Transcriptome analysis of beer-spoiling Lactobacillus brevis BSO 464 during growth in degassed and gassed beer

International Journal of Food Microbiology 235 (2016) 28–35

Contents lists available at ScienceDirect

International Journal of Food Microbiology journal homepage: www.elsevier.com/locate/ijfoodmicro

Transcriptome analysis of beer-spoiling Lactobacillus brevis BSO 464 during growth in degassed and gassed beer Jordyn Bergsveinson a, Vanessa Friesen b, Barry Ziola a,⁎ a b

Department of Pathology and Laboratory Medicine, College of Medicine, University of Saskatchewan, 2841 Royal University Hospital, 103 Hospital Drive, Saskatoon, SK S7N 0W8, Canada Contango Strategies Ltd., 15-410 Downey Road, Saskatoon, SK S7N 4N1, Canada

a r t i c l e

i n f o

Article history: Received 18 April 2016 Received in revised form 17 June 2016 Accepted 29 June 2016 Available online 30 June 2016 Keywords: Beer-spoilage lactic acid bacteria Biogenic amine Carbonated/pressurized beer Degassed beer Lactobacillus brevis RNAseq

a b s t r a c t Lactobacillus brevis BSO 464 (Lb464) is a beer-spoilage-related (BSR) isolate of interest given its unique physiological attributes; specifically, it is highly hop-tolerant and exhibits very rapid growth in pressurized/gassed beer. RNA sequencing was performed on Lb464 grown in pressurized and non-pressurized beer to determine important genetic mechanisms for growth in these environments. The data generated were compared against data in a previous transcriptional study of another lactic acid bacterium (LAB) during growth in beer, namely, Pediococcus claussenii ATCC BAA-344T (Pc344). Results revealed that the most important genetic elements for Lb464 growth in beer are related to biogenic amine metabolism, membrane transport and fortification, nutrient scavenging, and efficient transcriptional regulation. Comparison with the previous transcriptional study of Pc344 indicated that the total coding capacity (plasmid profile and genome size) of a LAB isolate allows for beer-spoilage virulence and adaptation to different beer environments, i.e., the ability to grow in degassed beer (during production) or gassed beer (packaged product). Further, differences in gene expression of Lb464 and Pc344 during mid-exponential growth in beer may dictate how rapidly each isolate exhausts particular carbon sources during. The presence of headspace pressure/dissolved CO2 was found to drive Lb464 transcription during midexponential growth in beer towards increasing cell wall and membrane modification, transport, osmoregulation, and DNA metabolism and transposition events. This transcriptional activity resembles transcriptional patterns or signatures observed in a viable, but non-culturable state established by non-related organisms, suggesting that Lb464 overall uses complex cellular regulation to maintain cell division and growth in the stressful beer environment. Additionally, increased expression of several hypothetical proteins, the hop-tolerance gene horC, and DNA repair and recombination genes from plasmids pLb464-2, −4, and −8 were observed in the gassed beer environment. Thus, plasmids can harbor genes with specific (gassed) beer growth advantages, and confirm that plasmid transfer and acquisition as important activities for adaptation to the beer environment. © 2016 Elsevier B.V. All rights reserved.

1. Introduction Lactic acid bacteria (LAB), and specifically isolates of Lactobacillus brevis, have long been recognized as common beer-spoilage contaminants (Back, 1994; Fuiji et al., 2005). Despite the recognized impact these beer-spoilage-related (BSR) LAB have on the brewing industry, we have limited understanding of the underlying genetic mechanisms necessary for their growth in beer, as well as the overall background genetic regulation induced by the beer environment. Determination of common, beer-specific genetic responses critical for growth in beer is important for development of screening methods for BSR organisms. Abbreviations: BSR, beer-spoilage related; DE, differential expression; CDS, coding DNA sequence; dCO2, dissolved CO2; FC, fold-change; GO, gene ontology; LAB, lactic acid bacteria; Lb464, Lactobacillus brevis BSO 464; Pc344, Pediococcus claussenii ATCC BAA344T; RNAseq, RNA sequencing; VBNC, viable, but non-culturable. ⁎ Corresponding author. E-mail addresses: [email protected] (J. Bergsveinson), [email protected] (V. Friesen), [email protected] (B. Ziola).

http://dx.doi.org/10.1016/j.ijfoodmicro.2016.06.041 0168-1605/© 2016 Elsevier B.V. All rights reserved.

Previous transcriptomic analysis (RNA sequencing; RNAseq) of beerspoilage isolate Pediococcus claussenii ATCC BAA-344T-NR (Pc344) when grown in beer and basic nutritive conditions revealed several important insights into the genetic strategy employed to grow in beer (Pittet et al., 2013; Wang et al., 2009). However, it is unclear if these same genetic pathways are important for the growth of all BSR LAB isolates. Here we present RNAseq for Lactobacillus brevis BSO 464 (Lb464) (Bergsveinson et al., 2015b), when grown in degassed and in pressurized/carbonated beer, with the goal of better defining candidate beerspoilage indicator genes or pathways that should be focused on in future transcriptomic studies with other BSR LAB. Lb464 is a highly virulent beer-spoilage isolate, containing all putative hop-tolerance genes (Bergsveinson et al., 2015b), and several BSR plasmids (Bergsveinson et al., 2015a); moreover, Lb464 can rapidly grow in and spoil pressurized/carbonated beer (Bergsveinson et al., 2015c). Thus, RNAseq analysis of Lb464 growing in both gassed and degassed beer was undertaken to delineate general BSR LAB genetic responses during growth in beer,

