Comparison of two commercial preparations of buffered peptone water for the recovery and growth of Salmonella bacteria from foods

Transcription

1 Journal of Applied Microbiology 2000, 89, Comparison of two commercial preparations of buffered peptone water for the recovery and growth of Salmonella bacteria from foods C.L. Baylis, S. MacPhee and R.P. Betts Campden & Chorleywood Food Research Association, Chipping Campden, UK 234/3/2000: received 17 March 2000, revised 17 May 2000 and accepted 24 May 2000 C. L. BAYLIS, S. MACPHEE AND R. P. BETTS This study compared the performance of two commercial preparations of buffered peptone water. Performance was assessed in terms of ability to resuscitate and recover low numbers of stressed cells, buffering capacity, growth of Salmonella bacteria in pure culture and growth of Salmonella in food pre-enrichments. Although both the preparations of BPW had similar chemical compositions, differences in their recovery performance were found. Brand A recovered signi cantly higher numbers of heat-injured Salmonella (mean ˆ 057 log 10 cfu ml 1 difference) in pure culture compared with brand B when dealing with very low inoculum levels. Although brand B had higher buffering capacity, the ph at the end of the pre-enrichment was found to be similar in both media, even in foods such as milk powder which showed the greatest decline in ph. Both brands were comparable in their ability to grow unstressed Salmonella from different food types. In unstressed cell studies, similar cell numbers were recovered at the end of a 24 h incubation period from both media, although brand B yielded a higher biomass. In the food study with unstressed cells, performance was related more to the food type and the likely association between this and the level and type of competitor organisms present, rather than to the brand of medium used. INTRODUCTION Recent advances in technology have brought about the introduction of a wide range of commercially available kits and alternative methods for the growth and detection of Salmonella in foods (Feng 1992; Blackburn 1993). Despite the high speci city and rapidity of results that some endpoint detection systems offer, the major limiting factor is the dependence on conventional cultural enrichment to yield suf cient quantities of target analyte, typically a minimum of 4 log 10 cfu ml 1 of the organism, to ensure detection and subsequent isolation (Jay 1992). The isolation and detection of Salmonella from foods can present practical dif culties because these bacteria may be present in low numbers and often in the presence of high numbers of competitor organisms. Moreover, cells may also be sub-lethally injured by food processing or by intrinsic factors associated with a particular food. For these reasons, pre-enrichment, often in a nutritionally complex Correspondence to: C.L. Baylis, Campden & Chorleywood Food Research Association, Chipping Campden, GL55 6LD, UK ( c.baylis@campden.co.uk). medium with few or no selective agents, is necessary to facilitate the recovery of these stressed cells, as well as growth of uninjured salmonellae, prior to selective enrichment. Buffered peptone water (BPW) is one of the most widely used pre-enrichment broths for Salmonella in a wide range of foods. Moreover, it is often the medium of choice in many published reference methods, including the current International Organization for Standardization (ISO) method (Anon. 1998), for the detection of Salmonella in foods. In addition to providing conditions for the resuscitation and growth of cells prior to selective enrichment, BPW buffers the ph of the growth system against ph changes brought about by the growth and metabolism of micro-organisms during enrichment and those imposed by the food sample. Previous studies (D'Aoust 1981) suggest that the nutrient level of pre-enrichment media is not a critical determinant for effective isolation of Salmonella bacteria from foods. Factors of greater importance include the length of time of incubation (Litch eld 1973; D'Aoust and Maishment 1979; D'Aoust 1981) and the presence of large numbers of competing Gram-negative bacteria. These can = 2000 The Society for Applied Microbiology

