Decarboxylation of Substituted Cinnamic Acids by Lactic Acid Bacteria ...

3 downloads 0 Views 669KB Size Report
cinnamic acid (p-coumaric acid) decarboxylase. With the exception of L. hilgardii, these bacteria decarboxy- lated p-coumaric acid and/or ferulic acid, with the ...
APPLIED AND ENVIRONMENTAL MICROBIOLOGY, Dec. 2000, p. 5322–5328 0099-2240/00/$04.00⫹0 Copyright © 2000, American Society for Microbiology. All Rights Reserved.

Vol. 66, No. 12

Decarboxylation of Substituted Cinnamic Acids by Lactic Acid Bacteria Isolated during Malt Whisky Fermentation SYLVIE

VAN

BEEK

AND

FERGUS G. PRIEST*

International Centre for Brewing and Distilling, Department of Biological Sciences, Heriot-Watt University, EH14 4AS Edinburgh, Scotland Received 2 May 2000/Accepted 22 September 2000

Seven strains of Lactobacillus isolated from malt whisky fermentations and representing Lactobacillus brevis, L. crispatus, L. fermentum, L. hilgardii, L. paracasei, L. pentosus, and L. plantarum contained genes for hydroxycinnamic acid (p-coumaric acid) decarboxylase. With the exception of L. hilgardii, these bacteria decarboxylated p-coumaric acid and/or ferulic acid, with the production of 4-vinylphenol and/or 4-vinylguaiacol, respectively, although the relative activities on the two substrates varied between strains. The addition of p-coumaric acid or ferulic acid to cultures of L. pentosus in MRS broth induced hydroxycinnamic acid decarboxylase mRNA within 5 min, and the gene was also induced by the indigenous components of malt wort. In a simulated distillery fermentation, a mixed culture of L. crispatus and L. pentosus in the presence of Saccharomyces cerevisiae decarboxylated added p-coumaric acid more rapidly than the yeast alone but had little activity on added ferulic acid. Moreover, we were able to demonstrate the induction of hydroxycinnamic acid decarboxylase mRNA under these conditions. However, in fermentations with no additional hydroxycinnamic acid, the bacteria lowered the final concentration of 4-vinylphenol in the fermented wort compared to the level seen in a pure-yeast fermentation. It seems likely that the combined activities of bacteria and yeast decarboxylate p-coumaric acid and then reduce 4-vinylphenol to 4-ethylphenol more effectively than either microorganism alone in pure cultures. Although we have shown that lactobacilli participate in the metabolism of phenolic compounds during malt whisky fermentations, the net result is a reduction in the concentrations of 4-vinylphenol and 4-vinylguaiacol prior to distillation. maric acid [PCA]), forming the volatile phenols 3-methoxy-4hydroxystyrene (4-vinylguaiacol [4-VG]) and 4-hydroxystyrene (4-vinylphenol [4-VP]), respectively (4) (Fig. 1). These compounds have clove- or spice-like and medicinal or phenolic flavor characteristics, respectively. Most brewing strains of Saccharomyces cerevisiae lack ferulate decarboxylase (Pof⫺) in order to minimize 4-VG and 4-VP production, because these flavors are undesirable in most beers (wheat beers are an exception). However, wild yeasts, such as Saccharomyces bayanus, are Pof⫹, produce large amounts of 4-VG and 4-VP, and are responsible for off-flavors in beers (15). Distillers ferment their worts with a mixture of cultured yeasts, the Pof status of which has not been reported, and spent brewer’s yeast, which is Pof⫺. Volatile phenols participate positively in the final aroma of whiskies to various extents, being particularly prevalent in those from the west coast island of Islay and least pronounced in those from the Speyside region in the northeastern area of Scotland (20). The compounds responsible are largely derived from the peat smoke used in the malting process and are extracted from the oakwood during maturation (12, 17), but in Speyside whiskies, very little or no peated malt is used (22). Here we explore the possibility that hydroxycinnamic acids from barley contribute to the phenolic content of whisky prepared from nonpeated malts. Lactobacillus plantarum has been shown to synthesize an inducible PCA decarboxylase (7), which converts PCA into 4-VP. In this article, we establish the wide distribution of similar pdc genes in various strains of Lactobacillus isolated from whisky fermentations (including L. brevis, L. crispatus, L. fermentum, L. hilgardii, L. paracasei, L. pentosus, and L. plantarum) and show that some of these genes are expressed, that the bacteria decarboxylate PCA during laboratory-scale fermentations, and that bacterial PCA metabolism may contribute to the final flavor of the spirit.

