Rapid differentiation and enumeration of the total, viable vegetative cell and spore content of thermophilic bacilli in milk powders with reference to Anoxybacillus flavithermus

Authors


Andreas Rueckert, Thermophile Research Unit, Department of Biological Sciences, University of Waikato, Private Bag 3105, Hamilton, New Zealand (e-mail: andreas.rueckert@alumni.tu-berlin.de).

Summary

Aims:  The development of a rapid method for the selective detection and enumeration of the total and viable vegetative cell and spore content of thermophilic bacilli in milk powder by PCR.

Methods and Results:  Quantitative PCR and microscopy indicate the presence of up to 2·9 log units more cells in milk powder than accounted for by plate counting due to the majority of cells being killed during milk processing. Two approaches for viable and dead cell differentiation of thermophilic bacilli by quantitative PCR were evaluated, these being the nucleic binding dye ethidium monoazide (EMA) and DNase I digestion. The former agent exposed to a viable culture of Anoxybacillus flavithermus caused considerable cell inactivation. In contrast, DNase I treatment had no effect on cell viability and was utilized to develop DNA extraction methods for the differential enumeration of total, viable vegetative cells and spores in milk powder. Moreover, the methods were further applied and evaluated to 41 factory powder samples taken throughout eight process runs to assess changes in numbers of vegetative cells and spores with time. DNase I treatment reduced vegetative cell numbers enumerated with PCR by up to 2·6 log units. The quantification of spores in the factory milk powders investigated indicates on average the presence of 1·2 log units more spores than determined by plate counting.

Conclusions:  The method presented in this study provides the ability to selectively enumerate the total and viable cell and spore content of reconstituted milk.

Significance and Impact of the Study:  The current study provides a tool to monitor the extent of thermophilic contamination during milk powder manufacturing 60–90 min after sampling.

Introduction

The contaminating role of thermophilic Bacillus species during the production of milk powder has been well documented (Stadhouders et al. 1982; Kwee et al. 1986; Murphy et al. 1999; Ronimus et al. 2003). In particular, strains of Anoxybacillus flavithermus strain C, Bacillus licheniformis strain F and Geobacillus stearothermophilus strain A have been found predominantly and nearly ubiquitously distributed in milk powders despite the country of origin (Rueckert et al. 2004). These thermophilic contaminants can have significant economic consequences when they exceed specification limits and may result in downgrading of the product. Typically, the number of contaminating thermophiles is determined by plate counting, producing results at least 16 h after milk processing has been terminated. With this testing regime, it is difficult to predict the optimum length of a process run yet still ensure that thermophile numbers are below specified limits. Real-time monitoring of thermophiles throughout the process run would offer a flexible and economic management strategy which allows processing to be continued until specified limits are reached.

A method for the quantitative detection of the seven most commonly occurring thermophilic milk powder bacilli including G. stearothermophilus (strain A), A. flavithermus (strains B, C and D), B. licheniformis (strains F and G) and Bacillus subtilis (Ronimus et al. 2003) has recently been developed (Rueckert et al. 2005). However, although PCR methods are extremely selective and accurate with regard to the amplification of a specific DNA target sequence, the major disadvantage arises in the inability to differentiate between DNA originating from viable or dead cells. This is problematic, in particular, with milk powder as a PCR sample. Growth of thermophiles primarily occurs in the preheaters, evaporators and heat exchangers and the majority of cells are killed by subsequent direct stream injection, concentrate formation and final spray drying (Thompson et al. 1978; Murphy et al. 1999). These dead cells still contain amplifiable DNA and would cause an overestimation of thermophile contaminant level by quantitative PCR. If no account is taken of the numbers of dead cells in any estimation of contamination, then processing runs would be terminated well before specified allowable numbers of thermophiles are met.

Nucleic acid-based methods for the differentiation of viable and dead cells have been applied using reverse transcription PCR (Herman 1997; Norton and Batt 1999) to detect viable Listeria monocytogenes. Another strategy employed by Nogva et al. (2003) and Rudi et al. (2005) uses the nucleic acid-binding dye ethidium monoazide bromide (EMA) to discriminate between living and dead cells of Escherichia coli, Salmonella sp., L. monocytogenes and Campylobacter jejuni by PCR. EMA is a DNA intercalating agent (Waring 1965; Bolton and Kearns 1978) which is reported to selectively penetrate the membrane of dead cells, but is purportedly unable to penetrate the intact membranes of live cells (Nogva et al. 2003; Rudi et al. 2005). Once photolysed, EMA covalently links to DNA and prevents PCR amplification. In addition, the utilization of DNase has been used in the assessment of mammalian cell viability by flow cytometric analysis (Frankfurt 1983) and is based on the assumption that dead cells lose membrane integrity, making their DNA accessible to DNase degradation.