J. Bergsveinson et al. / International Journal of Food Microbiology 235 (2016) 28–35

to determine responses shared with Pc344 which grows in degassed beer, but not gassed beer (Bergsveinson et al., 2015c), and to determine those responses that might be isolate (Lb464)-specific in response to the beer environment. The results provide insight into which genes are more important for LAB growth in beer both before and after packaging; i.e., before and after the introduction of headspace pressure and dissolved CO2 (dCO2). 2. Materials and methods 2.1. Growth in degassed and gassed beer The growth curves of L464 in degassed beer were previously presented in Bergsveinson et al. (2015a) and methodology for determining growth in gassed beer is described in Bergsveinson et al. (2015c). In the present study, Lb464 was taken from − 80 °C stock and grown overnight in MRS pH 5.5 broth at 30 °C. Twenty μL was then passaged into 85/15 medium (85% v/v lager beer 15% v/v double-strength modified MRS broth [without Tween 80]) and grown for 2 d at 30 °C. From this culture, 100 μL were inoculated into a tube of 16 mL of freshly degassed beer (1st degassed beer), and 2.1 mL inoculated into a 4 °C, 341 mL bottle of beer, which was immediately recapped (1st gassed beer; Bergsveinson et al., 2015c). The 1st beer cultures were incubated at 30 °C for 30 h and 55 h to achieve late-log growth (Fig. 1) for degassed and gassed beer, respectively. From the 1st beer degassed culture, 6.25 mL was used to inoculate 1 L of freshly prepared degassed beer, in duplicate. For gassed beer, 4 °C bottles were opened, 8 mL of beer removed, and then 10.3 mL of 1st gassed beer culture was added, followed by immediate bottle recapping. For one replicate, three separate beer bottles were inoculated and recapped. All cultures were grown at 30 ° C until mid-exponential growth was established, taking 22 h for degassed beer and 48 h for gassed beer (Fig. 1). Lb464 grown in MRS broth was used for assessing Lb464 transcription in nutritive growth conditions against Lb464 transcription during growth in degassed and gassed beer. One mL of an overnight MRS culture grown at 30 °C was inoculated into 100 mL of mMRS broth in duplicate. Cells were harvested at mid-exponential growth at 30 °C (14 h; Fig. 1). 2.2. RNA isolation and mRNA processing Cells in the 1 L volumes for degassed and gassed beer replicates were collected by centrifugation at 4000 ×g for 10 min at room temperature. For gassed beer, the volumes from three separate beer bottles were pooled before centrifugation. Resultant cell pellets for each condition

29

were re-suspended in 35 mL of degassed or gassed beer medium and centrifuged once more at 10,000 ×g for 3 min. Cell pellets were flashfrozen with liquid N2 and stored overnight at −80 °C. For mMRS control samples, 35 mL aliquots were centrifuged at 10,000 ×g for 3 min. These cell pellets were also flash-frozen and stored overnight. For each condition, the cell pellets for one replicate were pooled at the first RNA extraction step. Total RNA isolation was done with the PowerMicrobiome™ RNA Isolation Kit (MOBIO) according to the manufacturer's instructions, except that 70% v/v ethanol was used in place of solution PM4 to prevent small RNA species (5S, tRNA and degraded RNA) from co-precipitating with mRNA and ribosomal RNA (rRNA). A 15 min on-column DNase digest was included in this protocol (DNase I, MOBIO). To the 100-μL eluate of total RNA, 1 μL of SUPERNaseIn™ RNase Inhibitor (Ambion) was added. A further DNase treatment was then performed on the eluate using 6 U of TurboDNase (Ambion). To ensure that DNA removal was complete, cDNA was prepared using the SuperScript III First-Strand Synthesis SuperMix for qRT-PCR (Invitrogen) and assessed via qPCR, along with a no-reverse transcription (noRT) control, using primers for genes proC and rpoB as previously described (Bergsveinson et al., 2012). rRNA was removed from each sample with the Ribo-Zero™ Magnetic Kit for gram-positive bacteria (Epicentre). Samples were then concentrated using RNeasy® MiniElute® Cleanup kit to a volume of 12 μL (Qiagen). Samples were assessed for rRNA removal efficiency and overall quality both pre- and post-rRNA removal with use of the Experion™ RNA StdSens Assay (BioRad) and samples were quantified for DNA and RNA via a Qubit® 2.0 Fluorometer (Invitrogen). 2.3. mRNA sequencing and processing of reads Purified mRNA was prepared with the Illumina TruSeq RNA Sample Preparation Kit and sequenced via the Illumina HiSeq platform at the National Research Council Plant Biotechnology Institute, Saskatoon, SK. Samples were indexed and multiplexed on one lane along with four other samples to achieve paired-end, 100 bp reads. The Illumina reads were visualized via FastQC version 0.9.3 for quality and Trim Galore version 0.3.3 was used to remove Illumina adaptors from read ends and reads of poor quality such that a Phred quality score of ≥30 across the library was achieved (Barbraham Bioinformatics; www. bioinformatics.babraham.ac.uk/). Any resulting reads that were b20 nucleotides (nt) long were also discarded. Bowtie 2 version 2.2.3 (run in – M mode, − very-sensitive for end-to-end alignments and –X 400 for maximum fragment length; (Langmead and Salzberg, 2012) was then used to align reads to the Lb464 genome (NCBI BioProject Accession No. PRJNA203088; Bergsveinson et al., 2015b) (Table 1). Bowtie 2 output files in SAM format were converted via SAMtools to sorted, indexed BAM files (Langmead and Salzberg, 2012; Li et al., 2009). These BAM files, along with the Lb464 genome and the Prokaryotic Genome Automatic Annotation Pipeline (PGAAP) (Tautosova et al., 2013) annotation files for Lb464 available from NCBI, were loaded into Artemis v. 14.0.0 (Carver et al., 2012) to examine read coverage across the chromosome and each plasmid. 2.4. Differential expression (DE) analysis

Fig. 1. Growth of Lb464 in mMRS, and degassed and gassed beer at 30 °C. Cells were harvested for RNA extraction when mid-exponential growth was reached (indicated by the vertical dashed lines). Error bars indicate standard deviation (N = 3).