In the early stages of the production of malt whisky, the hot water extract of the malt (wort) is not boiled as it is in a brewery. In this way, the activity of the soluble enzymes from the malt is retained during the fermentation to maximize alcohol yield. Consequently, bacteria from the malt that can survive mashing (62 to 63°C for at least 30 min) enter the fermentation (14). Lactic acid bacteria dominate the bacterial flora of the fermentation because of their heat tolerance and ability to metabolize and multiply under the low-pH and anaerobic conditions of the fermentation. The presence of these bacteria can affect yeast fermentation in various ways. If large numbers enter the fermentation, they compete for nutrients with yeast cells, reducing yeast growth and ethanol yield (8, 10). Moreover, the fermentation end products of these bacteria (principally lactate) may limit the fermentative productivity of the yeast cells (14). However, in a well-managed distillery, lactic acid bacteria flourish only during the later stages of the fermentation, when the yeast has exhausted the available nutrients and is in stationary phase (10). This “late lactic fermentation” is thought to contribute positively to whisky flavor by providing an ester note (11), but the precise details of this process are unknown. Cereals, including barley, are particularly rich in hydroxycinnamic acids, which are esterified to cell wall polysaccharides (1, 21). These compounds are released during mashing (15) and may be further metabolized during fermentation. Many microorganisms have the ability to decarboxylate substituted cinnamic acids, such as trans-4-hydroxy-3-methoxycinnamic acid (ferulic acid [FA]) and trans-4-hydroxycinnamic acid (p-cou* Corresponding author. Mailing address: International Centre for Brewing and Distilling, Department of Biological Sciences, HeriotWatt University, EH14 4AS Edinburgh, Scotland. Phone: 44 131 451 3464. Fax: 44 131 451 3009. E-mail: [email protected]. 5322

VOL. 66, 2000

DECARBOXYLATION OF SUBSTITUTED CINNAMIC ACIDS

5323

FIG. 1. Pathway for the biotransformation of cinnamic acid derivatives by L. plantarum (6).

MATERIALS AND METHODS Strains, media, and culture conditions. The bacterial strains used were isolated from malt whisky distilleries in Scotland and Japan and were identified to species level by partial 16S rRNA sequence analysis (K. L. Simpson and F. G. Priest, Abstr. 6th Symp. Lactic Acid Bacteria Genet. Metab. Appl., p. A10, 1999; S. van Beek and F. G. Priest, Abstr. 6th Symp. Lactic Acid Bacteria Genet. Metab. Appl., p. A34, 1999). All strains were maintained in MRS medium (9) containing 30% (wt/vol) glycerol at ⫺70°C and were grown statically in MRS medium at 30°C. For laboratory-scale fermentations, distiller’s wort was prepared from nonpeated malt (cultivar Derkado) to an original gravity of 14.5°P in the pilot brewery of the International Centre for Brewing and Distilling. It was pasteurized at 90°C for 15 min, clarified by centrifugation, and stored at ⫺20°C until required. Fermentations were conducted with 2-liter conical flasks containing 800 ml of wort and inoculated with pressed distiller’s M yeast (Quest International, Menstrie, United Kingdom) at 3 g (wet weight)/liter. Bacteria were added at 0.03 g (wet weight)/liter as described in Results to simulate a typical bacterial flora of about 104 to 105 cells/ml. When needed, the wort was supplemented with FA or PCA to a total concentration of 100 ␮g/ml after pasteurization. Fermentations were conducted with a water bath at 22°C for 24 h, 27°C for 24 h, and finally 33°C until 72 h to simulate the conditions in a malt distillery fermentation. Bacterial growth during fermentation was determined by spread plating appropriately diluted samples on MRS agar and incubating the samples at 37°C for 24 h. Distillation procedure. A conical flask (500 ml) containing 400 ml of fermented wort was connected to a vessel (size 45/40; KIKO, Osaka, Japan) to collect foam before the vapors passed through the lyne arms (size 24/40; KIKO) and finally to the condenser (size 24/40; KIKO). Heating was provided by a gas burner. Detection of PCA and FA decarboxylase activities by UV spectrophotometry. Bacteria were grown for 18 to 24 h in MRS broth to an optical density (OD) of ⬃1.0. The cells were harvested by centrifugation and resuspended to an OD of 1.0 in 70 mM sodium phosphate buffer (pH 6.0) containing 100 ␮g of FA or PCA per ml. The suspension was incubated at 30°C for up to 8 h. Samples were centrifuged hourly, and the supernatants were kept on ice prior to analysis. Decarboxylation activity was determined from UV scans (250 to 350 nm) (UNICAM UV-Vis spectrometer; Helios, Cambridge, United Kingdom) using absorption peaks at 286 nm for PCA and at 284 and 312 nm for FA. All data are the averages of triplicate experiments. Quantitative determination of catabolic products from FA and PCA. Cultures were supplemented with FA and/or PCA and incubated for 24 h at 30°C. Supernatants were analyzed for 4-VP, 4-ethylphenol (4-EP), 4-VG, and 4-ethylguaiacol (4-EG) using a high-performance liquid chromatography (HPLC) apparatus (Gilson, Villiers le Bel, France) composed of a 231 autosampler, a Rheodyne 7010 injector, 306 and 302 pumps with 5SC pump heads, a manometric (model 802) controller, an 811C dynamic mixer, and a Waters (Watford, United Kingdom) 420-AC fluorescence detector. Separation was carried out with an Anachem (Luton, United Kingdom) HICHROM 5ODS2 column (150 by 4.6 mm) and a gradient of 1% glacial acetic acid in water (eluent A) and 1% acetic acid in acetonitrile (eluent B) at a flow rate of 1 ml/min. For quantification, 4-VG, 4-VP, 4-EG, and 4-EP (all from Lancaster Synthesis, Morecambe, United Kingdom) were used as external standards, and trimethylphenol (Lancaster Synthesis) was used as an internal standard. The detector was set on high sensitivity at an excitation wavelength of 254 nm and an emission wavelength of 360 nm. The peak table entries were 25.5 min for 4-VP, 27.25 min for 4-VG, 28 min for 4-EP, 29.5 min for 4-EG, and 33 min for trimethylphenol. The mean

