healthmedicinet

Complex food matrices and spoilage conditions

The canned food matrix

Canned creamy mushroom soup (canned food product, purchased in The Netherlands) was
used as food matrix for the assessment of bar-coded amplicon sequencing as a feasible
method for detection and quantification of bacterial spores. In order to validate
the method, a spike of defined bacterial spores was used. Spore crops were purchased
for G. stearothermophilus ATCC 7953 (Mesa Laboratories, Denver, CO) or produced according to methods described
by Zhao et al. 12] for B. subtilis A163 22], B. sporothermodurans IC4 23], B. cereus TNO 02.0143 (TNO food product isolate), and G. thermoglucosidans TNO-09.020 21]. Spore suspensions were mixed in equivalent amounts based on spore counts, as determined
by a hemocytometer according to methods described by Kort et al. 22]. In summary, six randomly selected squares were counted, with a surface area of 0.0025 mm
²
and a depth of 0.01 mm each. Spore suspensions containing all five species were added
to aliquots of 20 g of creamy mushroom soup in stomacher filter bags (BagFilter S/25,
Interscience, France), in tenfold dilutions to final concentrations ranging from 1?×?10
⁶
to 1?×?10
¹
spores of each species per ml canned food. After mixing the spores with the canned
food, a Pulsifier treatment was performed (1 min in a Pulsifier, LED Techno, Den Bosch,
The Netherlands) in order to release the spores or bacteria from the canned food matrix.
After passing through the filter of the stomacher bag, the samples were used for determination
of CFU counts on TSA plates incubated at 37 or 55 °C, and three samples of each mixture
(850 ?l, corresponding to 8.5?×?10
⁵
to 8.5?×?10
⁰
spores each) were taken for total DNA extraction. The outgrowth of added bacterial
spores was studied at 37 or 55 °C. Four aliquots (20 g each) of creamy mushroom soup
were spiked with the spore mixture until a final concentration of 1?×?10
⁶
spores per ml. Two spiked aliquots were incubated o/n at 37 °C and the other two spiked
aliquots were incubated o/n at 55 °C (Additional file 13), a starting number of 8.5?×?10
⁵
spores of each bacterium) for total DNA extraction.

The ready-to-eat food matrix

The bacterial outgrowth was studied in a ready-to-eat (RTE) meal consisting of fried
rice, vegetables, and meat. This meal was purchased in a local supermarket and was
packaged in a disposable plastic container, sealed airtight with a plastic cover.
The meal consisted of rice (rice, water, 62 %), leek (12 %), pork meat (7 %), ham
(pork meat, salt, potato starch, soy protein, aroma, 7 %), egg (7 %), vegetable oil,
yeast extracts, salt, and sugar. Several packages of the ready-to-eat rice meal of
the same batch were mixed and divided over five portions, with one serving as a control,
and the others treated with different organic acids, including 0.3 % propionic acid
99 % (Acros Organics, Belgium), 0.1 % potassium sorbate (Acros Organics, Belgium),
2.5 % acetic acid glacial 99 % (Fisher Scientific, United Kingdom), or 2.5 % Purac
FCC 80 (lactic acid; Corbion, The Netherlands). The pH of all five batches was adjusted
to 5.5 with HCl, and the batches were then distributed into samples of 15–20 g for
each sampling point (see Additional file 14). Each sample was placed into a plastic stomacher filter bag (BagFilter S/25, Interscience,
France). The plastic stomacher bags were sealed with a clip and stored at 7 °C until
sampling. Every 2 days, two samples were taken from the 7 °C incubator, and three
volumes of peptone-physiological salt solution (0.85 % (w/v) NaCl, 0.1 % peptone in demineralized water) were added to each portion, followed
by treatment with a stomacher (IUL Masticator, LA-Biosystems, Waalwijk, The Netherlands)
for 60 s onto TSA (Oxoid CM0131, 2 days, 37 °C for isolation of all aerobic bacteria)
and MRSA (Oxoid CM0361, 2 days, 37 °C under microaerophilic conditions for isolation
of lactic acid bacteria) plates.

Based on the observed CFU counts, samples were selected for 16S rRNA gene sequence
analysis in order to review the microbial population at that stage. All samples of
the untreated and sorbate-treated samples were selected for sequence analysis in duplicate
(i.e., from samples 1–2, 3–4, 13–14, 23–24, 33–34, 43–44, 53–54 and 7–8, 17–18,27-28,
37–38, 47–48, 57–58, indicated in bold in Additional file 14, two 300 ?l aliquots were analyzed individually). From the propionate-, acetate-,
and lactate-treated samples, only time points t?=?2 (samples 5–6, 9–10, 11–12) and t?=?12 (samples 55–56, 59–60, 61–61) were analyzed in duplicate.