Sorted, indexed BAM files and the Lb464 annotation file were processed using HTSeq-count version 06.1 (Anders et al., 2014) in –m union mode to produce count tables of gene features. These feature-count tables were generated using the PGAAP annotation file with all annotations for rRNA and tRNA sequences removed, and hop-tolerance genes re-annotated. Annotations of hypothetical proteins were verified and cross-referenced to a genomic annotation of Lb464 generated by the RAST server (Rapid Annotation using Subsystem Technology) (Aziz et al., 2008). DESeq 2 (v. 1.8.1) was implemented in RStudio to perform DE analysis on these read counts, with a false discovery rate (FDR) of 0.1 (Love et al., 2014). As this study aimed to elucidate general

30

J. Bergsveinson et al. / International Journal of Food Microbiology 235 (2016) 28–35

Table 1 Bowtie2 alignment of RNA sequencing reads for Lb464. Samplea

Total Paired Reads

% Aligned Readse

% rRNA Readsb

% Annotated CDSc

# Single reads mapping to CDSd

Degassed beer-I Degassed beer-II Gassed beer-I Gassed beer-II mMRS control-I mMRS control-II

14,418,996 13,954,174 13,658,680 14,240,958 13,652,921 14,027,270

98.7 99.1 99.0 98.5 94.9 91.6

15.9 18.2 0.3 0.8 81.1 77.3

75.4 75.1 89.4 90.1 12.2 12.6

21,748,136 20,985,400 24,425,970 24,620,116 3,324,704 3,537,528

a b c d e

I and II denote replicate samples. Total number of high-quality, single read fragments aligning to CDS regions. Percentage of all reads aligned to Lb464 genome using Bowtie2 alignment. Percentage of aligned paired-end reads corresponding to rRNA genes. Percentage of non-rRNA aligned paired-end reads corresponding to annotated CDS regions.

transcripts and processes induced by the beer environment, transcripts with FDR values b0.1 were taken as significant and examined further (Supplemental Tables A.1, A.2, A.3). For discussion of biologically relevant transcripts in each sample comparisons, transcripts expressed at or above 2 Log2 fold change (FC) (i.e., expression FC of ≥4 between conditions) are taken as highly biologically relevant in each set of sample comparisons. FC values are log transformed for reporting because this transformation minimizes skew in the data set by reducing variance in gene expression levels. Lastly, it should be noted that some transcripts within operons showing significant DE are expressed at levels b 2 Log2 FC (see Tables A.1 – A.3). These transcripts are not explicitly listed (Table 2) or discussed.

3. Results and discussion

samples was deemed successful by Experion analysis, Qubit quantification, and qPCR assessment. Following library preparation, the samples were multiplexed as part of a total of 12 samples in one lane and Illumina paired-end sequencing yielded a total of 170,794,473 reads with average insert size of 168 bp. After quality processing of these reads and subsequent mapping to the Lb464 genome via Bowtie 2, 91 to 99% of the remaining reads mapped to the genome (Table 1). The number of quality, non-rRNA read pairs that map to Lb464 coding DNA sequence (CDS) loci for all samples were sufficient, as previous studies have found that between 5 and 10 million nonrRNA fragments allow detection of all but a few of the least expressed genes in diverse bacteria growing under a variety of conditions (Haas et al., 2012). This study also found that the use of biological replicates provides for determination of differentially expressed genes with high statistical significance, even when the number of reads per sample is reduced to 2 to 3 million, as was the case for the mMRS control samples (Control; Table 1). Sub-sampling of the BAM files prior to DE testing only limited the number of features that exhibited coverage, which forced DESeq2 to not perform DE testing of these features. Therefore, the total number of QC reads for defined features in each sample were used for downstream DE testing (Table 1). As this study is intended to provide a qualitative, meta-understanding of how Lb464 survives in the beer environment, a FDR cutoff of 0.1 was used to determine which transcripts show significant DE.

3.1. RNA isolation, sequencing and read processing

3.2. General LAB genetic expression response to beer

Extraction of RNA was done with cells at mid-exponential growth (Fig. 1), and it is noted that in degassed beer, Lb464 is able to establish exponential growth more rapidly and multiplies to a higher level relative to growth in gassed beer (Fig. 1). Isolation of Lb464 mRNA in all

In both degassed and gassed beer growth conditions, transcripts related to agmatine deiminase and putrescine carbamoyl transferase (L747_12850 and L747_12860; energy production and pH regulation), ATPase (L747_06760 and L747_07730; maintenance of proton motive

2.5. Gene ontology (GO) annotation and enrichment analysis Lb464 proteins were annotated for gene ontology (GO) terms using Blast2GO v.3.0 by using a blastx search Expect value of 1.0− 3, and default settings for GO annotations (Altschul et al., 1990; Conesa et al., 2005). Proteins that were significantly expressed (FDR b 0.1) in DESeq 2 comparisons were taken and used to perform enrichment analysis against the complete genome GO annotation via Fisher's Exact Test in Blast2GO (using FDR b 0.1).

Table 2 Lb464 transcripts N2 Log2 FC in gassed beer vs. degassed beer.a Locus Tag

Protein

RAST annotation and/or Function

Log2 FC Location

L747_10285 L747_09230 L747_08465 L747_12895 L747_08475 L747_10935 L747_12900 L747_08470 L747_09365 L747_02045 L747_02040 L747_01900 L747_03425 L747_01235 L747_11200 L747_12445 L747_04675

Na+:H+ antiporter Inner membrane protein Membrane protein Hypothetical protein Fur family transcriptional regulator Peroxiredoxin Hypothetical protein Membrane protein D-alanyl-D-alanine carboxypeptidase Hypothetical protein Bactoprenol glucosyl transferase Metal ABC transporter substrate-binding protein Alanine glycine permease Hypothetical protein MFS transporter Hydrolase Amino acid ABC transporter permease

PMF maintenance Structural support or membrane transport Structural support or membrane transport Unknown Ferrous iron metabolite control Defense against toxic oxygen species Unknown Structural support or membrane transport Cross linking peptidoglycan in cell wall Unknown Cell membrane integrity Zinc-transporter binding protein; ZnuA Amino acid transport Unknown Multidrug transport Peptidoglycan lytic protein P45; cell wall-associated or secreted signal peptidase Amino acid transport

3.5 3.5 3.3 3.0 3.0 2.9 2.8 2.6 2.3 2.2 2.2 2.1 2.1 2.1 2.1 2.0 2.0

a

Protein annotation and predicted cellular function or role according to RAST server (Aziz et al., 2008).