relative standard deviation for 80 duplicates, taken at random over several runs, was 2.5%. All data are the averages of triplicate experiments. Molecular methods. DNA was isolated from 1 ml of a late-exponential-phase culture (OD at 600 nm of about 1.0) in MRS medium using a PUREGENE DNA isolation kit (Philip Harris/Flowgen, Shenstone, United Kingdom) modified by the addition of 140 U of mutanolysin (Sigma) per ml to the lytic enzyme solution and incubation of the cell suspension at 37°C for 45 min. The pdc gene was detected by PCR using primers PCD 489F (5⬘-AACGGCTGGGAATACGA-3⬘) and PCD 813R (5⬘-GCAAATTCGGGTACAAC-3⬘), derived from an alignment of three decarboxylase genes: L. plantarum pdc (accession no. U63827), Bacillus pumilus ferulate decarboxylase (accession no. X84815), and Bacillus subtilis phenolic acid decarboxylase (accession no. AF017117). The PCR mixture contained 5 ␮l of 10⫻ Taq DNA polymerase buffer, 1 ␮l of deoxynucleoside triphosphate mix (12.5 mM), 2 mM MgCl2, 100 nM each primer, 20 ng of genomic template DNA, and 1 U of Taq DNA polymerase (Bioline, London, United Kingdom) in a final volume of 50 ␮l. DNA amplification was performed for 35 cycles consisting of denaturation for 1 min at 94°C, annealing for 30 s at 50°C, and elongation for 1 min at 72°C with an automated Phoenix DNA Thermocycler (Helena BioSciences, Sunderland, United Kingdom). The PCR products were sequenced using 1 ␮l of the reaction mixture as a template and the same primers as those used for the amplification. The following cycling profile was used: denaturation at 96°C for 15 s, primer annealing at 45°C for 15 s, and elongation at 60°C for 60 s. The energy transfer dye terminator chemistry supplied with the MegaBACE dye terminator ready mix (Amersham Pharmacia Biotech AB, Uppsala, Sweden) was used as described by the manufacturer for labeling the fragments. The excess of dye and buffer components was removed by ethanol precipitation. The sequencing products were separated on a MegaBACE 96 capillary sequencing system (Amersham Pharmacia Biotech AB) at 9 kV for 120 min after they were electroinjected at 3 kV for 50 s. Total RNA was extracted, treated with DNase I (Boehringer Mannheim, Lewes, United Kingdom) to eliminate any genomic DNA contamination, and purified from cells grown to an approximate OD at 600 nm of 0.7 in MRS medium or in wort (with samples adjusted to the same dry biomasses) by using an RNeasy mini kit (Qiagen, Crawley, United Kingdom). Total RNA was quantified by UV scanning (GeneQuant RNA/DNA calculator; Pharmacia, Little Chalfont, Buckinghamshire, United Kingdom). The RNA integrity was checked by standard denaturing agarose gel electrophoresis. This RNA was used as a template for reverse transcriptase (RT) PCR (RT-PCR) with an Access kit (Promega, Southampton, United Kingdom). The reaction tube contained 20 pmol each of two primers (PCD 489F and PCD 813R), 20 ␮g of template RNA, avian myeloblastosis virus RT, and substrates provided in the kit. First-strand cDNA synthesis was performed at 48°C for 45 min; inactivation of avian myeloblastosis virus RT and primer-RNA-cDNA denaturation were done at 94°C for 2 min. Second-strand cDNA synthesis and PCR amplification were accomplished during 30 cycles of denaturation at 94°C for 30 s, annealing at 50°C for 30 s, and extension at 68°C for 1 min, followed by a final extension at 68°C for 5 min. Simultaneously, an RT-PCR negative control, without RT, was run with each RNA template (data not shown). The L-lactate dehydrogenase gene (ldh) was used as a positive control for the RT-PCR products obtained with two degenerate primers designed from an alignment of five ldh sequences (GenBank accession numbers D12591, M76708, X70926, E06645, and Z81318) using CODEHOP (http://www. blocks.fhcrc.org/blocks/codehop.html) online software (19): LDH3F [5⬘-GT(CT)G G(CT)GACGG(CT)GC(CT)GTTGTTT-3⬘] and LDH2BR [5⬘-CCGATGTAGA TGTCGTTCAA-3⬘]. All PCR and RT-PCR products were analyzed by 1% agarose gel electrophoresis.

5324

VAN

BEEK AND PRIEST

APPL. ENVIRON. MICROBIOL.

TABLE 1. Strains used and their hydroxycinnamic acid decarboxylation and reduction activities

Strain

L. L. L. L. L. L. L.

brevis 113 crispatus H8 fermentum 70 hilgardii 84 paracasei 69 pentosus 128 plantarum 72

pdc GenBank accession no.