2 502 C.L. BAYLIS ET AL. retard the growth of Salmonella by virtue of their metabolism, which lowers the redox potential of the enrichment medium (Oblinger and Kraft 1973), or by producing antagonistic substances (Litch eld 1973). In a previous study by Stephens et al. (1997), a method of assessing the ability of pre-enrichment media to recover low levels of stressed cells of Salmonella was devised. Although large differences in performance between commercially available media were highlighted, no attempt to provide explanations was made. In this study, the recovery and growth of Salmonella in two commercial brands of BPW was investigated. To compare the ndings of this study on the recovery of heat-injured cells with those reported previously (Stephens et al. 1997), similar methods and the same strain of Salmonella enterica subsp. enterica serotype Typhimurium (CRA 8378) were used. In addition to the single strain used by Stephens et al. (1997), a second strain of Salm. Typhimurium was also included along with a strain of Salm. Enteritidis. In the rst part, the recovery and growth of heat-injured cultures of these strains in pure culture was studied. In the second part, the growth of 12 Salmonella serotype strains in pure culture was determined and subsequently, the recovery and growth of eight of these strains arti cially inoculated into ve diverse foods types was studied. MATERIALS AND METHODS Bacterial strains Three strains of Salmonella spp. obtained from the Campden & Chorleywood Food Research Association Culture Collection (CRA) were used to compare the recovery and growth, in pure culture, of heat-injured cells. These included two strains of Salm. Typhimurium (CRA 8378 and CRA 8379) and one Salm. Enteritidis PT4 (CRA 8620). Twelve strains of Salmonella, including the three above, were used to compare growth in pure culture in each brand of BPW. The additional strains used included: Salm. Enteritidis PT4 (CRA 1004), Salm. Virchow (CRA 1012), Salm. Hadar (CRA 1017), Salm. Newport (CRA 1043), Salm. Heidelberg (CRA 1027), Salm. Infantis (CRA 1036), Salm. Indiana (NCTC 11304), Salm. Agona (CRA 1052) and Salm. Braenderup (CRA 1096), all previously isolated from foods. Eight of the strains were later used to study the growth of Salmonella in foods. Cultures were maintained on Nutrient Agar (NA; Oxoid) slopes stored at 4 C. Culture media Two commercial preparations of BPW were used in this study. Brand A was obtained from Oxoid (CM509) and brand B from Merck (7228). The chemical composition of both brands, as stated by the manufacturers, was: peptone, 10 g l 1 ; sodium chloride, 50g l 1 ; and potassium dihydrogen phosphate, 15g l 1. In addition to these components, brand A contained disodium hydrogen phosphate (anhydrous) 35g l 1 while brand B contained disodium hydrogen phosphate dodecahydrate, 90g l 1. Both media were made in accordance with the manufacturers' instructions. Two different production batches of each brand of media were compared in this study. Preparation of heat-injured cells Salmonella Typhimurium (CRA 8378 and CRA 8379) and Salm. Enteritidis (CRA 8620) were cultured in 9 ml Nutrient Broth No. 2 (NB; Oxoid) incubated overnight at 37 C. Cultures were inoculated into a 250 ml conical ask containing 100 ml NB pre-warmed to 37 C. Flasks were loosely sealed and incubated with shaking (100 rev min 1 ) in a water-bath at 37 C. The cultures were grown to mid-exponential phase using the procedure outlined by Stephens et al. (1997). After 2 h and at regular intervals thereafter, the cultures were incubated until their optical density (O.D.) at 600 nm was between 02 and 03 and the cultures were in midexponential phase of growth. The initial cell population in each culture prior to heat treatment was determined by diluting a portion (1 ml) of each culture (serial, decimal) in Maximum Recovery Diluent (MRD; Oxoid) and plating appropriate dilutions onto Tryptone Soya Agar (TSA; Oxoid) and TSA with added NaCl (25% w/v). Plates were incubated at 37 C for 48 h. The cells in the remaining cultures were incubated at 515 C for 25 min using a submerged coil apparatus (Cole and Jones 1990). The heat-treated cultures (approximately 5 ml) were collected and portions (1 ml) were diluted (serial decimal) in MRD and plated in the manner described previously. The proportion (%) of injured cells in each culture was calculated from differential counts on TSA, and TSA with added NaCl (25% w/v), after incubation at 37 C for 48 h. Recovery of heat-injured Salmonella A portion (2 ml) of each heat-treated culture was diluted (1 in 10) in BPW (Oxoid). Further dilutions (serial decimal), up to fold, were performed in each of the two BPWs under investigation. Each brand of BPW was tested in duplicate using media from a different production batch for each of the replicates to compare differences in performance. To provide suf cient volumes of inoculated med-