chromosome chromosome chromosome chromosome chromosome chromosome chromosome chromosome chromosome chromosome chromosome chromosome chromosome pLb464–8 chromosome chromosome chromosome

J. Bergsveinson et al. / International Journal of Food Microbiology 235 (2016) 28–35

force), manganese transport protein (L747_13605; oxidative stress and potential mediation of hop stress), methionine sulfoxide reductases MsrA (L747_05475; oxidative stress), glutathione reductase (L747_11980; oxidative stress), as well as other metal transport (L747_09900) and energy homeostasis proteins (L747_06040, L747_10025, L747_05335) show significant DE relative to the mMRS control (Tables A.1 and A.2). These transcripts were also previously implicated in the transcriptomic study of Pc344 growing in beer (Pittet et al., 2013). It should be noted that the earlier Pc344 study involved slightly different beer growth conditions compared to those used here for growth of Lb464. Specifically, the control medium was MRS pH 6.5 with added Tween 80, not mMRS pH 5.5, and, although the same lager beer was used as here, the dCO2 content was not as well defined (Pittet et al., 2013). Despite these differences, however, any commonalities in genetic expression for Lb464 and Pc344 under beer growth conditions lend credence to the above noted genes being critical for LAB (non species-specific) survival in beer. 3.2.1. Metabolism of biogenic amines in beer Agamatine, putrescine, spermidine, and spermine, all have been found to be natural constituents in beer due to malt enzyme activity (Izquierdo-Oulido et al., 1994, 1996), with levels affected by raw materials, brewing techniques and microbial contamination (Beatriz and Gloria, 2005; Izquierdo-Oulido et al., 1996). Histamine (formed from histidine), tyramine (formed from tyrosine) and cadaverine (formed from lysine) can also found in beer and are considered to be specifically correlated with LAB contamination (Geissler et al., 2016; Izquierdo-Oulido et al., 1996; Kalač et al., 2002). The production of biogenic amines facilitates mediating an acidic environment and can produce metabolic energy through coupling amino acid decarboxylation with electrogenic amino acid and amino transporters (Pessione et al., 2005). The finding that agmatine and putrescine metabolism genes are actively transcribed in Lb464 and Pc344 during rapid growth in beer (Pittet et al., 2013) confirms previous work supporting biogenic amines as indicators of microbial contamination of beer (Geissler et al., 2016; Kalač et al., 2002). It is noted that the arginine deiminase pathway (L747_02720, arginine deiminase; L747_11765, ornithine carbamoyltransferase; L747_04900 carbamoyl phosphatase) are not expressed during mid-exponential Lb464 growth in beer. Nonetheless, several Lb464 biosynthetic pathways shared by histidine metabolism are up-regulated in beer, including pathways involved in purine (L747_08270), alanine (L747_05595), aspartate (L747_07260) and glutamate metabolism (L747_10940, L747_01690) (Pessione et al., 2005) (Table A.3). Further, there are several transcripts specifically involved with the export and import of biogenic amines (spermidine/putrescine ABC transporter binding protein, L747_12245 and arginine:ornithine antiporter, L747_11800), which exhibit significant DE in both beer growth conditions, with a spermidine/putrescine ABC transporter system (L747_0837008385) and an arginine:ornithine antiporter (L747_1170) both showing significant DE compared to the mMRS control. This is also in line with previous analyses demonstrating BSR L. brevis has increased uptake (consumption) of protienogenic amino acids and increased production of amines (Geissler et al., 2016). 3.2.2. Membrane transport and nutrient scavenging in beer Transcripts related to various membrane transport mechanisms (e.g., ATP-binding cassette (ABC) type, major facilitator superfamily (MFS) transporters, multidrug transporters, efflux (ion) pumps, and permeases) were found to be up-regulated in Lb464, as was earlier found for Pc344 (Pittet et al., 2013). In contrast to Pc344, however, a greater number of peptide and amino acid transport transcripts are up-regulated in Lb464 (Tables A.1 and A.2) (Pittet et al., 2013). Peptide and amino acid transport are not only coupled to biogenic amine production and nitrogen metabolism, but can also be advantageous for

31

bacterial cells to defend against osmotic stress conditions (Csonka, 1989; Pessione et al., 2005; Zaprasis et al., 2013). Lb464 expresses several genes involved in nitrogen cycling: (L747_08270, xanthine permease/purine transport; L747_04210, glutamine synthetase/nitrogen metabolism; L747_07260, asparagine synthase; L747_07255, ammonia permease/transport; L747_01930l, and aminopeptidase C/amino acid processing; L747–01500, serine protease/peptide bond cleavage). Pc344 also has coding capacity for amino acid and peptide transport and metabolism (Pittet et al., 2013), but the limited dependence by Pc344 on these transcripts relative to Lb464 may be a function of differences in genome size and coding capacity, suggesting that efficient or increased nitrogen metabolism is advantageous for rapid growth of LAB in beer. Alternatively, it is possible that the much more rapid growth of Lb464 in beer sooner exhausts the carbohydrates that Pc344 actively up-takes and metabolizes at mid-exponential growth in beer (Pittet et al., 2013). In this event, Lb464 would have to rely on scavenging for nitrogen sources, and biogenic amine production and metabolism to power cellular processes by the time mid exponential growth in beer is established. However, during mid-exponential growth, Lb464 expresses at N2 Log2 FC key genes of operons involved in metabolism of maltose (L747_07715 and L747_09905; maltose O-acetyltransferase), arabinose (pentose) uptake and metabolism (L747_12740 and L747_07720), and butanoate (L747_08660, alpha-acetolactate decarboxylase). This supports previous analyses showing BSR L. brevis to have increased consumption of arabinose (and another pentose, xylose), relative to non-BSR L. brevis strains (Geissler et al., 2016). As arabinose is found in trace amounts in beer (Goni et al., 2009), LAB having the coding capacity to import and utilize this scarce nutrient (and other pentoses) will be most successfully propagated and have opportunity to share these genetic elements with other bacteria present.