Origin

Scotland Japan Scotland Scotland Scotland Scotland Scotland

AF257164 AF257159 AF257162 AF257158 AF257160 AF257161 AF257163

% Identity with reference pdca N

P

81 99 100 80 100 99 100

89 92 100 88 100 100 100

Decarboxylation activityb UV spectroscopy

Reduction activityb

HPLC

PCAc

FAd

PCAc

FAd

PCA and FAe

4-VP

4-VG

⫹ ⫹ ⫹ ⫺ ⫺ ⫹⫹ ⫹⫹

⫺ ⫺ ⫹ ⫺ ⫹ ⫹ ⫹

ND ⫹ ⫺ ND ⫺ ⫹⫹ ⫹⫹

ND ⫹ ⫺ ND ⫺ ⫹ ⫹

ND ⫹⫹ ⫺ ND ⫹ ⫹⫹ ⫹⫹

ND ⫹ ⫺ ND ⫺ ⫺ ⫹⫹

ND ⫺ ⫺ ND ⫺ ⫺ ⫺

a

The reference pdc gene was from L. plantarum (accession no. U63827). N, nucleotide sequence; P, protein sequence. ⫹⫹, high quantity of products (35 to 95 ␮g/ml) detected by HPLC, high level of change (e.g., Fig. 2A) in UV spectroscopy; ⫹, medium quantity (5 to 35 ␮g/ml) of products detected by HPLC and some level of change (e.g., Fig. 2B) in UV spectroscopy; ⫺, low quantity (0 to 5 ␮g/ml) of products detected by HPLC and no change in level of UV spectroscopy. ND, not determined. c Cultures were exposed to 100 ␮g of PCA per ml. d Cultures were exposed to 100 ␮g of FA per ml. e Cultures were exposed to PCA and FA each at 50 ␮g/ml. b

RESULTS Decarboxylation of substituted cinnamic acids by Lactobacillus strains. All Lactobacillus strains isolated from malt whisky fermentations contained the pdc gene (Table 1). The sequences of the PCR products of about 330 bp were highly conserved with the published sequence for L. plantarum (Table 1). We therefore examined the ability of these bacteria to decarboxylate hydroxycinnamic acids by monitoring the changes in UV absorbance which accompany the removal of PCA and FA and the accumulation of 4-VP and 4-VG (7, 13). Typical results are shown in Fig. 2. The decarboxylation of PCA was indicated by the spectra shown in Fig. 2A for L. pentosus 128, in which the loss of an absorbance peak at 286 nm was accompanied by an increase at about 250 nm over an 8-h period. Most of the Lactobacillus strains decarboxylated PCA in this assay (L. hilgardii 84 and L. paracasei 69 were exceptions). FA decarboxylation was less pronounced, as shown in Fig. 2B for L. fermentum 70 and in Fig. 2C for L. pentosus 128, and was not detected in L. brevis 113, L. crispatus H8, and L. hilgardii 84 (Table 1). We analyzed the products of PCA and FA decarboxylation by selected strains of Lactobacillus using HPLC. Bacteria were incubated in distiller’s wort in the presence of PCA, FA, both PC and FA, or neither substrate at 30°C for 24 h. The culture supernatants were assayed for 4-VP and 4-VG as products of PCA and FA decarboxylation, respectively. Of the organisms tested, L. pentosus 128 produced large amounts of 4-VP (83 ␮g/ml) from PCA but showed virtually no FA decarboxylase activity (2.5 ␮g of 4-VG per ml produced from 100 ␮g of substrate per ml). When both substrates were added to the same culture, PCA decarboxylase activity again dominated over FA decarboxylase activity. L. paracasei 69 showed metabolism of hydroxycinnamic acids similar to that of L. pentosus, albeit at lower levels. L. crispatus H8, however, showed a different physiology (Fig. 3). The level of decarboxylation of hydroxycinnamic acids was low when the bacterium was induced with PCA (1.7 ␮g of 4-VP per ml and virtually no 4-VG [0.005 ␮g/ml]). However, FA induced the synthesis of a decarboxylase activity(s) which resulted in the accumulation of 11.4 ␮g of 4-VG per ml from the added FA and 0.8 ␮g of 4-VP per ml from the natural PCA in the wort. When both substrates were supplied to this bacterium, PCA was decarboxylated almost totally into the corresponding 4-VP (42.5 ␮g/ml), and

FIG. 2. Changes in UV absorbances of PCA (A) and FA (B and C) by decarboxylation activity of L. pentosus 128 (A and C) and L. fermentum 70 (B) over assay periods of 0 h (solid line), 2 h (diamonds), 5 h (asterisks), and 8 h (circles).

VOL. 66, 2000

DECARBOXYLATION OF SUBSTITUTED CINNAMIC ACIDS

5325

FIG. 4. Induction of pdc mRNA in Lactobacillus strains, as demonstrated by RT-PCR. (A) RT-PCR products from L. pentosus 128 mRNA after exposure of cells to FA at 100 ␮g/ml for 5 min (lane 1) and 30 min (lane 2) and to PCA at 100 ␮g/ml for 0 min (lane 4), 5 min (lane 5), and 30 min (lane 6). Ten microliters of RT-PCR product was loaded on the gel. Lane 3, PCR (DNA template) control; lane M, 100-bp DNA molecular size marker (Gibco). (B) RT-PCR products from L. crispatus H8 mRNA after exposure of cells to PCA at 100 ␮g/ml for 0 min (lane 2), 5 min (lane 3), and 30 min (lane 4). Twenty microliters of RT-PCR product was loaded on the gel. Lane 1, PCR (DNA template) control; lane M, 100-bp DNA molecular size marker (Gibco).