3 COMPARISON OF BPW FOR SALMONELLA 503 ium, at the highest dilution levels, larger volumes (5 ml) in the nal dilution series ( ) were transferred. From each BPW under investigation, the four highest dilutions were each inoculated into duplicate 96-well microtitre plates (Nunc 1±67008 A, Nunc A/S, Denmark). A total of 200 ml was dispensed into each well and plates were covered and incubated at 37 C for 24 h and 48 h. After incubation, the number of wells showing evidence of turbidity, indicating growth of Salmonella, was counted. The inoculum level in each dilution was determined from a most probable number (MPN) calculation on the number of wells scored for growth. To con rm the reproducibility of these results, this procedure was performed on a further three occasions using a single production batch of each brand of media. These same production batches were used for subsequent studies. The results of this microtitre platebased MPN method, as outlined by Stephens et al. (1997), were used to determine the differences in the number of heat-injured cells recovered from the heat-treated cultures by each of the two media.data analysis of most probable number recovery performance Statistical analysis of the data from three Salmonella strains, two Salm. Typhimurium (CRA 8378 and 8379) and one Salm. Enteritidis (CRA 8620), was performed to determine whether there were signi cant differences in recovery performance between the two brands of BPW. Analysis of variance (ANOVA) was carried out on the MPN data obtained from the 10 7 dilution from both BPWs on the ve occasions they were tested. Growth of Salmonella in pure culture The growth of 10 Salmonella strains in each of the two media was studied. Two different production batches of each BPW were selected and prepared in accordance with the manufacturers' instructions. Each strain was cultured by incubation in NB overnight at 37 C, after which they were each diluted (serial decimal) in each of the media under test. Portions (400 ml) of the appropriate dilutions were used to inoculate 100-well honeycomb plates with levels of approximately 2 log 10 cfu 400 ml 1. Inoculum levels were con rmed by plate counts on NA incubated at 37 C for 24 h. Growth of each Salmonella strain was studied by placing the plates into the reading chamber of a Labsystems Bioscreen analyser (Life Sciences International, Basingstoke, UK) at 37 C. Growth was automatically monitored at 15 min intervals over a 24 h period by recording measurements of the O.D. of all the wells at 600 nm. The data generated were used to construct growth curves for each Salmonella strain during incubation in each of the BPWs under test. After 24 h, the level of cells in each well was determined using plate counts on NA incubated at 37 C for 24 h. This experiment was repeated in triplicate to con rm the reproducibility of the work. Comparison of medium buffering capacity The buffering capacity of each brand of BPW was determined. This was achieved by gradually adding 01ml volumes of 1 mol 1 HCl to 100 ml of media during continuous mixing using a magnetic stirrer device. Measurements were taken using a ph meter (Orion SA520, Orion Research Inc., Boston, MA, USA) after the ph reading had stabilized. Further readings were taken after subsequent additions of acid. This experiment was repeated on a separate occasion to con rm the results obtained. Growth and analysis of Salmonella inoculated into food Strains were incubated in NB at 37 C overnight. Each strain was diluted (serial decimal) in MRD, and samples (25 g) of milk powder, cooked turkey, cottage pie readymeal, uncooked minced beef and bean sprouts were inoculated separately with Salmonella. Inoculum levels of approximately <100 cfu 25 g 1 sample were con rmed by plate counts on NA. Samples were enriched in 225 ml of each test BPW incubated at 37 C. At intervals of 2, 4, 6, 8 and 24 h, samples of each enrichment broth were diluted (serial decimal) in MRD and plated onto NA to determine the total plate count, and Xylose Lysine Desoxycholate agar (XLD; Oxoid) to provide a differential Salmonella count. Plates were incubated at 37 C for 48 h and 24 h, respectively. After incubation, all colonies on NA were counted. On XLD, typical black colonies (H 2 S-positive) were counted. A separate count of atypical colonies growing on XLD was also recorded. Counts were expressed as cfu ml 1 of enrichment broth. Uninoculated samples (negative controls) were also tested for the presence of naturally occurring Salmonella or other H 2 S-producing bacteria capable of growth on XLD that would interfere with the Salmonella count. The ph of each enrichment broth was also measured using a ph meter (Orion SA520) and recorded at each time interval. The Salmonella counts from the enriched cultures of each different food type and eight different strains of Salmonella, after 24 h incubation, were subjected to statistical analysis. A three-way ANOVA was used to compare the counts obtained from the two media brands, ve food types and each Salmonella strain used in the food study. The effect of food type was considered separately using a twoway ANOVA.