The goal of this study was to adapt and apply the previously developed quantitative PCR method for enumeration of thermophilic bacilli (Rueckert et al. 2005) to milk powder samples throughout process runs to assess changes in the numbers of vegetative cells and spores of thermophilic bacilli with time, and to determine the contribution of dead cells to the PCR quantification. As a consequence, methods have been developed which allow for the selective enumeration of both viable cells and spores. The current study provides the ability to monitor the extent of thermophilic contamination during milk powder manufacturing 60–90 min after sampling.

Materials and methods

Bacterial strains and culture preparation

Geobacillus stearothermophilus strain A, A. flavithermus strain C and B. licheniformis strain F were derived from New Zealand milk powder samples (Ronimus et al. 2003). The organisms were routinely grown at 55°C in either tryptic soy broth (BactoTM; Becton Dickinson, Sparks, MD, USA) (TSB) under agitation or in tryptic soy agar (TSA) supplemented with 0·2% (w/v) soluble potato starch (Sigma; S2004). Spores of A. flavithermus C and B. licheniformis F were produced in liquid Castenholz medium (DSMZ-Deutsche Sammlung von Mikroorganismen und Zellkulturen GmbH, Braunschweig, Germany) by inoculating 1 ml of the TSB starch stationary grown precultures to 1 l of Castenholtz medium. The cultures were grown at 55°C for 72 h and the spores harvested and purified as described by Rueckert et al. (2005).

Factory powders

A total of 41 individual whole milk powder samples were derived from eight milk powder processing runs from a New Zealand dairy factory in the seasonal period between January and February 2004. The factory runs GO10, GO11, GO12, GO13, GO14, GO15 and EN22 and EN23 are successive and constitute continuous processing of 17–19 h separated by a cleaning in place (CIP) regime. Inconsistencies in the CIP regime applied prior to EN23 were reported from the plant management. The samples were collected from the powder drying belt immediately after formation and stored at room temperature until required. When samples were used for enumeration care was taken to ensure that contamination was minimized. This included processing under a laminar flow cabinet and using aerosol-resistant filter tips. The total and viable vegetative cell and spore content for each powder was determined by plate counting, microscopy and quantitative PCR (Rueckert et al. 2005).

Total plate counts and spore counts were obtained on TSA starch by pouring duplicates of decimal dilution series of reconstituted milk into warm liquid medium. The medium was allowed to solidify and the plates incubated at 55°C for 16–48 h. Spore counts were obtained by heat treatment of reconstituted milk samples in a water bath at 80°C for 20 min prior to plating. The number of vegetative cells was calculated as the difference between total count and spore count.

Ethidium monoazide bromide treatment

Five milligrams of solid EMA bromide were purchased from Biotium, Inc. (Hayward, CA, USA) and according to the manufacturer's recommendation dissolved in N,N-dimethylformamide in absence of light. Aliquots of 50 μl of the 10 mg ml−1 EMA stock solution were stored at −20°C in light-impermeable microtubes.

The effect of EMA on cell viability was investigated using a modification of the standardized protocol described by Nogva et al. (2003). Accordingly, a culture of A. flavithermus C diluted in sterile 0·9% NaCl was exposed for 5 min to 0·1, 1, 10 and 100 μg ml−1 EMA on ice, respectively. The samples were prepared in two sets of duplicates for each EMA concentration in the absence of light in clear 1·5 ml Eppendorf microtubes using a final volume of 1 ml. Subsequently, the samples were exposed for 1 min to a 500 W halogen light source (Osram, T3Q clear halogen) which was 20 cm distant from the sample tubes. Duplicates of each EMA concentration were then immediately diluted 10-fold in sterile deionized water and poured into warm liquid TSA starch. Similarly, duplicates of control samples treated in the same way using the equivalent amounts of N,N-dimethylformamide as was used for the EMA-treated samples were also included. Furthermore, duplicates of positive control cultures containing no additives but treated as outlined above were used as reference to determine the survival rates. The recovery of cell numbers was determined by colony counting. In addition, duplicates of the same experimental batches, e.g. EMA, control and positive samples were sonicated for 125 s at 120 W at 20 kHz (Liquid Processor XL-2020; Misonix, Farmingdale, NY, USA) as described by Rueckert et al. (2005). The samples were then centrifuged for 2 min at 16 100 g and PCR analysis performed on the supernatant.