FIG. 3. Accumulation of 4-VP (■) and 4-VG (u) derivatives during growth in wort, supplemented or not supplemented with substrate, of L. pentosus 128 (A), L. casei 69 (B), and L. crispatus H8 (C). 1, with PCA at 100 ␮g/ml; 2, with FA at 100 ␮g/ml; 3, with PCA at 50 ␮g/ml and FA at 50 ␮g/ml; 4, no additional hydroxycinnamic acid.

14.3 ␮g of 4-VG per ml was produced from the added FA. It is not clear if these activities were due to a single enzyme or two separate decarboxylases. Although most organisms were unable to further metabolize the vinyl derivatives, L. plantarum and, to a lesser extent, L. crispatus reduced 4-VP to the corresponding ethyl derivative (Table 1). Expression of the Lactobacillus decarboxylase genes. Initially, we examined the expression of the pdc gene during growth in MRS medium at 37°C to avoid native hydroxycinnamic acids in the wort affecting enzyme induction. Hydroxycinnamic acids (100 ␮g/ml) were added to cultures, and samples were taken immediately and after 5 and 30 min of induction. Total RNA was extracted, and the pdc mRNA was amplified by RT-PCR. RT-PCR products were evident soon after the addition of either PCA or FA to cultures of L. pentosus 128, with PCA giving rise to a higher hybridization signal than FA (Fig. 4A). Products from L. crispatus H8, on the other hand, were faintly visible in the agarose gel when the culture was exposed to 100 ␮g of PCA per ml (Fig. 4B) and undetectable when FA was used as the inducer (data not shown), despite the earlier observation that decarboxylase activity was induced primarily by FA (Fig. 3). The experiment was repeated with bacteria in wort to determine if the natural levels of cinnamic acids could induce gene expression. Strains were grown in wort alone or induced with FA or PCA (100 ␮g/ml) for 60 min. The pdc gene was induced in L. pentosus 128 by PCA and, to a lesser extent, by FA, as noted for MRS medium above, but the indigenous hydroxycinnamic acids in wort also induced gene expression (Fig. 5A). Similar results were obtained with L. fermentum 70 (Fig. 5A), but we obtained no evidence for induction of the L. paracasei 69 or L. crispatus H8 pdc gene under these conditions. Since we had observed that wort constituents could inhibit the RT-PCR in poorly purified RNA samples, we used

the lactate dehydrogenase gene (ldh) as a positive control for our template RNA from L. paracasei 69. The product shown in Fig. 5B confirms that the template was suitable and that the lack of a pdc gene product from this bacterium indicates a lack of gene expression under these conditions. Decarboxylation of substituted cinnamic acids during yeast fermentation. Since it is not known if distiller’s M yeast has hydroxycinnamic acid decarboxylase activity, we investigated the production of decarboxylation products during the growth of pure yeast in wort. Over a period of 75 h, the yeast converted virtually all the added PCA into 4-VP but accumulated relatively little 4-VG from the FA (Fig. 6A). Distillery fermentations contain a varied Lactobacillus flora, comprising at least two species (3, 16). We therefore examined a mixed culture of L. crispatus H8 and L. pentosus 128 as typical of a distillery

FIG. 5. Induction of pdc mRNA in Lactobacillus strains grown in distiller’s wort, as demonstrated by RT-PCR. (A) RT-PCR products from L. fermentum 70 mRNA (lanes 1 to 3) and L. pentosus 128 mRNA (lanes 4 to 6) after exposure for 1 h to wort alone (lanes 1 and 4), wort supplemented with FA at 100 ␮g/ml (lanes 2 and 5), or wort supplemented with PCA at 100 ␮g/ml (lanes 3 and 6). Lane M, 100-bp DNA molecular size marker (Gibco). (B) RT-PCR products from L. casei 69 mRNA after exposure for 1 h to wort alone (lanes 2 and 4) or wort supplemented with FA at 100 ␮g/ml (lanes 1 and 3). Lane M, 100-bp DNA molecular size marker (Gibco).

5326

VAN

BEEK AND PRIEST

APPL. ENVIRON. MICROBIOL.

FIG. 6. Decarboxylation of FA and PCA in laboratory-scale wort fermentations and accumulation of 4-VP (filled symbols) and 4-VG (open symbols). (A) Distiller’s M yeast in wort with PCA at 100 ␮g/ml (triangles) and FA at 100 ␮g/ml (squares). (B) Mixed culture of L. pentosus 128 and L. crispatus H8 in wort with PCA at 100 ␮g/ml (triangles) and FA at 100 ␮g/ml (squares). (C) Distiller’s M yeast and mixed bacteria in wort with PCA at 100 ␮g/ml (triangles) and FA at 100 ␮g/ml (squares). (D) Mixed bacteria (triangles) and bacteria with distiller’s M yeast (circles) in unsupplemented wort (note the change in the y axis).