4 504 C.L. BAYLIS ET AL. RESULTS Recovery of heat-injured Salmonella The mean MPN estimates of the number of cells from heat-injured cultures of Salm. Typhimurium (CRA 8378 and CRA 8379) and Salm. Enteritidis (CRA 8620) from the ve replicates of each brand of BPW, comprising two from different production batches tested on one occasion and three from one of these batches but prepared and used on different occasions, are presented in Table 1. The amount of `heat injury' in each culture, calculated from differential counts on TSA and TSA NaCl, was between 59% and 86% on the rst occasion and 90% and 98% on the second occasion. Although each culture was subjected to the same heat treatment, differences between the strains and the initial number of cells in the inoculum may account for the slight differences in heat injury obtained. Nevertheless, both brands of BPW were inoculated with the same dilutions of each of the heat-treated cultures. The number of positive wells at each of the four highest dilutions therefore highlight the differences in the number of cells recovered by each brand of BPW. Brand A (Oxoid) consistently recovered greater numbers of cells at the highest dilutions (10 7 and 10 8 ) and hence, the lowest numbers of cells compared with brand B (Merck) (Table 1). Calculated MPN estimates of the number of cells recoverable in the 10 7 dilution ranged between 0060 and 2372 cells ml 1 (or 0024±949 cells 400 ml 1 ). It was therefore assumed that the inoculum level in the majority of wells containing the 10 8 dilution was 1 cell. Differences in the numbers of heat-injured cells of Salm. Typhimurium (CRA8378 and CRA8379) and Salm. Enteritidis (CRA8620) in each of the two brands of BPW, after incubation, are presented in Fig. 1. These results represent the mean differences in cell numbers between the two commercial brands of BPW from the ve replicates tested on two separate occasions. The replicates of each brand of BPW comprised four replicates made from one production batch of media and a single broth made from a second production batch. Comparison of the recovery of heat-injured Salmonella from the three individual replicates of each BPW, tested at the same time, showed only slight differences in the numbers recovered (data not presented). Brand A showed greatest variation between individual replicates with differences (approximately 01±04 log 10 cfu ml 1 ) in the number of cells recovered. By comparison, differences between replicates from brand B showed the least variability. Irrespective of the small variations between replicates, the results showed that brand A consistently yielded signi cantly (P R 001) higher numbers of heat- Table 1 Comparison of MPN values of heat-injured cultures of Salmonella Typhimurium and Salm. Enteritidis recovered in two brands of buffered peptone water Mean no. of positive wells at each dilution (192 set up) Strain *Amount of injury Medium brand No. of Mean MPN (cfu ml 1 ) replicates at 10 7 dilution Salm. Typhimurium (CRA 8378) 59% A 2x B 2x % A 3{ B 3{ Salm. Typhimurium (CRA 8379) 86% A 2x B 2x % A 3{ B 3{ Salm. Enteritidis (CRA 8620) 84% A 2x B 2x % A 3{ B 3{ *Determined from differential plate counts on TSA and TSA with added NaCl (25% w/v). {Same production batch prepared on three separate occasions. xtwo different production batches prepared on the same day.