DNase I treatment

The effect of DNase I (Sigma, DN25) treatment on the viability of A. flavithermus C was investigated in duplicate experiments by plate counting and quantitative PCR using a 1·1 × 105 CFU ml−1 culture diluted to this concentration in sterile 0·9% NaCl. Aliquots of 1 ml of culture were exposed for 15 min at 37°C to 0 (positive control), 50, 100, 150 and 200 Kunitz units of DNase I respectively. The samples contained additionally 100 μl 10x DNase buffer (10 mmol l−1 EDTA, 75 mmol l−1 MgCl2 and 200 mmol l−1 Tris–HCl, pH 7·5). Following incubation, duplicates of the samples were serially diluted in sterile deionized water and immediately poured into warm liquid TSA starch medium followed by incubation at 55°C for 16–24 h to determine their colony counts. Additionally, aliquots of each treatment were centrifuged at 16 000 g for 10 min and the cell pellets resuspended in 1 ml of sterile deionized water. The samples were then boiled for 10 min and sonicated for 125 s at 120 w at 20 kHz as described and quantitative PCR performed on the supernatants. All treatments were performed in duplicate.

The effect of DNase I on the viability and growth of thermophilic bacilli was extended to include G. stearothermophilus strain A and B. licheniformis strain F. For this purpose, the organisms were grown in TSB starch until the cultures reached early-mid exponential growth phase at an OD600 of approx. 0·5 (Ultraspec 3000; Pharmacia Biotech, Cambridge, UK). Subsequently, the cultures were stored overnight at 4°C and on the following morning 100 μl of each cultures was used to inoculate duplicate flasks of the following three media: (i) TSB starch, (ii) TSB starch supplemented with 5 ml of 10x DNase buffer and (iii) TSB starch supplemented with 5 ml of 10x DNase buffer and 200 Kunitz units ml−1 of DNase I. The final volume of each flask was 50 ml. The OD600 of the cultures were monitored hourly over an incubation period of 12 h at 55°C using 500 μl of culture for spectrometer reading.

DNA preparation for the enumeration of total and viable vegetative cells

Total bacterial DNA extraction from milk was performed by the method described in Rueckert et al. (2005). This included the addition of 200 μl of 1·2 mol l−1 tri-sodium citrate and 200 μl of n-decane to 1 ml of reconstituted milk (0·1 g ml−1), followed by brief vortexing and centrifugation at 16 100 g (5415 D, Eppendorf) for 10 min. The top layer of cream was carefully extracted from the microcentrifuge tube using a sterile 200 μl pipette-tip and the supernatant poured off by gently inverting the tube. The samples were re-centrifuged for 2 min at 16 100 g and the remaining supernatant removed with a pipettor, while avoiding the disruption of the cell pellet. The pellet was re-suspended in 1 ml of sterile deionized water and the sample subjected to ultra-sonication (Liquid Processor XL-2020, Misonix) for 125 s at 120 W at 20 kHz. The samples were then centrifuged for 2 min at 16 100 g and PCR analysis performed on the supernatant.

DNA from viable vegetative cells was obtained by adding DNase I to the reconstituted milk, before the extraction protocol was applied. This removed contaminating DNA from dead cells, but had no effect on DNA in viable cells. For this method, 100 μl of 10x DNase buffer and 200 Kunitz units of DNase I were added to 1 ml reconstituted milk, followed by brief vortexing and incubation at 37°C for 15 min in a water bath. Subsequently, the samples were tri-sodium citrate and n-decane extracted, re-suspended in 1 ml sterile deionized water and boiled for 10 min to denature the DNase. The DNA of viable vegetative bacteria was then released by ultra-sonication as described above.

DNA preparation for the enumeration of total spores

DNA purification from spores was accomplished by adding 100 μl of 10x DNase buffer and 200 Kunitz units of DNase I to 1 ml of reconstituted milk. The sample was then sonicated for 125 s at 120 W at 20 kHz and further incubated for 15 min at 37°C in order to release and degrade DNA of both live and dead vegetative cells. Subsequently, the sample was tri-sodium citrate and n-decane extracted and the spores re-suspended in 1 ml of sterile deionized water. Residual DNase I was inactivated by boiling the sample for 10 min and the spore DNA released by sonication for 6 min in the presence of 50 mg of 0·1 mm glass beads (Biospec Products, Bartlesville, OK, USA) at 120 W at 20 kHz (Rueckert et al. 2005). Prior to use, the glass beads were baked for 8 h at 250°C to eliminate any contaminating DNA. The samples were centrifuged for 2 min at 16 100 g and 10 μl of the supernatant used for quantitative PCR analysis.

In addition, negative controls were performed on the DNase-treated spore samples prior to heat treatment. For this purpose, the samples were centrifuged for 10 min at 16 000 g and an aliquot of 50 μl spore-free supernatant boiled and 10 μl subjected to quantitative PCR.