3.3. Transcriptional response in gassed beer vs degassed beer The defining difference between degassed and gassed beer is that gassed beer is maintained with roughly 150 KPa headspace pressure at 30 °C (in a standard North American beer bottle with a fermenting LAB isolate growing in it) (Bergsveinson et al., 2015c). This headspace pressure means approximately two volumes of CO2 are forced into solution in packaged beer. Neither this headspace pressure nor level of dCO2 are lethal for LAB. Nonetheless, the presence of pressure and dCO2 in packaged beer strongly influences the beer-spoilage ability of BSR LAB (Bergsveinson et al., 2015c).

Fig. 2. DESeq2 differentially expressed Lb464 genes during growth in different media. For each set of compared growth conditions, the left and right two bars correspond to genes significantly DE (FDR b 0.1) in the first compared to the second growth condition indicated, and vice versa, respectively. The actual number of genes involved in each case is given above each bar.

32

J. Bergsveinson et al. / International Journal of Food Microbiology 235 (2016) 28–35

The influence of pressure/dCO2 on cellular activities is confirmed by present transcriptional data, as a set of genes were found to have significant DE in degassed beer at N 2 Log2 FC that were not similarly expressed in gassed beer, and vice versa (Fig. 2 and Table A.3). Interestingly, there was a greater number of genes with significant DE at this level in degassed relative to gassed beer (73 vs 16 genes) (Fig. 2 and Table 2). It is important to note that genes transcribed in degassed beer are also transcribed in gassed beer, however, the presence of pressure/dCO2 in gassed beer selects for stronger transcription from other loci, thus causing the degassed-beer-specific transcripts appear to relatively down-regulated in the gassed beer dataset. Regardless, the fact that there is a set of degassed-beer-specific genes that are transcribed highlights the importance of these physiological responses in beer without increased pressure/dCO2 present. Overall, gassed-beer-specific transcripts appear to be primarily involved in synthesis of cellular membrane structural components (Tables A.1 – A.3), cell wall fortification (L747_02040, bactoprenol glucosyl transferase; and L747_07965, cell surface protein), and energy and transcriptional regulation (L747_11320, cold-shock protein/transcriptional regulation; L747_04040, 50S ribosomal protein L32; L747_08550, DEAD/DEAH box helicase; and L747_10825, ribonulceotide reductase). There are also three specific transporter proteins up-regulated (L747_10285, Na+:H+ antiporter; L747_04825, ammonia permease; and L747_11245, glycine/betaine ABC transporter), all enabling energy maintenance and osmoregulation. Interestingly, there are more plasmid-based transcripts that exhibit high significant DE in the gassed relative to the degassed beer environment, with the latter growth environment involving more strong expression from chromosomally located loci (Tables A.2 and A.3). Importantly, the shift towards cellular wall defense and osmoregulation cannot explicitly be designated as either a pressure-response or as a dCO2-specific response. Extracellular pressure undoubtedly creates osmotic stress for the bacterial cell, and can affect cell morphology and cytoskeleton integrity, and disrupt cell division (Ishii et al., 2004; Molina-Gutiérrez et al., 2002). Concurrently, dCO2 can increase cell membrane permeability, thus causing osmotic shock and disruption of cellular processes and transcription (Suzuki et al., 2006). Thus, the

resultant damage to cell wall integrity by these two stress elements likely enhances the membrane and proton motive force-damaging effects of other beer stresses such as ethanol and acidic pH. 3.3.1. GO enrichment: absence/presence of dCO2 In the absence of dCO2, transcription of genes related to cellular structural components, fatty acid synthesis, and lipid metabolism is elevated in the beer environment (Fig. 3). These findings fit with the previous transcriptional analysis of Pc344, which indicated these are important adaptations for LAB to deal with the stress of ethanol and oxidative stress (Pittet et al., 2013). Biosynthetic processes and primary cell metabolism processes are expectedly enriched given that cells were harvested for analysis during mid-exponential growth in beer. In the presence of dCO2, there is a notable enrichment (up-regulation) of transcripts involved in processes associated with DNA metabolism (both recombination and integration) and subsequent nitrogen metabolism (Fig. 3). DNA metabolism is critical for repairing cellular damage and during cell growth, suggesting that the dCO2 beer environment elicits more damage to the cell writ-large. Further, the enrichment of transcripts related to recombination, integration and transposition activity suggests that the Lb464 genome is more prone to damage under conditions of pressurized/gassed beer and therefore amenable to genomic rearrangement and gene loss. While Table 2 and Tables A.1 – A.3 indicate hypothetical proteins and membrane-associated proteins are critical for growth in gassed beer, and thus possible candidacy as selective markers for LAB survival and growth in pressurized/gassed-beer, the associated transcripts are either not amenable to GO categorization or comprise proportionally less of the transcripts showing DE during growth in gassed beer. 3.4. Viable, but non-culturable (VBNC) state and the beer environment In response to continuous culturing in beer, BSR LAB are found to enter a VBNC state, wherein cells modify their metabolism and physiology so as to survive the environmental stresses and, as a result, they lose their ability to be cultured on standard laboratory media (Suzuki et al., 2006). A VBNC state is also likely to be strongly induced in most LAB

Fig. 3. Enriched common GO functions for Lb464 in different media. Enriched GO terms were determined at a FDR level of b0.1. Proximity of boxes indicates linked GO functions. Groupings of boxes indicate related or interconnected GO terms.