5 COMPARISON OF BPW FOR SALMONELLA 505 Fig. 1 Comparison between mean log MPN values of `heatinjured' Salmonella in two brands of buffered peptone water injured cells of Salmonella, approximately 054 log 10 cfu ml 1 (Fig. 1), compared with brand B. Fig. 2 Growth during pre-enrichment of Salmonella Typhimurium (CRA 8378) in brand A (W) and brand B (&) buffered peptone waters measured using the Bioscreen analyser Growth profiles of Salmonella Growth patterns of 12 Salmonella strains in pure culture in each brand of BPW, measured using the Bioscreen analyser, showed no noticeable differences between the Salmonella serotypes (not shown). However, consistent differences in the pattern of growth, characterized by changes in O.D. over time, were observed in each brand of BPW. On all occasions, brand B yielded the greatest changes in O.D. (Fig. 2). Despite the consistent differences in observed O.D. values, nal populations (log 10 cfu ml 1 )as determined from plate counts revealed little difference in these media (all approximately 10 8 cfu ml 1 ). Differences in O.D. were therefore attributed to biomass as opposed to cell numbers. in the same food type. The growth pro les were, however, affected by the food type and the level of indigenous micro ora present in the samples. For the purpose of this publication, the growth of Salm. Typhimurium (CRA 8378) and Salm Enteritidis (CRA 1004) only are presented (Fig. 4a±e), together with the respective background micro- ora in each food type over the 24 h pre-enrichment period. All eight Salmonella strains inoculated into milk powder showed similar patterns of growth in both brands of BPW Comparison of medium buffering capacity Brand B had a greater buffering capacity than brand A (Fig. 3). On addition of acid, a more rapid decline in the ph of brand A was observed, brand B requiring a higher amount of acid to achieve the same reduction. Growth of Salmonella strains inoculated into a range of foods Irrespective of the strain, growth pro les of the different salmonellas generally followed the same pattern of growth Fig. 3 Difference between the buffering capacity of brand A (~) and brand B (&) buffered peptone water

6 506 C.L. BAYLIS ET AL. Fig. 4 Growth of Salmonella in (a) milk powder, (b) cooked turkey, (c) cottage pie ready-meal, (d) raw minced beef and (e) raw bean sprouts. Growth patterns of Salmonella Typhimurium in brand A (W) and brand B (~), and Salm. Enteritidis in brand A (^) and brand B (&) buffered peptone waters. Corresponding total counts in each medium are represented by open symbols and dotted lines

7 COMPARISON OF BPW FOR SALMONELLA 507 over the 24 h period. Plate counts on XLD revealed increases in the numbers of Salmonella from <070 log 10 cfu ml 1 to approximately 25 log 10 cfu ml 1 after 8 h of incubation. All nal populations of Salmonella exceeded 6 log 10 cfu ml 1 after 24 h, irrespective of the serotype or the brand of medium used (data not presented). The nal populations in brand B were consistently higher that those in brand A after 24 h of incubation. In some instances, the nal population was >1 log 10 higher in brand B (Fig. 4a). The total plate counts on NA of the uninoculated samples at the beginning of incubation revealed numbers of background micro ora to be approximately 2±3 log 10 cfu ml 1, which increased to >8 log 10 cfu ml 1 after 24 h. The observed pattern of growth was similar to that of the Salmonella, albeit at a higher level, throughout the enrichment period (Fig. 4a). The ph of the growth systems remained relatively stable (about ph 68±74) during the rst 8 h. By 24 h, both brands had decreased to about ph 48. The results of growth of Salmonella in cooked turkey revealed similar patterns of growth of each serotype in both brands of BPW. After an initial increase of approximately 3 log 10 cfu ml 1 during the rst 8 h of enrichment, the Salmonella populations after 24 h were typically 8 log 10 cfu ml 1. The greatest difference in numbers was observed with Salm. Enteritidis, which was 05 log 10 cfu ml 1 higher in brand B after 24 h (Fig. 4b). The nal numbers of the remaining species were the same (<1 log 10 cfu ml 1 difference). As with the milk powder, the growth pattern of the total population, including the indigenous micro ora, was similar to the Salmonella, although numbers were approximately 2±3 log 10 cfu ml 1 higher during enrichment (Fig. 4b). The ph of all the growth systems remained relatively stable over the 24 h period, typically decreasing by about 02 ph units. In the prepared ready-meal (cottage pie), the background micro ora was approximately 5 log 10 cfu ml 1 at the beginning of incubation. After remaining at this level for the rst 8 h of enrichment, the total numbers at 24 h were in excess of 8 log 10 cfu ml 1. By comparison, the levels of Salmonella remained at or below 3 log 10 cfu ml 1 during the rst 8 h but had grown to levels of approximately 7 log 10 cfu ml 1 after 24 h (Fig. 4c). Some differences in the Salmonella