Two

Stephen Jay, Dianne Davos, Mark Dundas, Elizabeth Frankish and Diane Ughtfoot

pork and liquid egg (251). Growth has been observed on beef, crab meat, and fish when these have been stored at 8°C. Generation times on meat at lOOC are long (8-26 h) compared to those for spoilage organisms. Lower minimum growth temperatures than these have sometimes been reported when salmonellae have been cultured in laboratory media.. Growth has been recorded in broth at 5.9°C (257), and as low as 5.2°C (259) and even 4°C (277) on agar media. In the last study, one of 109 strains tested grew at 4°C, and 44 failed to grow at 7°C. There are difficulties in determining the minimum temperature for growth. Close control and monitoring of the temperature over extended periods are needed. Frequently, near the minimum, there may be a period when there is an increase in cell mass (and hence optical density) without cell division, and growth is not sustained. The QIJtimum tp.mpp.rature for growth is 35-37°('..., with growth of most strains occurnng up to 45-47°C though at a :r;:educedrate. Growth does not occur at 50°C. In Figure 8.1 the ~e root of the aerobic growth rate of Salmonella Typhimurium in mutton mince (352) is plotted against temperature, and is compared with rates predicted from a USDA ARS Microbial Food Safety Research Unit model (52). Agreement between the two sets of data is good, especially considering one was obtained for one strain on meat and the other from different strains growing in laboratory medium. Above the maximum temperature for growth, salmonellae die. Salmonellae are sensitive to heat and heat resista~ strains are uncommon. The m~, decImal reduction tIme at b'/uC of stationary phase cultures of 296 strains of salmonellae heated in trypticase soy broth with 2% yeast extract (pH 6.8) was found to be about 1.3 minutes (285). One strain of S. Blockley had a D-value of 5.8 minutes and S. Senftenberg strain 775W had a D-value of 31 minutes. All other strains had D-values between 0.7 and 2.6 minutes. Similarly, another survey (27) found that most of 221 strains had a D-value at 60°C of 0.4-0.6 minutes, and that only three strains (a strain of S. Bedford, a strain of S. Senftenberg, and S. Senftenberg 775W) were heat-resistant strains (D-values at 60°C of 4-6 min). The z-value (CO needed to reduce the decimal reduction time by 10-fold) of normal heat-sensitive salmonellae is 4-5 Co. Heat resistance depends on what conditions the Sal;'!.onella cells heatinglQells grown at (240), or at high pH increased heat resistance neutral pH. The higher 228

are exposed to before slightly low pH (pH 5.8) \ (pH 8-9.75) (192) hav-eJ over cells grown at near the growth temperature,

the more resistant the cells (193, 285), and even exposure for relatively short periods to sub-lethal temperatures (37°-52°C) increases heat resistance (53, 193). Storage of cells at 4-8°C significantly decreased the heat resistance of S. Enteritidis PT 4 (187) and, therefore, S. Enteritidis in eggs should be more easily killed by cooking following refrigerated storage than following ambient storage. Stationary phase cells are more resistant than log phase cells (194, 285). Heat resistance of salmonellae in foods also .jepe~s on the composition of the food, the pH and the type of aCIdulant, and the ~. For instance, S. Enteritidis is more readily destroyed in egg albumen than in homogenised egg, and survives heat better in egg yolk than in either of the other two (194). Differences in pH account in part for the greater heat resistance in egg yolk than in egg white (282). If homogenised egg is acidified with hydrochloric acid, heat resistance increases to reach a maximum near pH 5.5; if acetic or lactic acids are used, heat resistance decreases with acidification (282). Foods high in fat and low in moisture (e.g. dried egg, dried animal feeds, chocolate, syrups) may need severe heat treatments to kill salmonellae: for instance, in milk chocolate with less than 2% moisture the Dso'c value for S. Typhimurium has been measured at 222 minutes (204). The addition of small amounts of water can have a marked effect on reducing this high heat tolerance: for instance, increasing the moisture content to 3.7% decreased the D71'c value from 20 to four hours for S. Anatum in milk chocolate (33). Cells dried on membranes showed little or no death when heated in an oven at lOOoe for one hour, and one percent or more survived even heating at 135°C for 20 minutes (221). The solute responsible for the lowered awalso

:I:

1.75

.

u; 1.5 c 0 1.25 Cl> c Cl>

0

0

. 0

1

.