J. Bergsveinson et al. / International Journal of Food Microbiology 235 (2016) 28–35

by the gassed beer environment (Bergsveinson et al., 2015c). Although Lb464 has been shown to not enter a VBNC state following one growth cycle in degassed or gassed beer (Bergsveinson et al., 2015c), it is evident that the gassed beer environment is more stressful for cellular growth than is the degassed beer environment (Fig. 1). Interestingly, the general processes up-regulated in response to the gassed beer environment closely resemble those described in recent transcriptomic studies of gram-positive Rhodococcus (Su et al., 2016) and of gram-negative Vibrio parahaemolyticus in VBNC states (Meng et al., 2015). In Rhodococcus, pathways related to butanoate metabolism, and synthesis and degradation of ketone bodies were up-regulated during the formation of a VBNC state (Su et al., 2016), as they are in Lb464 (Tables A.2 and A.3). Further, in V. parahaemolyticus, glutamine synthetase, which regulates the expression of genes involved in nitrogen metabolism, is up-regulated, as it is in Lb464 (L747_04210N2 Log2 FC in both beer condition relative to mMRS controls and 0.7 Log2 FC in gassed beer relative to degassed beer) (Tables A.1 – A.3). Glutamine synthetase enables the utilization of a variety of different nitrogen sources, and thus supports the survival of the cell (Meng et al., 2015). Activities directed at DNA metabolism, recombination, and repair were also induced in the VBNC state in V. parahaemolyticus (Meng et al., 2015), as they are in Lb464 during growth in beer (Fig. 3). Although the VBNC state of LAB is not well understood, it is clear that upon first introduction into gassed beer, Lb464 performs extreme-stress survival mechanisms that nonLAB VBNC cells have been transcriptionally shown to perform, all the while maintaining cellular division and growth. This is of note given that a true VBNC state is not established by Lb464 under the conditions (i.e., number of passages) analyzed here, in contrast to other BSR LAB that enter a VBNC state at first introduction to gassed beer (Bergsveinson et al., 2015c). Additionally, work with gram-negative Vibrio vulnificus has identified a LysR-type transcriptional regulator, OxyR, which regulates oxidative stress-related genes that when mutated result in induction of the VBNC state (Kong et al., 2004). Lb464 has four LysR transcriptional regulators, with one showing significant DE during mid-exponential growth in gassed beer relative to mMRS (L747_12175; 1.5 Log2 FC). Lb464, however, has a number of different transcriptional regulator families. DE testing between the degassed and gassed beer expression datasets reveals that transcriptional regulator Fur (ferric uptake regulator; L747_08475) is up-regulated approximately 3 Log2 FC in gassed relative to degassed beer (Table 2). This family of regulators is important in protecting against reactive oxygen species, controlling the uptake of other metals such as manganese, and bacterial virulence (Troxell and Hassan, 2013). Indirect activation of Fur can also be via small RNAs, which are up-regulated simultaneously (Troxell and Hassan, 2013). This highlights the potential complexity of the Lb464 transcriptional regulatory network for survival in gassed beer. 3.5. Transcriptional response of Lb464 plasmids and hop-tolerance genes The importance of plasmids in the beer-spoilage ability and hop tolerance of Lb464 in degassed beer has been previously demonstrated (Bergsveinson et al., 2015a). pLb464-2 was found to most greatly contribute to the beer-spoilage capacity, with pLb464-4 and pLb464-8 contributing in an ancillary way. These findings are reflected in the current transcriptional data analysis via DESeq 2 (Tables A.1 – A.3). pLb464-2 has the highest proportion of transcripts showing significant DE in the beer environment, lending credence to the idea that it is a highly niche-adapted genetic element (Fig. 4). Indeed, it is the only plasmid that has plasmid-replication genes (i.e., L747_00265; RepB) up-regulated during mid-exponential growth in both beer environments relative to in mMRS medium (Tables A.1 – A.3). Interestingly, the hop-tolerance gene, putative MFS transporter horC, is the most highly transcribed pLb464-2 transcript (4.6 Log2 FC in degassed beer and 5.4 Log2 FC in gassed beer vs in mMRS medium), while its putative transcriptional regulator horB is not transcribed higher in either beer

33

Fig. 4. Plasmid transcripts significantly DE N Log2 FC in different growth conditions. Transcripts located on pLb464-6 and -7 are not shown given they have limited coding capacity and/or are cryptic plasmids.

environment relative to mMRS medium. In fact, transcription of horB is significantly increased during Lb464 growth in mMRS medium, suggesting it is a transcriptional repressor of horC in nutritive (or nonbeer stress) conditions. Also encoded on pLb464-2 is a ferritin protein responsible for storing and controlling the release of intracellular iron (L747_00270) that is transcribed in both degassed (2.1 Log2 FC) and gassed beer (1.7 Log2 FC) relative to in mMRS medium. This finding, in conjunction with the chromosomal Fur regulator being important for growth in gassed beer relative to degassed beer (Tables A.1 – A.3), suggests that pLb464-2 provides a specific advantage to Lb464 for iron regulation. Though understanding of iron levels in beer is limited, it has been shown that iron is present in differing levels across light and dark beers, and that beer consumption correlates with increased ferritin intake (Sancho et al., 2011; Whitfield et al., 2001). Controlling environmental ferrous iron likely contributes to regulating oxidative stress, which is important during rapid LAB growth in pressurized/gassed beer. pLb464-4 and pLb464-8 previously have been suggested to provide important functions for Lb464 growth in beer that are either redundant with chromosomal processes and/or synergistic with the coding capacity of pLb464-2 (Bergsveinson et al., 2015a). These two plasmids are maintained in low plasmid copy number in degassed beer (Bergsveinson et al., 2015a), however, the transcriptional data presented here suggests that several genes located on each plasmid are induced during Lb464 growth in both degassed and gassed beer (Tables A.1 –