populations in each brand of BPW were observed and although this was often related to a particular species, it was apparent that brand B consistently yielded higher numbers of Salmonella after 24 h. The largest difference in numbers was found with Salm. Enteritidis (Fig. 4c) and Salm. Virchow (data not shown), which were both 12 log 10 cfu ml 1 higher in brand B after 24 h. All other species were typically 05 log 10 cfu ml 1 higher, while numbers of Salm. Typhimurium were the same in both brands of BPW (Fig. 4c). The ph of these growth systems remained virtually unchanged at ph 69±71 during the rst 8 h of enrichment, but declined to ph 59 in brand B and 53 in brand A. Overall, the number of salmonellas generated was not signi cantly different between the media (P ˆ 008). In the minced beef samples (Fig. 4d), the initial level of total organisms in the samples was 7 log 10 cfu ml 1. Consequently, the total populations within each broth increased by only about 15 log 10 cfu ml 1 over the 24 h pre-enrichment period. Salmonella inoculated at levels of approximately 2 log 10 cfu 25 g 1 displayed varying lengths of lag phase depending on the species and media used. In some instances, interference by competitor organisms made counting of the Salmonella dif cult. The numbers of some Salmonella spp. increased to detectable levels after 4 or 6 h in one or both brands of BPW, while Salm. Newport, Salm. Heidelberg and Salm. Infantis all remained at levels below the limit of detection for the rst 8 h of pre-enrichment in both brands of BPW. However, nal levels of each Salmonella species were generally >6 log 10 cfu ml 1 after 24 h. The observed growth patterns of each Salmonella in the two brands of media were similar, although numbers of some species, for example Salm. Typhimurium (Fig. 4d), attained levels above the limit of detection sooner in one brand than the other, but this was not consistent to a particular brand. The ph of each broth during enrichment remained stable at about 67 for the duration of the incubation period and decreased by no more than 04 ph units in 24 h. The samples of bean sprouts contained high levels of background organisms, typically r6 log 10 cfu ml 1 at the beginning of incubation, which increased to 8 log 10 cfu ml 1 over the 24 h period. Consequently, owing to the high ratio of competitor organisms to Salmonella, which were inoculated at levels of 2 log 10 cfu 25 g 1, certain salmonellas remained below levels of detection for the duration of the enrichment period and were not subsequently recovered. These included Salm. Heidelberg, Salm. Infantis and Salm. Agona, which could not be recovered from either BPW after 24 h (data not presented). Salmonella Newport and Salm. Virchow were not recovered from brand B while brand A yielded 52 and 47 log 10 cfu ml 1 of these species, respectively. In contrast, Salm. Enteritidis could not be recovered from brand A whereas brand B yielded 3 log 10 cfu ml 1 of this serotype (Fig. 4e). On the four occasions when Salmonella was enumerated after 24 h, brand A yielded the highest number of Salmonella from the bean sprout samples. The level of competitor organisms showed a slight increase during this period or remained at a relatively constant level. The ph of each growth system remained stable during the 24 h enrichment period, with ph values remaining within the

8 508 C.L. BAYLIS ET AL. range that would normally facilitate good recovery and growth of the Salmonella (ph 60±80). Analysis of the counts of each Salmonella from the different food types after 24 h incubation did not reveal overall statistical differences between the two brands of media. The in uence of food type on the performance of each brand showed no signi cant differences in the numbers of salmonellas obtained from the two brands of media from each of the food types, with the exception of the cooked turkey (P ˆ 004). Levels of Salmonella generated in both media were, however, still in excess of that needed to guarantee transfer into selective enrichment. DISCUSSION The importance of a suitable pre-enrichment medium for the recovery of heat-injured Salmonella prior to selective enrichment has been demonstrated previously (Clark and Ordal 1969; Edel and Kampelmacher 1973). Despite similarities in their composition, commercial preparations of BPW may still contain certain types of component, particularly peptones, or a concentration of a component that affects the recovery performance of that brand. The ndings of this study highlighted variations in performance between the two commercial preparations of preenrichment media when recovering heat-injured cells of Salmonella. This difference in performance between commercial preparations of the same medium type during recovery of heat-injured cells supports the ndings of Stephens et al. (1997). In the present study, brand A (Oxoid) recovered between 18 and 91 times more Salmonella than brand B (Merck) at the 10 7 dilution, depending on the