fermentation. These bacteria rapidly decarboxylated PCA into 4-VP but had little or no activity on FA (Fig. 6B). When wort supplemented with PCA was fermented by yeast in the presence of the mixed bacteria, the production of 4-VP followed the bacterial pattern, with rapid decarboxylation of PCA (Fig. 6C). In the presence of FA, however, the 4-VG level produced was typical of the levels found in the pure-yeast fermentation (Fig. 6C). In wort which had not been supplemented with hydroxycinnamic acids (Fig. 6D), the synthesis of 4-VP and 4-VG peaked after incubation for about 55 h (the typical length of a distillery fermentation is between 40 and 85 h, depending on local practice). The presence of the bacteria led to rapid decarboxylation of the native PCA in the wort, but the maximum concentration of 4-VP was lower than that in the pure-yeast fermentation. Moreover, the presence of bacteria led to the removal of 4-VG from the fermentation. Overall, in native wort with no additional hydroxycinnamic acids, the lactic acid bacteria reduced the concentrations of the volatile phenols despite the decarboxylase activity associated with these bacteria. Finally, we examined pdc gene expression from the mixed bacterial population of L. fermentum 70 and L. pentosus 128 during a yeast fermentation. Wort was inoculated with distiller’s M yeast and bacteria in the ratios used previously (see Materials and Methods), and total RNA was extracted from samples taken at the intervals shown in Fig. 7. The indigenous cinnamic acids induced the bacterial gene(s), particularly during the early stages of the fermentation, when PCA decarboxylation was noted in the presence of bacteria (Fig. 6D). These results show that the bacteria were likely to be responsible for the production of 4-VP during the early stages of the fermentation.

Recovery of volatile phenols from distilled spirits. To determine the contribution that decarboxylated hydroxycinnamic acids might make to spirit flavor, fermentations were distilled 72 h after inoculation and the complete distillate (referred to as low wines) was analyzed by HPLC. The contribution of lactobacilli (L. crispatus H8 and L. pentosus 128) was determined in the presence of yeast, with and without additional mixed FA and PCA (each at 50 ␮g/ml). The recovery of 4-VG in the low wines was higher than that of 4-VP (Table 2). Indeed, given that the distillation resulted in a 2.85-fold concentration, virtually all the 4-VG present in the fermentation was recovered in the low wines. However, the recovery of 4-VP in the low wines was lower than the initial quantity in the wort (between 42 and 56% recovery). As shown in Fig. 4D, the concentration of 4-VG was higher in the absence of bacteria in the fermentation, and this result was evident as a fourfold reduction in the low wines (1.5 ␮g/ml from the pure-yeast fermentation compared with 0.38 ␮g/ml from the mixed fermentation). On the contrary, the concentration of 4-VP was higher when bacteria accompanied the yeast, but only when hydroxycinnamic acids were added to the wort. In the absence of supplemental hydroxycinnamic acids, the concentration of 4-VP was about 2.5-fold lower in the presence of bacteria (0.25 ␮g/ml) than in the pure-yeast fermentations under the same conditions (0.66 ␮g/ml). DISCUSSION Genes for decarboxylases active on PCA from L. plantarum (6) or active on both PCA and FA from B. pumilus (23) and B. subtilis (5) have been cloned and sequenced. The enzyme from L. plantarum is restricted to PCA and caffeic acid as substrates,

VOL. 66, 2000

DECARBOXYLATION OF SUBSTITUTED CINNAMIC ACIDS

5327

FIG. 7. Growth of L. pentosus 128 and L. crispatus H8 during a laboratory-scale wort fermentation in the presence of S. cerevisiae and expression of the bacterial pdc gene(s). Samples were taken after 24, 30, 46, and 55 h for extraction of RNA, and 20 ␮l of RT-PCR product was loaded on the agarose gel.

while the B. subtilis enzyme has a broader substrate specificity, including PCA, FA, and caffeic acid. In B. pumilus, the enzyme uses only FA and PCA as substrates. Here we show that PCA decarboxylase activity is widespread among Lactobacillus species (Table 1). The high degree of homology among the partial pdc gene sequences from the lactobacilli is interesting, since it indicates a possible conservation of the active site of the enzyme or recent lateral gene transfer. Future analysis of the proximal and distal regions of the genes, in which most of the heterogeneity is thought to occur (5), will indicate the phylogeny of these genes more completely. The decarboxylation of substituted cinnamic acids can be conveniently estimated by UV spectrophotometry, which reveals the removal of substrate but does not necessarily identify the products of the enzyme reaction (4, 13). We therefore examined the production of 4-VP and 4-VG from PCA and FA, respectively, by HPLC. These complementary approaches indicated a range of decarboxylase activities among the strains. Some strains, such as L. fermentum 70, catabolized the substrates when examined in the UV assay but did not produce detectable 4-VP or 4-VG when assayed by HPLC. It is possible that this strain rapidly reduced the vinyl derivatives into the corresponding ethyl forms (4-EP and 4-EG) or conducted some other form of degradation; at least five distinct routes of microbial biotransformation of FA have been described (18). PCA induced high levels of PCA decarboxylase activity in both L. pentosus 128 and L. paracasei 69, although the latter was not evident by UV spectroscopy, perhaps due to poor transport of the substrate into the cell. These bacteria had a limited ability