Evaluation of the DNA extraction methods for total, viable vegetative cells and spores

The methods for the separate detection and quantification of total, viable vegetative cells and spores in milk were evaluated in triplicate by adding three dilutions of known numbers of viable and dead cells and spores of A. flavithermus C to both reconstituted sterile milk (EN22, 1st hour) and sterile deionized water respectively. The cell and spore numbers added were chosen to simulate the bacteriological load of low, medium and high count milk powders (Ronimus et al. 2003; Rueckert et al. 2004). Viable vegetative cell and spore numbers were determined by plate counting and microscopy. Dead vegetative cells were obtained by heat-killing of an aliquot of the viable cell culture at 95°C for 10 min in a water bath. The extent of cell death was confirmed by plate counting. The different cell fractions from both media were then recovered and their DNA released as described above.

Quantitative real-time PCR

Quantitative real-time PCR analysis amplifying small regions of the 16S rRNA genes was performed with a Smart Cycler II system (Cepheid, Sunnyvale, CA, USA) in 25 μl Smart Cycler reaction tubes. The amplifications were performed using a modified protocol as described by Rueckert et al. (2005) with 1·5 units of the TaKaRa TaqTM Hot Start enzyme (TaKaRa Bio Inc., Shiga, Japan), 4 mmol l−1 MgCl2, 0·2 mmol l−1 dNTPs (TaKaRa Bio Inc.), 10x TaKaRa Taq PCR reaction buffer, 600 nmol l−1 of forward and reverse primer, 150 nmol l−1 of TaqMan probe and 10 μl of sample. The PCR reaction was cycled once at 95°C for 90 s followed by 42 repetitions at 95°C for 5 s and at 62°C for 20 s. DNA quantities of the samples were calculated using the standard curves for vegetative cells and spores (Rueckert et al. 2005).

Phase contrast and fluorescence microscopy

The total number (viable and dead) of bacterial rod forms obtained after extraction from reconstituted milk powder was determined by phase contrast microscopy (Olympus, BH2, 1000× magnification; London, UK) using a Thoma counting chamber. In order to detect spores a malachite green and safranin red (Sigma; M9015, M323950) staining protocol was employed as outlined by Clark (1973). The staining protocol was used on spores and cells extracted from reconstituted milk powder.

The differentiation of viable and dead cells of selected samples was assessed using fluorescence microscopy (Leica DMRE, 50 W Hg burner, I3 block, excitation filter BP 450–490 nm, dichroic mirror at 510 nm and long pass filter at 515 nm; Leica, Bensheim, Germany) with SYTO BC (Molecular Probes, Eugene, OR, USA; S-34855) and propidium iodide (Sigma; P4170). An exponential culture of A. flavithermus C grown in TSB starch was used as a positive control and included in the SYTO BC and propidium iodide two-colour staining assay. A control culture was additionally subjected to the extraction regime used for concentrating bacteria from milk constituents to assess the effect these treatments had on cell viability. The staining procedures were then performed according to protocols outlined by Green (1990).

Results

Effect of EMA treatment on cell viability

The effect of EMA on the viability of A. flavithermus strain C cells diluted to 3·6 × 105 CFU ml−1 was investigated in the presence of 0·1, 1, 10 and 100 μg ml−1 of EMA. In addition, the same culture was also exposed to 0·1 and 1% (v/v) of N,N-dimethylformamide which corresponds to the amount of solvent used for delivering 10 and 100 μg ml−1 of EMA. The results are shown in Table 1. Significantly, EMA concentrations of 10 and 100 μg ml−1 killed all cells. Based on plate count data the survival rates of cultures exposed to 1 and 0·1 μg ml−1 of EMA were 0·75% and 58% relative to the positive control respectively. In contrast, the viability of the control cultures incubated with N,N-dimethylformamide was not affected by the solvent and quantitative PCR on these samples showed high correlations to the positive control. However, PCR inhibition was observed for the EMA-treated cultures indicating that the agent must have penetrated the membrane of viable cells and covalently cross-linked with the DNA during photolysis. Based on quantitative PCR, the signal log value reductions of the 0·1, 1, 10 and 100 μg ml−1 EMA-treated samples to the positive control were 0·8, 2, 2·3 and 4.

Table 1.  Effect of EMA and N,N-dimethylformamide only treatment on viability of Anoxybacillus flavithermus strain C
 Ethidium monoazide bromide (μg ml−1)Dimethylformamide (%)Positive control
0·11·0101000·11·0
  1. *CFU ml−1.

  2. †Mean of six replicates.

  3. ‡Mean of duplicate experiments.

Plate count*†2·1 × 1052·7 × 103003·9 × 1053·8 × 1053·6 × 105
PCR*‡9·8 × 1046·2 × 1033·0 × 1036·2 × 1016·7 × 1057·4 × 1056·1 × 105

Effect of DNase I treatment on cell viability

The effect of DNase treatment on the viability of A. flavithermus C was investigated by plate count and quantitative PCR. The results are shown in Table 2. The plate counts of the DNase-treated cultures exceeded those of the untreated positive control by 0·13 to 0·7 log units with the recovery rate having its maximum using 200 Kunitz unit of enzyme. This was somewhat in contrast to the results obtained by PCR where the positive control contained approximately two times more cells as was found for DNase-treated cultures.