Cl

0

'0 0.75 0 0: 0.5

0 Mutton . USDA Model

<11

i

5- 0.25

rJ)

0 5

10

15

25

20 Temperature

30

35

'C

Figure 8.1. Comparison of growth rate for Salmonella on mutton with rates predicted by USDA model -

Salmonella

influences heat resistance. When sodium chloride was used to reduce the aw in broth, there was a small increase in resistance of salmonellae to a maximum at near aw 0.95 (6.1% salt w/w) followed by a decrease in resistance to aw 0.90 (13.8% salt w/w) (27). When glycerol was used to lower the aw, the heat resistance increased with concentration of glycerol to a final aw of 0.85 (43.3% w/w). The increase in D60°Cvalues was only 4-5 fold. When sucrose was used to lower the aw, the heat resistance also increased progressively with concentration to an awof 0.90 (57.6% w/w), but, in

this case, the increase in D60°Cvalues was over a lOO-fold.Furthermore, the z-value was considerably higher in the presence of salt than in the presence of either glycerol or sugar. Corry (68) found a linear relationship between the log D65'c value and the % concentration (w/w) of solute, but not between aw and heat resistance, for sucrose, glucose, fructose, sorbitol and polyethylene glycol. There was not a similar linear relationship for glycerol which behaved differently from the other solutes and which also gave a much lower heat resistance. At the same concentration, the heat resistance of cells suspended in solutions of different solutes increased in the order glycerol, fructose, sorbitol, glucose, sucrose. The length of habituation time of Salmonella at reduced water activity also influences heat tolerance. Mattick et al. (261) found optimal habituation time and the extent of increase in heat tolerance depended on the solute. Habituation in glucose-fructose to aw 0.95 for 12 h resulted in maximal heat tolerance, with more than a fourfold increase in D54 values. Habit,uation under the same conditions for 72 h, however, resulted in heat sensitivity comparable to non-habituated salmonellae. The combined effects of food composition, acid and solute type, pH and aw mean that it is not possible to predict the heat resistance of salmonellae in foods of low

aw. At chill temperatures below that permitting growth, salmonellae can survive on foods for long periods. They have been shown to survive for more than 28 days at 2-4°C on a variety of vegetables such as green beans, cabbage, lettuce, beets, carrots, peppers, and tomatoes (207). On inoculated pecan halves, little decrease in the viable population of three strains of salmonellae over 32 weeks storage at 5°C was found by Beuchat and Heaton (38). Survival is dependent predominantly on other factors like pH and aw, and survival is longer at chilled than at ambient temperatures: for instance, S. Enteritidis PT 4 inoculated into commercial mayonnaise is destroyed more rapidly when the mayonnaise is held at 20°C than when held at 4°C (249).

Freezing will reduce the numbers of salmonellae by an amount that depends on the freezing rate and the type of food. Rapid freezing promotes survival. Log phase cells are more sensitive than stationary phase cells. There is an initial decrease in viable count as a result of the freezing damage, followed by a slower rate of decline during storage. Lower storage temperatures and less fluctuations in temperature give greater survival. Storage temperatures near the freezing point result in most death or injury. In minced chicken breast (pH 5.8), 60-83% of Salmonella cells survived storage at -20°C for 126 days, whereas at -2° and -5°C only 1.3% to 5.8% were still viable after 5 days (140). Salmonellae can survive for long periods in frozen foods and their frequent isolation from frozen foods is ample evidence of their ability to survive. pH In laboratory media, salmonellae can grow from about pH 4.0 to near pH 9.J5!with the optimum bemg in the range ~\The minimum pH allowmg growth is influenced by the temperature of incubation (66), and by the presence of salt and nitrite (151). The minimum pH also depends on the acidulant used. In broth acidified with hydrochloric, gluconic, lactic, or acetic acids, the minimum pH for growth has been reported to be 4.05, 4.20, 4.40, and 5.40 respectively (66). Outside the range of pH allowing growth, salmonellae die. At low pH values the nature of the acidulant determines the rate of death. Volatile fatty acids are more bactericidal than acids such as lactic and citric acids. The undissociated molecule is responsible for the lethal action so that the effectiveness of acids increases as the pH is lowered (157). For formic, acetic, propionic and butyric acids, the bactericidal effect decreases somewhat with increasing chain length. The rate of death decreases as the temperature is reduced, and, at a given pH and temperature, is dependent on the concentration of the acid. Organic acids are more effectively bactericidal under anaerobic than under aerobic conditions. Tolerance of acidic conditions, apart from being an advantage for survival in the environment, is important for virulence because ingested salmonellae have to pass through the acid (pH less than 3) of the stomach. Tolerance of acid conditions is influenced by how the organisms are grown. Salmonellae have at least three systems that enable cells to adapt to survive potentially lethal acid exposure (234). One is a pH independent general stress resistance produced by stationary phase cells that is dependent on RpoS and the production of the stationary-phase229