strain and the degree of injury (Table 1). Differences in pre-enrichment media of up to 3 log 10 cycles between the worst and best medium were reported by Stephens et al. (1997). The product formulations for the two BPWs were apparently similar, assuming that the waters of hydration account for the different levels of phosphate salt. The type or blend of peptones used could be a contributing factor to these differences, although this is dif cult to ascertain without further investigation. The in uence of medium components on the recovery and survival of damaged bacteria has been reviewed previously (Harris 1963). Media preparation, particularly autoclaving and over heating, is an important aspect of culture media that is often overlooked but which may adversely affect medium performance and reliability. During the autoclaving process, auto-oxidation of phosphate buffers and sugars may potentially occur (Baumgartner 1938), resulting in the subsequent generation of toxic oxygen species, particularly hydrogen peroxide, within the medium (Mackey and Derrick 1986). Consequently, poor recovery performance of sub-lethally injured cells in a particular medium, as observed in this study, may result. In this study, the preparation of both media was identical. Growth patterns of 12 Salmonella strains not exposed to heat injury revealed brand B to consistently yield the highest biomass of cells, compared with the other medium, although the differences in cell numbers were generally less than 05 log 10 cfu ml 1. In this study, brand B was shown to have a much greater buffering capacity than brand A. One possible bene t of having a higher buffering capacity in the medium is to counteract extreme changes in ph within the growth system, caused either by the food or the metabolic activity of microbial populations during incubation. Phosphate as a buffering agent has been reported to enhance the growth of certain organisms, including Salmonella, at low ph (Nojoumi et al. 1995). In this study using inoculated food samples, the ph of the growth systems generally remained within the range necessary for the recovery and growth of the Salmonella. The greatest fall in ph was observed with the samples of milk powder. This observation was also reported in a previous study (van Schothorst and Renaud 1985) in which it was observed that competitive growth by high numbers of Gram-positive bacteria in milk powder and dust sweepings from milk powder plants caused a drastic lowering of the ph of the broth. The higher levels obtained in brand B may be attributable to a favourable ph for growth being maintained for longer. Despite brand B showing a higher buffering capacity, both media had a nal ph of about 48 after 24 h. Furthermore, in both media, numbers of Salmonella generated were in excess of those needed to guarantee successful detection. During the investigation into the growth and recovery of Salmonella inoculated into foods, inhibition of Salmonella by competitor organisms was observed. Irrespective of the Salmonella strain present and the type of media used, in foods containing a high ratio of background organisms to Salmonella, the ability to recover Salmonella was greatly reduced. This phenomenon was observed during the growth of Salmonella from bean sprouts and, to a lesser extent, from minced beef. In a previous study by Jameson (1962), the population dynamics of Salmonella enrichments were discussed, in particular the effect of a dominant population on a minority population. Earlier still, it was proposed by Bail (1929) that multiplication of an organism leads to a rise in population to a level de ned as molar concentration. This is when exponential multiplication ceases and the organism enters stationary phase. In mixed cultures, the growth of Salmonella may be terminated prematurely because the competitor organisms have reached the maximum concentration that the BPW can support, resulting in reduced numbers of Salmonella after enrichment. In this study, the high levels of competitor organisms present

9 COMPARISON OF BPW FOR SALMONELLA 509 in the bean sprouts may explain the failure to recover satisfactory numbers of Salmonella from these samples. However, the effects of changes in redox potential in the growth system during incubation, depletion of essential nutrients and the production of toxic products of metabolism, cannot be ruled out as alternative or contributing reasons for the observed inhibition or retarded growth of the Salmonella. The overall ndings of this study have shown that the performance of commercial brands of BPW in the recovery of heat-injured cells of Salmonella can vary. In this study, brand A gave the highest recovery performance for heatinjured cells. It is known that high levels of phosphate may subsequently result in raised levels of toxic oxygen species in the medium after autoclaving. Whether the phosphate component of brand B is the source of its reduced recovery performance, or whether another