34

J. Bergsveinson et al. / International Journal of Food Microbiology 235 (2016) 28–35

A.3). pLb464-4 transcripts expressed in response to beer code for hypothetical or phage-related proteins, as well as several membrane and transport proteins (L747_00425, multidrug ABC transporter permease; L747_00980, MFS transporter; and L747_00905, metal ABC transporter substrate-binding protein). There are several Lb464-4 transcripts involved in either DNA excision or DNA repair that have significant DE in degassed beer (L747_00985, pyrimidine dimer DNA glycosylase; L747_00680, HNH endonuclease; L747_00785, recombinase; and L747_00900, DNA integrase). These transcripts, which range from 3.4 to 6.7 Log2 FC in degassed beer relative to mMRS controls, however, do not have significant in gassed beer. During growth in gassed beer, six hypothetical proteins, an ABC transporter binding-protein, and transposase-related transcripts are the only pLb464-4 transcripts with significant DE (Tables A.1 – A.3). In contrast to pLb464–4, pLb464–8 has several more transcripts showing significant DE during growth in gassed beer relative to degassed beer (Fig. 4). In both beer environments, transcripts with significant DE include transposase proteins, two hypothetical proteins (L747_01235 and L747_01295), a NAD (FAD)-dependent dehydrogenase (L747_01140), and a FAD-dependent pyridine nucleotide-disulfide oxidoreductase (L747_ 01290), the latter two of which both participate in oxidative stress protection. In gassed beer, there is increased transcription of DNA damage-repair protein UrvX (L747_01230) and a DeoR family global transcriptional regulator (L747_01115). Although these plasmids pose a large energy burden to the stressed Lb464 cell due to their large size, there is apparent favoritism for transcription from specific plasmids under differing stress conditions. Plasmids pLb464-5, -6, and -7 appear to play no major role or confer no advantage for Lb464 growth in beer (Fig. 4). pLb464-5 cannot be cured from the Lb464 cell (Bergsveinson et al., 2015a) and possesses genes involved in plasmid replication, genes involved in Type 1 restriction modification system (L747_01030,_01035), and a RelB antitoxin gene (L747_01015). No genes from pLb464-5 were significantly DE in either beer environment; instead, several were significantly transcribed in the mMRS growth environment (Tables A.1 – A.3). Small plasmid pLb464-6 and cryptic plasmid pLb464-7 were also both up-regulated during Lb464 growth in mMRS medium. As such, pLb464-5, -6 and -7 transcripts are likely beneficial for plasmid maintenance (i.e., through RepA proteins) or transcriptional regulation or DNA repair when growing in nutritive conditions (Tables A.1 – A.3). Interestingly, transcriptional coverage of all three plasmids took place in both degassed and gassed beer, indicating that the overall plasmid profile of Lb464 is highly stable. For pLb464-1, the hop-tolerance gene horA had significant DE in both degassed (0.7 Log2 FC) and gassed beer (1.2 Log2 FC) relative to mMRS medium. Interestingly, the cassette of genes always associated with horA did not show comparable transcriptional activity as occurred during Pc344 growth in beer (Pittet et al., 2013). On pLb464-3, the putative hop-tolerance gene, manganese transporter hitA, barely exhibits significantly DE in degassed beer (0.4 Log2 FC) and is not significantly expressed at all in gassed beer. The only pLb464-3 transcript that has significant DE in both conditions is L747_00305, a transcriptional regulator (1.2 and 0.7 Log2 FC in degassed and gassed beer, respectively). These results clearly indicate that plasmid-based gene expression is important for growth in beer, and confirm that not all plasmids present harbor genes contributing directly to the beer-spoilage phenotype (Bergsveinson et al., 2015a). The analyses presented here overall indicate that a complex multifactorial transcriptional response occurs during LAB growth in beer and should caution against the notion that there are just a few absolute genetic markers for BSR LAB. Rather, this data supports the conclusion that BSR LAB share a broad set of stress response mechanisms with non BSR LAB. The most notable differences in beer-tolerant organisms are unique membrane modifications and nutrient scavenging abilities, as well as potentially acid-tolerance through biogenic amine production and metabolism.