to decarboxylate FA, even when induced only with FA; they resemble L. plantarum, in which PCA induces a high level of PCA decarboxylase activity but no detectable FA decarboxylase activity (7). On the other hand, L. crispatus showed a relatively high level of FA decarboxylase activity when induced by FA and little PCA decarboxylase activity when induced by PCA. Mixed induction with FA and PCA resulted in a high level of PCA decarboxylation, suggesting that the FA decarboxylase of this bacterium can use PCA as a substrate or, alternatively, that a PCA decarboxylase is induced by FA and not by PCA. However, the results obtained with the RT-PCR experiments (Fig. 5B), in which pdc gene mRNA was evident in PCA-induced but not FA-induced cells, argues against PCA decarboxylase activity being induced by FA and rather for the presence of two enzymes: a PCA decarboxylase induced by PCA and an FA decarboxylase with activity on PCA. The latter is induced by FA but is not detectable using the primers designed for the pdc genes. The results of similar studies of enzyme induction in L. plantarum with FA and PCA also could be explained only by the presence of two enzymes: a dominant PCA decarboxylase and an elusive FA decarboxylase (7). Lactic acid bacteria grow to high population densities (about 108 bacteria/ml) in whisky fermentations (10), and it is likely that bacterial PCA decarboxylase activity contributes to the flavor of the fermented wort. However, if yeast hydroxycinnamic acid decarboxylation activity is high, the bacterial contribution will be correspondingly low. Brewer’s yeast, by definition, does not synthesize hydroxycinnamic acid decarboxylase (Pof⫺), but here we show that distiller’s M yeast

TABLE 2. Recovery of 4-VP and 4-VG after distillation of laboratory-scale fermentations Vinyl compound

4-VP 4-VG a

Recovery (␮g/ml)a fromb: W

LW

W⫹CA

LW⫹CA

W⫹B

LW⫹B

W⫹B⫹CA

LW⫹B⫹CA

1.01 ⫾ 0.07 0.45 ⫾ 0.06

0.66 ⫾ 0.0 1.50 ⫾ 0.23

28.41 ⫾ 2.71 8.77 ⫾ 1.14

17.94 ⫾ 0.95 21.92 ⫾ 1.00

0.40 ⫾ 0.20 0.06 ⫾ 0.02

0.25 ⫾ 0.11 0.38 ⫾ 0.17

47.88 ⫾ 1.63 7.17 ⫾ 1.42

39.57 ⫾ 1.92 15.92 ⫾ 3.16

Mean and standard deviation for four samples. W, wort fermented with yeast; LW, the low wines resulting from distilling the fermented wort; W⫹CA, wort supplemented with FA and PCA (each at 50 ␮g/ml) and fermented with yeast; LW⫹CA, the resultant distillate; W⫹B, wort fermented with yeast and mixed L. pentosus 128-L. crispatus H8; LW⫹B, the resultant distillate; W⫹B⫹CA, fermented wort supplemented with cinnamic acids and bacteria; LW⫹B⫹CA, the resultant distillate. b

5328

VAN

BEEK AND PRIEST

has both PCA and, to a lesser extent, FA decarboxylation activities (Fig. 6A). However, the accumulation of 4-VP is relatively slow compared with the bacterial decarboxylation of PCA (Fig. 6B). In a mixed fermentation, the bacterial contribution is therefore noticeable as a rapid accumulation of 4-VP which subsequently declines. The rapid induction of the bacterial pdc gene(s) by PCA (Fig. 5) and the demonstration of pdc mRNA during fermentation (Fig. 7) support the notion of bacterial involvement in hydroxycinnamic acid decarboxylation, at least during the early stages of fermentation. In particular, the bacteria may contribute to the phenolic characteristics of fermented worts in distilleries that practice short fermentations and transfer the wort to the still once the yeast has exhausted its fermentable sugars (about 40 h after inoculation), but they are less likely to be contributory in distilleries that operate long fermentations (55 to 70 h). The most intriguing observation was that the concentrations of decarboxylated hydroxycinnamic acids were consistently lower in mixed bacterial-yeast fermentations than in pure-yeast fermentations (Fig. 6). The only explanation for this result is an interaction between bacteria and yeast. Perhaps the rapid substrate decarboxylation effected by the bacteria results in the 4-vinyl derivatives accumulating at an early stage, followed by reduction to the 4-ethyl derivatives by the yeast. If decarboxylation by the yeast is rate limiting in this process, mixed cultures will provide rapid transformation into the ethyl forms. Alternatively, one of the major differences between a pureyeast fermentation and a mixed fermentation with lactic acid bacteria is a greater reduction in pH due to lactic acid production by the bacteria (2, 14). It is possible that the reduction of 4-VP occurs more favorably under these conditions. The reduced concentrations of 4-VP in mixed bacterial-yeast fermentations were also evident after distillation. The recovery of 4-VG from a single distillation was greater than that of 4-VP (about 45% more), but it must be remembered that malt whisky undergoes two distillations, which would result in reduced levels in the final product. Therefore, while our results show that lactobacilli decarboxylate FA and PCA to produce phenolic compounds during whisky fermentation, their influence on the phenolic content of the final, matured spirit is probably small compared to the contributions of the peated malt (when used) and the oakwood maturation casks. ACKNOWLEDGMENTS We are grateful to Hisato Ikemoto for providing strains and for many stimulating discussions by e-mail, James MacKinlay for HPLC analysis, and Bertil Pettersson for DNA sequencing. Sylvie van Beek thanks Suntory Ltd., Osaka, Japan, for financial support. REFERENCES 1. Akin, D. E., R. D. Hartley, L. L. Rigsby, and W. H. Morrison III. 1992. Phenolic acids released from bermudagrass (Cynodon dactylon) by sequential sodium hydroxide treatment in relation to biodegradation of cell types. J. Sci. Food Agric. 58:207–214. 2. Barbour, E. A., and F. G. Priest. 1986. The preservation of lactobacilli: a