Table 2.  Effect of DNase I treatment on viability of Anoxybacillus flavithermus strain C
 DNase I (Kunitz units)Positive control
50100150200
  1. *CFU ml−1.

  2. †Mean of six replicates.

  3. ‡Mean of duplicate experiments.

Plate count*†1·5 × 1053·2 × 1052·7 × 1055·8 × 1051·1 × 105
PCR*‡2·0 × 1051·5 × 1052·0 × 1052·3 × 1054·6 × 105

We further investigated the effect of 1x DNase buffer and DNase I to 200 Kunitz units per ml in TSB starch on the growth properties of G. stearothermophilus A, A. flavithermus C and B. licheniformis F. All three cultures showed similar growth in terms of curve progression and OD readings to the positive control of the same culture in TSB starch without additives (results not shown).

Evaluation of differential DNA extraction for total, viable vegetative cells and total spores

The results of the evaluation of the three differential extraction methods are shown in Tables 3 and 4. The correlation coefficients (r) of the dilution samples (high, medium and low count samples) between plate count and quantitative PCR for total vegetative cells were greater than 0·99 for both the reconstituted milk and water suspension. The average PCR signal reduction between total and viable vegetative cells was 0·34 log units for both preparations indicating that a mean of 56% of the total DNA content had been DNase I degraded. Further, the correlation coefficients for viable cells determined by plate count and quantitative PCR were 0·96 for the water and 0·97 for the milk preparation respectively.

Table 3.  Recovery of Anoxybacillus flavithermus C added to deionized sterile water by colony counting and quantitative PCR
 Plate count (CFU ml−1)†16S rDNA PCR (CFU ml−1)‡
ViableDead*SporesTotal cellsViable cellsTotal spores
  1. *Vegetative cell count determined prior to heat-kill.

  2. †Mean of six replicates.

  3. ‡Mean of triplicate experiments.

High3·7 × 1063·7 × 1067·0 × 1067·1 × 106 ± 6·6 × 1041·9 × 106 ± 4·3 × 1053·8 × 106 ± 1·3 × 106
Medium3·7 × 1043·7 × 1047·0 × 1047·5 × 104 ± 1·4 × 1043·0 × 104 ± 2·9 × 1033·2 × 104 ± 3·4 × 103
Low3·7 × 1023·7 × 1027·0 × 1021·1 × 103 ± 9·7 × 1018·6 × 102 ± 1·6 × 1021·5 × 103 ± 8·8 × 101
Table 4.  Recovery of Anoxybacillus flavithermus C added to reconstituted sterile milk by colony counting and quantitative PCR
 Plate count (CFU ml−1)†16S rDNA PCR (CFU ml−1)‡
ViableDead*SporesTotal cellsViable cellsTotal spores
  1. *Vegetative cell count determined prior to heat-kill.

  2. †Mean of six replicates.

  3. ‡Mean of triplicate experiments.

High3·7 × 1063·7 × 1067·0 × 1065·9 × 106 ± 3·2 × 1052·9 × 106 ± 7·8 × 1055·2 × 106 ± 3·0 × 105
Medium3·7 × 1043·7 × 1047·0 × 1048·3 × 104 ± 8·2 × 1035·6 × 104 ± 4·3 × 1032·4 × 104 ± 5·1 × 103
Low3·7 × 1023·7 × 1027·0 × 1021·0 × 103 ± 2·6 × 1027·9 × 102 ± 1·7 × 1016·1 × 102 ± 1·2 × 102

Similarly, the correlations of spore preparations between plate count and PCR was r = 0·97 for the milk and r = 0·96 for the water suspension respectively. PCR performed on the negative controls did not reach the threshold set within 42 cycles.

Enumeration of thermophiles in factory milk powder samples

The results for plate counts, microscopic counts and quantitative real-time PCR of thermophilic bacilli in the factory samples are shown in Table 5. Plate counts for vegetative cells and spores during the first hour of processing were typically in the range between 50 and 890 CFU g−1 for six of the eight factory runs presented. However, the initial plate counts for GO10 and EN23 were higher with between 1·0 × 103 and 3·9 × 103 thermophiles per gram of powder. The number of cultivable thermophilic bacilli increased with processing time commonly by one to two orders of magnitude over the initial count. Based on the number of thermophiles in the last sample taken in each run, GO12 would be regarded as a low count powder with thermophile counts ≤550 g−1. Accordingly, GO10, GO11, GO13, GO14, GO15 and EN22 are medium count powders with thermophile counts below 3 × 104 CFU g−1 and EN23 is a high count powder with numbers exceeding 3 × 104 CFU g−1 during the last 4 h of processing (Ronimus et al. 2003; Rueckert et al. 2004).