Sample pre-treatment and DNA extraction

Prior to the total DNA extraction, spores were isolated from the canned food samples
by adding 150 ?l of ?-amylase (Sigma Aldrich A7595-250ML 240 l) to each 850-?l sample
and subsequent incubation for 6 min at 65 °C, followed by addition of 20 ?l proteinase
K (15 mg/ml, Sigma P2308). After 2 min of centrifugation in an Eppendorf microcentrifuge,
the supernatant was discarded and the pellets were frozen at ?20 °C for subsequent
DNA extractions. DNA extractions were performed using phenol bead beating in combination
with the Agowa Mag mini DNA extraction kit (catalogue 40401, LGC genomics, Berlin,
Germany).

To the frozen canned food pellets containing the bacterial spores, a volume of 500 ?l
phenol, 600 ?l zirconium beads, and 500 ?l Agowa lysis buffer was added. To each 300-?l
RTE meal sample, 600 ?l zirconium beads (diameter 0.1 mm, catalogue 11079101z, Biospec
Products, Bartlesville, OK), 400 ?l lysis buffer (Agowa Mag mini DNA extraction kit),
and 300 ?l phenol, pH 8.0 (Phenol solution BioUltra, catalogue P4557, Sigma Aldrich,
St Louis, MO), were added.

Mechanical disruption of bacterial cells or spores was done by bead beating for 2 min
in a mini-beadbeater-8 cell disruptor (Merlin Bio-products, Breda, The Netherlands)
at setting fast (homogenize). After bead beating, the samples were cooled on ice prior
to a 10 min 10,000 RPM (9300 RCF) centrifugation step. After another phenol extraction
step of the aqueous phase, 500 ?l of the aqueous phase (corresponding to 0.7 or 1
original volume of the ready-to-eat and canned food samples, respectively) was transferred
to a new centrifugation tube prefilled with 1000 ?l binding buffer (Agowa) and 20 ?l
magnetic beads (Agowa). After mixture, the suspension was left for 10 min to allow
binding of the chromosomal DNA to the magnetic beads. After washing the beads according
to the Agowa Mag mini DNA extraction protocol, the DNA was extracted from the beads
with 63 ?l elution buffer (Agowa) according to the manufacturer’s instructions. For
the canned food samples, this corresponds to DNA from 8.5?×?10
⁵
spores.

Bar-coded 16S amplicon sequencing

Quantitative 16S-PCR was performed to determine the relative amount of bacterial template
in the isolated DNA samples 16]. Based on this quantification, a maximum of 5 ?l of template was used for the generation
of a 16S rRNA gene amplicon library spanning variable regions V5–V7 24]. Sequence analysis of the amplicon library was performed on a 454 GS-FLX-Titanium
Sequencer (Life Sciences (Roche), Branford, CT).

Sequence processing and analysis

FASTA-formatted sequences and corresponding quality scores were extracted from the
.sff data file generated by the GS-FLX Titanium sequencer using the GS Amplicon software
package (Roche, Branford, CT) and processed using modules implemented in the Mothur
v. 1.22.2 software platform 25]. Samples yielding insufficient reads (1000) were excluded in further analysis. The
sequences were de-noised using a pseudo-single linkage algorithm with the goal of
removing sequences that are likely due to pyrosequencing errors (“pre.cluster” command)
26]. Potentially chimeric sequences were detected and removed using the “chimera.uchime”
command 27]. High-quality aligned sequences were classified using the RDP-II naive Bayesian classifier
28]. Aligned 16S rRNA gene sequences were clustered into operational taxonomic units
(OTUs) using the average linkage clustering method.

For the study on the ready-to-eat meal, OTUs were defined by 97 % identity (“3 % OTU’s”),
and each OTU was classified at tax level 6, resulting in sequence read frequencies
for 422 different genera. For the study on the canned food, OTUs were defined by 100 %
identity (facilitating comparison of unique sequence reads with the known 16S sequences
of the five applied spore isolates), resulting in read frequencies for 2037 OTUs.
For both studies the total number of sequence reads per sample was normalized to 10,000
resulting in normalized read frequencies per genus (ready-to-eat meal study) or per
OTU (canned food study).

For determination of the detection limit in the canned food experiment, the normalization
was performed by using the average number of A. encheleia sequences in the unspiked canned food samples as reference and adjusting all other
frequencies of A. encheleia in the spiked samples to this average (assuming a constant distribution of inactivated
A. encheleia cells in the single can of canned food that was used, calculations in Additional
file 3).

Calculation of detection limits amplicon sequencing in a canned food matrix

Using the above described normalized data for (non-)spiked canned food data (normalized
reads/OTU/sample with total reads/sample?=?10,000, followed by a second normalization
based on A. encheleia reads in (non-)spiked samples), we plotted 10log(reads/OTU/sample) against 10log(spores/sample)
for each of the five pairs of OTU/spore-crop and calculated the detection limit (spores/sample)
at one read/OTU using linear regression. Samples yielding no reads (at higher spore
dilutions) were excluded from the linear regression (Additional file 3). Note that detection limits linearly increase at higher read/OTU levels (e.g., tenfold
higher at ten instead of one read/OTU).

Availability of supporting data

The sequence data are available in the European Nucleotide Archive (ENA) under accession
number PRJEB7698 (http://www.ebi.ac.uk/ena/data/view/PRJEB7698).