Supplementary data to this article can be found online at http://dx. doi.org/10.1016/j.ijfoodmicro.2016.06.041. Competing interests The authors declare they have no competing interests. Acknowledgments JB was supported through a Graduate Scholarship from the College of Graduate Studies at the University of Saskatchewan. She also was awarded the 2012 and 2013 Ecolab Scholarships, and the 2014 Roger C. Briess Scholarship by the American Society of Brewing Chemists Foundation. This study was funded through NSERC Discovery Grant 24067. References Altschul, S.F., Gish, W., Miller, W., Meyers, E.W., Lipman, D.J., 1990. Basic local alignment search tool. J. Mol. Biol. 215, 403–410. Anders, S., Pyl, T., Huber, W., 2014. HTSeq — A Python framework to work with highthroughput sequencing data. Bioinformatics 31, 166–169. Aziz, R.K., Bartels, D., Best, A.A., DeJongh, M., Disz, T., Edwards, R.A., Formsma, K., Gerdes, S., Glass, E.M., Kubal, M., Meyer, F., Olsen, G.J., Olson, R., Osterman, A.L., Overbeek, R.A., McNeil, L.K., Paarmann, D., Paczian, T., Parrello, B., Pusch, G.D., Reich, C., Stevens, R., Vassieva, O., Vonstein, V., Wilke, A., Zagnitko, O., 2008. The RAST server: rapid annotations using subsystems technology. BMC Genomics 8, 75. Back, W., 1994. Secondary contamination in the filling area. Brauwelt Int. 4, 326–328. Beatriz, M., Gloria, A., 2005. Chapter 13: Bioactive amines. In: Hui, Y.H., Sherkat, F. (Eds.), Handbook of Food Science, Technology, and Engineering vol. 1. CRC Press, Boca Raton FL, pp. 1-13–13-38. Bergsveinson, J., Pittet, V., Ziola, B., 2012. RT-qPCR analysis of putative beer-spoilage gene expression during growth of Lactobacillus brevis BSO 464 and Pediococcus claussenii ATCC BAA-344T in beer. Appl. Microbiol. Biotechnol. 96, 461–470. Bergsveinson, J., Baecker, N., Pittet, V., Ziola, B., 2015a. Role of plasmids in Lactobacillus brevis BSO 464 hop tolerance and beer spoilage. Appl. Environ. Microbiol. 81, 1234–1241. Bergsveinson, J., Pittet, V., Ewen, E., Baecker, N., Ziola, B., 2015b. Genome sequence of rapid beer-spoiling isolate Lactobacillus brevis BSO 464. Genome Announc. 3, e01411–e01415. Bergsveinson, J., Redekop, A., Zoerb, S., Ziola, B., 2015c. Dissolved carbon dioxide selects for lactic acid bacteria able to grow in and spoil packaged beer. J. Am. Soc. Brew. Chem. 73, 331–338. Carver, T., Harris, S.R., Berriman, M., Parkhill, J., McQuillan, J.A., 2012. Artemis: an integrated platform for visualization and analysis of high-throughput sequence-based experimental data. Bioinformatics 28, 464–469. Conesa, A., Gotz, S., Garcia-Gomez, J.M., Terol, J., Talon, M., Robles, M., 2005. Blast2GO: a universal tool for annotation, visualization and analysis in functional genomics research. Bioinformatics 21, 3674–3676. Csonka, L.N., 1989. Physiological and genetic responses of bacteria to osmotic stress. Microbiol. Mol. Biol. Rev. 53, 121–147. Fuiji, T., Nakashima, K., Hayashi, N., 2005. Random amplified polymorphic DNA-PCR based cloning of markers to identify the beer- spoilage strains of Lactobacillus brevis, Pediococcus damnosus, Lactobacillus collinoides and Lactobacillus coryniformis. J. Appl. Microbiol. 98, 1209–1220. Geissler, A.J., Behr, J., von Kamp, K., Vogel, R.F., 2016. Metabolic strategies of beer spoilage lactic acid bacteria in beer. Int. J. Food Microbiol. 216, 60–68. Goni, I., Diaz-Rubio, M.E., Saura-Calixto, F., 2009. Chapter 28: Dietary fiber in beer. In: Preedy, V.R. (Ed.), Beer in Health and Disease Prevention. Academic Press, San Diego, CA, pp. 299–308. Haas, B.J., Chin, M., Nusbaum, C., Birren, B.W., Livny, J., 2012. How deep is deep enough for RNA-Seq profiling of bacterial transcriptomes? BMC Genomics 13, 734. Ishii, A., Sato, T., Wachi, M., Nagai, K., Kato, C., 2004. Effects of high hydrostatic pressure on bacterial cytoskeleton FtsZ polymers in vivo and in vitro. Microbiology 150, 1965–1972. Izquierdo-Oulido, M., Marine-Font, A., Vidal-Carou, M.C., 1994. Biogenic amines formation during malting and brewing. J. Food Sci. 59, 1104–1107. Izquierdo-Oulido, M., Hernandez-Jover, T., Marine-Font, A., Vidal-Carou, M.C., 1996. Biogenic amines in European beers. J. Food Sci. 44, 3159–3163. Kalač, P., Savel, J., Kr, M., Prokopová, M., 2002. Biogenic amine formation in bottled beer. Food Chem. 79, 431–434. Kong, I., Bates, T.C., Hülsmann, A., Hassan, H., Smith, B.E., Oliver, J.D., 2004. Role of catalase and oxyR in the viable but nonculturable state of Vibrio vulnificus. FEMS Microbiol. Ecol. 50, 133–142. Langmead, B., Salzberg, S.L., 2012. Fast gapped- read alignment with Bowtie 2. Nat. Methods 9, 357–359. Li, H., Handsaker, B., Wysoker, A., Fennell, T., Ruan, J., Homer, N., Marth, G., Abecasis, G., Durbin, R., 2009. The sequence alignment/map format and SAMtools. Bioinformatics 25, 2078–2079. Love, M.I., Huber, W., Anders, S., 2014. Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol. 15, 550.

J. Bergsveinson et al. / International Journal of Food Microbiology 235 (2016) 28–35 Meng, L., Alter, T., Aho, T., Heuhn, S., 2015. Gene expression profiles of Virbio parahaemplyticus in viable but non-culturable state. FEMS Microbiol. Ecol. 91, fiv035. Molina-Gutiérrez, A., Stippl, V., Delgado, A., Gänzle, M.G., Vogel, R.F., 2002. In situ determination of the intracellular pH of Lactococcus lactis and Lactobacillus plantarum during pressure treatment. Appl. Environ. Microbiol. 68, 4399–4406. Pessione, E., Mazzoli, R., Giuffride, M.G., Lamberti, C., Garcia-Moruno, E., Barello, C., Conti, A., Giunta, C., 2005. A proteomic approach to studying biogenic amine producing lactic acid bacteria. Proteomics 5, 687–698. Pittet, V., Phister, T.G., Ziola, B., 2013. Transcriptome sequence and plasmid copy number analysis of the brewery isolate Pediococcus claussenii ATCC BAA-344T during growth in beer. PLoS One 8, e73627. Sancho, D., Blanco, C.A., Caballero, I., Pascual, A., 2011. Free iron in pale, dark and alcoholfree commercial lager beers. J. Sci. Food Agric. 91, 1142–1147. Su, X., Guo, L., Ding, L., Qu, K., Shen, C., 2016. Induction of a viable but nonculturable state in Rhodococcus and transcriptome analysis using RNA-seq. PLoS One 11, e0147596. Suzuki, K., Iijima, K., Asano, S., Kuriyama, H., Kitagawa, Y., 2006. Induction of viable but nonculturable state in beer spoilage lactic acid bacteria. J. Inst. Brew. 112, 295–301.

35

Tautosova, T., DiCuccio, M., Badretdin, A., 2013. Prokaryotic Genome Annotation Pipeline; In: The NCBI Handbook. second ed. National Center for Biotechnology Information, Bethesda, MD (http://www.ncbi.nlm.nih.gov/genome/annotation_prok/. Accessed 18 Dec 2015). Troxell, B., Hassan, H.M., 2013. Transcriptional regulation by Ferric Uptake Regulator (Fur) in pathogenic bacteria. Front. Cell. Infect. Microbiol. 3, 59. Wang, Z., Gerstein, M., Snyder, M., 2009. RNA-Seq: a revolutionary tool for transcriptomics. Nat. Rev. Genet. 10, 57–63. Whitfield, J.B., Zhu, G., Heath, A.C., Powell, L.W., Martin, N.G., 2001. Effects of alcohol consumption on indices of iron stores and of iron stores on alcohol intake markers. Alcohol. Clin. Exp. Res. 25, 1037–1075. Zaprasis, A., Brill, J., Thüring, M., Wünsche, G., Heun, M., Barzantny, H., Hoffmann, T., Bremer, E., 2013. Osmoprotection of Bacillus subtilis through import and proteolysis of proline-containing peptides. Appl. Environ. Microbiol. 79, 576–587.