Table 5.  Summary of plate counts*, microscopic counts† and 16S rDNA quantitative real-time PCR counts‡ of thermophilic bacilli in milk factory powders
Factory runSampling hourVegetative cells (CFU g−1)Spores (CFU g−1)
Plate countMicroscope countTotal count 16S rDNAViable count 16S rDNAPlate countTotal count 16S rDNA
  1. *Mean of six replicates.

  2. †Mean of eight replicates.

  3. ‡One observation contributed to each data point; ND, not detectable.

EN2210NDNDND0ND
121·5 × 1031·2 × 1051·9 × 1052·9 × 1041·8 × 1031·6 × 104
162·5 × 1044·0 × 1052·6 × 1061·4 × 1051·5 × 1041·6 × 105
171·9 × 1048·0 × 1059·4 × 1059·7 × 1045·9 × 1034·3 × 104
182·5 × 1044·8 × 1051·2 × 1066·7 × 1049·6 × 1032·5 × 104
191·4 × 1041·6 × 1061·8 × 1069·4 × 1041·0 × 1048·9 × 104
EN2313·9 × 1037·0 × 1058·7 × 1055·5 × 1042·0 × 1031·0 × 105
161·8 × 1052·2 × 1061·8 × 1077·1 × 1046·8 × 1043·2 × 105
171·6 × 1053·1 × 1062·4 × 1071·8 × 1051·5 × 1057·1 × 105
182·4 × 1052·8 × 1062·2 × 1071·2 × 1052·2 × 1052·5 × 105
192·7 × 1053·6 × 1062·1 × 1071·6 × 1051·9 × 1053·7 × 105
GO1011·0 × 1031·6 × 1051·1 × 1054·9 × 1031·1 × 1033·7 × 104
124·7 × 1021·2 × 1059·0 × 1043·3 × 1034·0 × 1024·9 × 103
163·3 × 1037·6 × 1055·0 × 1055·4 × 1032·3 × 1032·4 × 104
171·4 × 1048·8 × 1056·9 × 1056·6 × 1039·5 × 1036·5 × 104
181·1 × 1048·5 × 1058·9 × 1051·2 × 1041·9 × 1046·1 × 104
GO1118·9 × 1021·2 × 1056·3 × 1045·4 × 1024·8 × 102ND
122·0 × 1021·2 × 1051·9 × 104ND3·0 × 102ND
165·2 × 1022·4 × 1051·8 × 1051·6 × 1033·8 × 1021·5 × 103
171·0 × 1035·2 × 1054·6 × 1057·6 × 1031·3 × 1034·0 × 104
183·7 × 1031·0 × 1064·1 × 1062·5 × 1043·3 × 1036·4 × 104
GO1215·0 × 101ND1·5 × 104ND1·0 × 1022·3 × 104
122·9 × 102ND5·6 × 104ND2·5 × 1021·8 × 104
161·0 × 1024·0 × 1045·9 × 1041·7 × 1032·0 × 1021·7 × 104
175·5 × 1022·0 × 1058·1 × 1042·2 × 1033·0 × 1021·5 × 104
GO1316·0 × 101ND7·1 × 104ND1·5 × 1021·8 × 104
121·5 × 1022·0 × 1041·0 × 104ND3·0 × 102ND
169·6 × 1021·2 × 1059·6 × 1045·2 × 1024·6 × 1028·1 × 103
175·3 × 1022·4 × 1059·7 × 1044·1 × 1026·0 × 1021·1 × 104
189·7 × 1022·0 × 1051·6 × 1056·7 × 1021·7 × 1039·4 × 103
GO1417·7 × 101ND2·6 × 104ND6·5 × 1013·4 × 103
122·4 × 1024·0 × 1041·5 × 105ND1·6 × 1024·6 × 104
167·0 × 1026·0 × 1054·1 × 1053·2 × 1031·4 × 1032·0 × 104
175·6 × 1037·6 × 1057·7 × 1051·1 × 1044·5 × 1036·5 × 104
189·8 × 1032·0 × 1056·1 × 1051·6 × 1049·2 × 1034·4 × 104
GO1514·5 × 102ND6·9 × 1041·6 × 1027·5 × 1026·9 × 103
124·2 × 1024·0 × 1041·0 × 105ND5·5 × 1021·9 × 104
162·5 × 1031·2 × 1052·5 × 1051·8 × 1039·8 × 1034·4 × 104
174·6 × 1032·4 × 1057·5 × 1053·5 × 1032·8 × 1034·1 × 104
181·6 × 1046·4 × 1059·2 × 1053·7 × 1031·0 × 1049·1 × 104
192·8 × 1046·0 × 1051·5 × 1063·4 × 1032·9 × 1041·1 × 105

The enumeration of total thermophilic bacilli by quantitative PCR (Rueckert et al. 2005) indicated thermophile numbers exceeded those determined by plate counts (Table 5). On average, PCR quantification overestimated the plate counts for vegetative cells and spores by 2·2 and 1·2 log units respectively. As expected, when the contribution of dead cells was removed by the use of the DNase I treatment, the apparent number of vegetative cells more closely correlated with the plate count (reductions of between 0·8 and 2·6 log units were achieved). For milk powder samples with thermophilic loads below 420 CFU g−1 no enumeration was achieved by PCR as any amplification of signal did not reach the threshold set within 42 PCR cycles.