APPL. ENVIRON. MICROBIOL. comparison of three methods. Lett. Appl. Microbiol. 2:69–71. 3. Bryan-Jones, G. 1975. Lactic acid bacteria in distillery fermentations, p. 165–176. In J. G. Carr, C. V. Cutting, and G. C. Whiting (ed.), Lactic acid bacteria in beverages and foods. Academic Press Ltd., London, England. 4. Cavin, J. F., V. Andioc, P. X. Etievant, and C. Divies. 1993. Ability of wine lactic acid bacteria to metabolize phenol carboxylic acids. Am. J. Enol. Vitic. 44:76–80. 5. Cavin, J. F., V. Dartois, and C. Divies. 1998. Gene cloning, transcriptional analysis, purification, and characterization of phenolic acid decarboxylase from Bacillus subtilis. Appl. Environ. Microbiol. 64:1466–1471. 6. Cavin, J. F., L. Barthelmebs, and C. Divies. 1997. Molecular characterization of an inducible p-coumaric acid decarboxylase from Lactobacillus plantarum: gene cloning, transcriptional analysis, overexpression in Escherichia coli, purification, and characterization. Appl. Environ. Microbiol. 63:1939–1944. 7. Cavin, J. F., L. Barthelmebs, J. Guzzo, J. Van Beeumen, S. Bart, J. F. Travers, and C. Divies. 1997. Purification and characterization of an inducible p-coumaric acid decarboxylase from Lactobacillus plantarum. FEMS Microbiol. Lett. 147:291–295. 8. Chin, P. M., and W. M. Ingledew. 1994. Effect of lactic acid bacteria on wheat mash fermentations prepared with laboratory backset. Enzyme Microb. Technol. 16:311–317. 9. De Man, P. J., M. Rogosa, and M. Sharpe. 1960. A medium for the cultivation of lactobacilli. J. Appl. Bacteriol. 23:130–135. 10. Dolan, T. C. S. 1976. Some aspects of the impact of brewing science on Scotch malt whisky production. J. Inst. Brew. 82:177–181. 11. Geddes, P. A., and H. L. Riffkin. 1989. Influence of lactic acid bacteria on aldehyde, ester and higher alcohol formation during Scotch whisky fermentations, p. 193–199. In J. R. Piggot and A. Paterson (ed.), Distilled beverage flavour. Ellis Horwood, Chichester, United Kingdom. 12. Goldberg, D. M., M. Hoffman, J. Yang, and G. J. Soleas. 1999. Phenolic constituents, furans, and total antioxidant status of distilled spirits. J. Agric. Food Chem. 47:3978–3995. 13. Lindsay, R. F., and F. G. Priest. 1975. Decarboxylation of substituted cinnamic acids by enterobacteria: the influence on beer flavour. J. Appl. Bacteriol. 39:181–187. 14. Makanjuola, D. B., A. Tymon, and D. G. Springham. 1992. Some effects of lactic acid bacteria on laboratory-scale fermentations. Enzyme Microb. Technol. 14:350–357. 15. McMurrough, I., D. Madigan, D. Donnelly, J. Hurley, A. M. Doyle, G. Hennigan, N. McNulty, and M. R. Smyth. 1996. Control of ferulic acid and 4-vinyl guaiacol in brewing. J. Inst. Brew. 102:327–332. 16. Priest, F. G., and E. A. Barbour. 1985. Numerical taxonomy of lactic acid bacteria and some related taxa, p. 137–164. In M. Goodfellow, D. Jones, and F. G. Priest (ed.), Computer-assisted bacterial systematics. Academic Press Ltd., London, England. 17. Puech, J. L., and M. Moutonnet. 1990. Oakwood chemistry and extractable substances, p. 209–225. In I. Campbell (ed.), Proceedings of the Third Aviemore Conference on Malting, Brewing and Distilling. Institute of Brewing, London, England. 18. Rosazza, J. P. N., Z. Huang, L. Dostal, T. Volm, and B. Rousseau. 1995. Biocatalytic transformation of ferulic acid: an abundant aromatic natural product. J. Ind. Microbiol. 15:457–471. 19. Rose, T. M., E. R. Schultz, J. G. Henikoff, S. Pietrokovski, C. M. McCallum, and S. Henikoff. 1998. Consensus-degenerate hybrid oligonucleotide primers for amplification of distantly-related sequences. Nucleic Acids Res. 26:1628–1635. 20. Swan, J. S., and D. Howie. 1983. Sensory and analytical studies of regional influence on the composition of Scotch malt whisky, p. 129–142. In F. G. Priest and I. Campbell (ed.), Current developments in malting, brewing and distilling. Institute of Brewing, London, England. 21. Wall, J. S., L. C. Swango, D. Tessari, and R. J. Dimler. 1961. Organic acids of barley grain. Cereal Chem. 38:407. 22. Watson, D. C. 1983. Factors influencing the congener composition of malt whisky new spirits, p. 79–92. In J. Piggott (ed.), Flavour of distilled beverages. Ellis Horwood, Chichester, United Kingdom. 23. Zago, A., G. Degrassi, and C. V. Bruschi. 1995. Cloning, sequencing, and expression in Escherichia coli of the Bacillus pumilus gene for ferulic acid decarboxylase. Appl. Environ. Microbiol. 61:4484–4486.