component such as the peptone is responsible, would require further investigation. In foods, however, and with cells not exposed to sub-lethal injury, neither brand was superior for all food types. Levels of Salmonella in both media after incubation were in excess of those needed to achieve successful transfer to a selective enrichment medium, or even detection by a commercial immunoassay or similar detection methods. In the food study, growth performance was not related to the brand of medium or the Salmonella strain. Instead, medium performance was affected by the food type and most probably by the level and type of competitor organisms associated with certain foods. To provide a more realistic assessment of the recovery of Salmonella and other pathogens from foods during enrichment, further work using cultures previously exposed to an appropriate sub-lethal injury, such as acid, heat or freeze/ thaw stress, should be considered when comparing media or when evaluating a method. This would better mimic the types of stresses experienced by cells in foods during processing, and would give a better understanding of the recovery performance of an enrichment medium. The effect of competitor organisms on the growth of Salmonella and other pathogens in growth systems should not be underestimated. Where appropriate, it may be valuable to con rm the types of competitor organisms in a growth system to understand the mechanisms or factors that enable them to inhibit or retard the growth of the pathogens, if recovery is to be maximized and detection methods improved. ACKNOWLEDGEMENTS The authors would like to thank all the staff involved in this study and the nancial support of Oxoid Limited. The authors also gratefully acknowledge Dr Peter Stephens for providing some of the strains used in this study and for helpful discussions. REFERENCES Anon. (1998) BS EN (ISO 6579) Microbiology of food and animal feeding stuffs ± Horizontal method for the detection of Salmonella. London: British Standards Institution. Bail, O. (1929) Ergebnisse experimenteller populationsforschung. Zeitschrift fuèr Immunitatsforschung 60, 1±22. Baumgartner, J.G. (1938) Heat sterilised reducing sugars and their effects on the thermal resistance of bacteria. Journal of Bacteriology 36, 369±382. Blackburn, C. de W. (1993) Rapid and alternative methods for the detection of salmonellas in foods. Journal of Applied Bacteriology 75, 199±214. Clark, C.W. and Ordal, Z.J. (1969) Thermal injury and recovery of Salmonella typhimurium and its effect on enumeration procedures. Applied Microbiology 18, 332±336. Cole, M.B. and Jones, M.V. (1990) A submerged-coil heating apparatus for investigating thermal inactivation of microorganisms. Letters in Applied Microbiology 12, 232±239. D'Aoust, J.-Y. (1981) Update on pre-enrichment and selective enrichment conditions for detection of Salmonella in foods. Journal of Food Protection 44, 369±374. D'Aoust, J.-Y. and Maishment, C. (1979) Pre-enrichment conditions for the effective recovery of Salmonella in foods and feed ingredients. Journal of Food Protection 42, 153±157. Edel, W. and Kampelmacher, E.H. (1973) Comparative studies on the isolation of sublethally injured salmonellae in nine European laboratories. Bulletin of the World Health Organisation 48, 167±174. Feng, P. (1992) Commercial assay systems for detecting foodborne Salmonella: a review. Journal of Food Protection 55, 927± 934. Harris, N.D. (1963) The in uence of the recovery medium and the incubation temperature on the survival of damaged bacteria. Journal of Applied Bacteriology 26, 387±397. Jameson, J.E. (1962) A discussion of the dynamics of salmonella enrichment. Journal of Hygiene, Cambridge 60, 193±207. Jay, J.M. (1995) Modern Food Microbiology 6th edn. London: Chapman & Hall. Litch eld, J.L. (1973) Salmonella and the food industry ± methods for isolation, identi cation and enumeration. CRC Critical Reviews in Food Technology 3, 415±456. Mackey, B.M. and Derrick, C.M. (1986) Peroxide sensitivity of cold-shocked Salmonella typhimurium and Escherichia coli and its relationship to minimal medium recovery. Journal of Applied Bacteriology 60, 501±511. Nojoumi, S.A., Smith, D.G. and Rowbury, R.J. (1995) Tolerance to acid in ph 50-grown organisms of potentially pathogenic Gram-negative bacteria. Letters in Applied Microbiology 21, 359± 363. Oblinger, J.L. and Kraft, A.A. (1973) Oxidation-reduction potential and growth of Salmonella and Pseudomonas uorescens. Journal of Food Science 38, 1108±1112.

10 510 C.L. BAYLIS ET AL. Stephens, P.J., Joynson, J.A., Davies, K.W., Holbrook, R., Lappin-Scott, H.M. and Humphrey, T.J. (1997) The use of an automated growth analyser to measure recovery times of single heat-injured Salmonella cells. Journal of Applied Microbiology 83, 446±455. van Schothorst, M. and Renaud, A.M. (1985) Malachite green pre-enrichment medium for improved salmonella isolation from heavily contaminated samples. Journal of Applied Bacteriology 59, 223±230.