.5E+04	1.5E+07 ± 6.5E+06	1.4E+07 ± 9.3E+06	95.6	Pediococcus parvulus	Pediococcus parvulus
Zinfandel	NG	7.9E+03 ± 7.0E+03	1.2E+04 ± 4.8E+03
White blend 1	NG	4.7E+03 ± 6.3E+03	1.5E+03 ± 8.1E+02
White blend 2	NG	1.0E+04 ± 1.2E+04	1.1E+03 ± 1.2E+03	13.3	Mixed	Pediococcus
Mourvedre	NG	2.3E+05 ± 2.9E+04	1.4E+04 ± 4.5E+03	6.2	O. oeni	Lactobacillus kunkeei
Mourvedre SF	NG	1.2E+03 ± 7.8E+02	6.2E+02 ± 8.1E+00
Syrah-2	9.0E+04 ± 1.8E+04	4.3E+07 ± 1.3E+07	4.0E+07 ± 2.5E+07	94.4	Pediococcus parvulus	Pediococcus parvulus
Syrah-2 SF	NG	3.6E+03 ± 4.6E+03	1.3E+03 ± 1.2E+03

^aFor those wines with significant LAB populations (>10⁴ CFU/ml), prediction of the percentage of non-O. oeni LAB in the wine is given as well as a sequence identification (via BLAST analysis) for the WLAB1-2 and Lac1-2 QPCR assays. For each wine, a minimum of two separate QPCR assays were run in triplicate. NG, no growth; SF, sterile filtration; ID, identification.

Interestingly, both assays recorded LAB populations at the lower threshold (<10³ cells per ml) in Mourvedre and Syrah wines that had been passed through a 0.2-μm filter (Table 2). In both cases the QPCR results for the unfiltered wine were higher (∼1.5 log in the case of Mourvedre and ∼4 log in the case of Syrah). Since “free” DNA in the wine would pass through the filter unhindered, this finding suggests that the majority of DNA in wine is cell associated.

During wine production it is useful to know (i) if the total LAB concentration is high enough to start significant malolactic conversion and (ii) whether or not that LAB population is O. oeni. The differential QPCR assays developed here—which can be run in the same QPCR plate using the same thermal cycler conditions—allow for a rapid and quantitative determination of the concentration of total LAB and non-O. oeni LAB in a wine and thereby facilitate wine-processing decisions.