Phase contrast and fluorescence microscopy

The total number of Bacillus rod forms in each milk powder was also determined by phase contrast microscopy, and results are shown in Table 5. Typically, direct microscopic counts exceeded plate counts for vegetative cells between 1·1 and 2·9 log units and differed from quantitative PCR for total vegetative count between −0·8 and 0·9 log units. Although rod morphologies could easily be differentiated by phase contrast, spores were indistinguishable from milk colloids of similar size and refraction. In order to observe spores and vegetative forms a malachite green and safranin red counterstaining was employed on tri-sodium citrate- and n-decane-extracted milk samples (Fig. 1d).

Figure 1.

The live/dead stain of: (a) the TSB control culture of Anoxybacillus flavithermus C, (b) the control culture after tri-sodium citrate and n-decane treatment, (c) an extracted milk powder. (d) Malachite green and safranin red stain of tri-sodium citrate and n-decane extracted reconstituted milk

When milk powder samples were subjected to the live/dead differentiating fluorescence dyes SYTO BC and propidium iodide only dead cells were observed, i.e. all cells stained red (Fig. 1c). Further, when the test was performed on an exponentially grown culture of A. flavithermus strain C all cells were alive and stained green (Fig. 1a). The same culture, subjected to tri-sodium citrate treatment, showed that only a small proportion (<5%) of cells were either killed or their membranes partially permeablized during this procedure. The tri-sodium citrate treatment also increased the number of colonies obtained on subsequent plating, as the agent disrupts cell aggregates formed during culturing (Fig. 1b, data not shown).

Discussion

The rapid enumeration of thermophilic contaminants during milk processing is advantageous to the manufacturer as it enables the optimization of the plant running time to a specified upper limit of contamination. Crucially, before quantitative PCR can be applied on powders, DNA originating from dead cells needs to be removed to avoid overestimation of contaminants. One approach undertaken in this study to discriminate viable and dead cells was the utilization of the nucleic acid binding dye EMA. However, this agent applied under the standardized protocol of Nogva et al. (2003) was not suitable for the purpose of this study, as the dye was cytotoxic towards the most commonly occurring milk powder contaminant (Rueckert et al. 2004). Additionally, quantitative PCR performed on the same batch of samples resulted in a signal reduction of up to 4 log units relative to the untreated positive control indicating that the dye had penetrated the membrane of viable cells and covalently cross-linked their DNA.

DNase from bovine pancrease is unable to penetrate the membrane of viable cells (Fischer 1982; Frankfurt 1983). Not unexpectedly therefore DNase concentrations of up to 200 Kunitz units ml−1, which corresponds to 0·33 mg ml−1 of protein, had no effect on either viability or growth properties of A. flavithermus C, B. licheniformis F and G. stearothermophilus A. Fischer (1982) reported that DNase in concentrations of up to 1 mg ml−1 added to the growth medium was not cytotoxic during mammalian cell culturing and prevented the culture from forming unwanted cell-clumping. Similar findings were observed in this study with A. flavithermus C where cell numbers from cultures exposed to DNase exceeded those of the positive control by up to 0·7 log units which we attribute to the disaggregation of cell clumps by the enzyme. Quantitative PCR on the same batch of samples, however, showed a reverse effect with cell numbers halved for the DNase-treated cultures (Table 2). This discrepancy can possibly be explained by the removal of either DNA from dead cells and/or DNA adhering to the surface of viable cells from the DNase-treated cultures. Disaggregation of cell clumps was also observed following tri-sodium citrate treatment of A. flavithermus C grown in TSB (Fig. 1a,b).

The enumeration of the total and viable vegetative cell and spore content by quantitative PCR in milk powders requires three separate DNA preparations. The most rapid is DNA extraction for the total vegetative cell content, i.e. live and dead cells which requires milk reconstitution, dissolution of milk casein aggregates and DNA release by sonication. This can be completed within 30 min. The extraction of DNA originating from viable cells can be completed in approx. 50 min. This requires an additional DNase treatment of reconstituted milk to degrade DNA of dead cells, followed by milk extraction, DNase inactivation and DNA release by sonication. The extraction of spore DNA requires initial sonication of reconstituted milk to release DNA from vegetative cells, followed by DNase-treatment. The milk sample is then extracted in order to concentrate the spores, any residual DNase is inactivated by heat and the spore DNA released by sonication. DNA samples ready to use for spore enumeration by PCR were typically obtained after 55 min.

In general, DNase I concentrations above 100 Kunitz units per ml were essential for the rapid and rigorous elimination of unwanted DNA from the samples. However, due to the diversity of milk powders with respect to cell number or heat treatment applied to produce the powder, e.g. high, medium or low heat, the proportions of viable and dead cells can vary significantly (Thompson et al. 1978; Murphy et al. 1999). To ensure that DNA from either dead cells or total cells is removed rigorously under all circumstances we regard the use of 200 Kunitz units per ml of DNase as advisable. The enzyme was completely inactivated by both, tri-sodium citrate extraction which chelates divalent magnesium ions and subsequent boiling of the samples for 10 min.

The methods established for the selective and quantitative detection of viable vegetative cells and spores in reconstituted milk were evaluated by adding defined numbers of viable and dead vegetative cells and spores of A. flavithermus C to both sterile milk and sterile deionized water. The different DNA fractions were then extracted and subjected to quantitative PCR. The results from these experiments show that in either suspending medium good correlations were obtained between enumeration based on plate count data and quantitative PCR for low, medium and high count samples.

DNase treatment for the enumeration of viable vegetative cells from the factory samples reduced cell numbers by up to 2·6 log units indicating that the vast majority of cells in milk powder are dead. This fact was confirmed by comparing total counts derived by phase contrast microscopy and plate counting, in which cell numbers determined by the former method exceeded colony counting by up to 2·9 log units. Microscopic counting cannot differentiate between psychrotrophic, mesophilic or thermophilic rod forms, but only the latter would be expected to increase during milk processing, and numbers of psychrophiles and mesophiles in the milk supply were always low or nonexistent (results not shown), so their contribution can be discounted. This supports the contention that the majority of thermophilic bacilli in the powder have grown in the processing line and been killed by the processing conditions. However, their DNA is still largely intact and detectable by PCR. Furthermore, the assessment of the proportions of viable and dead cells in reconstituted milk using fluorescence microscopy was limited by the sensitivity of the method due to the fact that numbers of viable cells were under the threshold of the detection limit of this method. This infers that processing conditions inactivate the vast majority of cells formed in the process line.

The quantification of thermophilic spores by PCR for the milk powders investigated indicates on average the presence of 1·2 log units more spores than determined by plate counting. The excess of spores in milk powders was indirectly supported by the numbers of total cells enumerated by microscopy and PCR as spores evolve from vegetative cells. Surprisingly, all attempts to increase the efficacy of germination by varying the heat activation conditions of 80 and 100°C for 10, 20 or 30 min or inducing germination by the addition of amino acids, carbohydrates (Thrane et al. 2000) or calcium chelates of dipicolinic acid (Donnellan et al. 1964; Lewis 1972) did not increase the germination (data not shown). The highest recovery of thermophilic spores was obtained by heat activation at 80°C for 20 min, which concurs with the findings of McGuiggan et al. (2002). It is possible that no single germination regime will be optimal for all thermophilic strains present in milk powders, and that optimum conditions will vary between species or even strains (McGuiggan et al. 2002). For example, the thermophilic milk powder isolate B. licheniformis strain F requires no heat activation to germinate efficiently and outgrowth occurs within 40 min after incubation in TSB starch at 55°C, whereas A. flavithermus C has an obligate requirement for heat activation for maximal germination and cell recovery (data not shown). The findings made in this study suggest that spore germination is not only influenced by germination conditions. It is possible that conditions during spore formation might also affect germination. Spores of A. flavithermus C produced in Castenholz medium were enumerated equally well by phase contrast microscopy, conventional plate counting and quantitative PCR indicating near complete germination rates (Tables 3 and 4). Thus, the discrepancies noted when these methods were applied to factory milk powder samples must be related directly to the processing conditions when the powder is formed.

In conclusion, the methods presented in this study provide the ability to selectively enumerate the total and viable vegetative cell and total spore content of reconstituted milk, and can be achieved within 90 min of sampling. The total vegetative bacterial load, i.e. live and dead cells can be achieved even more rapidly, within 60 min of sampling, and although this value may not have a direct bearing on the specified limits for viable thermophiles in milk powder, it can be useful as an indicator of overall plant hygiene. If both viable numbers and dead cell numbers are low then plant hygiene is satisfactory, whereas powders with a low viable count but a high dead cell count indicate a low-hygienic operation, which is being masked by the final heat treatment and spray drying. It is envisaged that monitoring of production runs in factories would be performed continuously so that the trends in the counts throughout the entire runs can be assessed and ultimately lead to a reliable accounting of the overall level of thermophilic contaminants.

Ancillary