Salmonella

Combinations As is clear already, when salmonellae are in foods, several factors act simultaneously to influence the growth rate or the extent of survival. It is the combination of parameters like pH, aw and temperature, and the presence of inhibitors like nitrite or short chain fatty acids that are important in determining the response of salmonellae. A knowledge of the minimum and maximum values for temperature, pH and aw that allow growth under otherwise ideal conditions is useful. Similarly, a knowledge of the combinations of these factors that will prevent growth is important in judging whether salmonellae may grow in a food. More useful information is obtained by taking advantage of 'user-friendly' models that predict the growth of salmonellae in foods once the temperature, pH, and aware known. One is a USDA ARS Microbial Food Safety Research Unit Model (52) and a second is Food MicroModelTM from the UK which was developed from Ministry of Agriculture Fisheries and Food funded research. The agreement between predicted growth rates and rates observed in many foods has been good. One example is shown in Figure 8.1. In many situations the major controlling factors are temperature, pH and aw. Nitrite, high carbon dioxide concentrations, and some natural antimicrobial materials in foods may be important in some circumstances. In these cases the prediction will be 'fail-safe' in that it will predict faster growth than in fact occurs. The models can also be used to predict the likely extent of growth under changing conditions of

6 5 4 ::I

LL 0 3 Cl 0 .J

2

0 0

5

10

15

20

25

Days

Figure 8.2. Effect of pH on the survival of SalmonellaTyphimurium in pepperoni manufacture; fermentation at 35°C, maturation at 12°C,final Sw0.878 (350)

..

temperature such as occur during cooling and during storage and distribution. Models for heat destruction depend on knowing the D- and z-value for the food. Models that can be used to predict survival of salmonellae in foods in other non-growth conditions are being developed. A preliminary one that is available examines the response of salmonellae to the effects of temperature, percent salt (aw), and pH with lactic acid as acidulant. The kinetics of survival are not linear or simple, and as a result this model predicts the time for a 10 OOO-fold reduction in numbers, or a 4-log decrease. Figure 8.3 has been derived from predictions given by this model (USDA ARS Microbial Food Safety Research Unit Model) and shows how combinations of temperature (5-30°C), aw (0.88 and 0.90) and pH (4.8-5.2; lactic acid) influence the time taken for viable salmonellae to decrease by 4-log units.

Methodology Methods for the detection of salmonellae in foods have been intensely researched for many decades, and this has led to a proliferation of analy.tical procedures for their isolation and identification. Most of the procedures are lengthy to perform, and none are optimal for all food categories or for all known serovars. In the following discussion both conventional cultural techniques and rapid methods for the isolation or detection of salmonellae in foods are reviewed. Despite the increased use of rapid methods, it should be recognised that developments in cultural methods are still of considerable importance, as they rely on cultural procedures to provide enough cellular material for detection. Cultural methods The conventional examination of food products for salmonellae requires the use of cultural methods which are different from those used for clinical specimens. This is mainly because salmonellae in foods are subjected to ecological conditions which are unlike those of their preferred environment, the gastrointestinal tract of man and animals. During processing and storage of foods, for example, they may be exposed to heat, desiccation, preservatives, osmotic stress or changes in pH. Furthermore, they usually have to exist in association with a relatively large number of competing microorganisms which represent the natural flora of the food. Therefore methods for their isolation must be designed to enhance the survival and multiplication of salmonellae while 231

Stephen Jay, Dianne Davos, Mark Dundas, Elizabeth Frankish and Diane Lightfoot

suppressing the growth of competitors. Because expert opinion is divided on many aspects of Salmonella methodology, laboratories usually choose techniques and procedures which have been found to be best suited for their purpose. Comprehensive reviews on the diverse, and often contradictory reports on Salmonella methodology have been presented by Litchfield (248), Fagerberg and Averts (123), Flowers et al. (134) and Fricker (141). There will always be disagreement between microbiologists on the best choice of media and incubation temperatures, but there has been general approval of the following basic procedure: pre-enrichment in non-selective broth, selective enrichment in broth, isolation on differential selective agar, biochemical confirmation of suspected colonies, and serological identification. Isolation of Salmonella Typhi With the long incubation period of typhoid, it is not always possible to isolate Salmonella Typhi from the implicated food. In an outbreak situation and as the ultimate. source of infection is the human carrier, efforts must be concentrated on screening all possible food handlers by culture of both faeces and urine as well as by serological methods for antibody detection, especially Vi antibodies. It may be necessary to examine more than one sample of faeces to detect the intermittent excretor. Sewer swabs (273) can be used to trace contamination up a sewer or watercourse to its origin to detect the source of infection of the water supply. This method has been used successfully in many outbreaks. Swabs are also useful in theinvestigation of outbreaks at institutions. Development of a DNA probe could prove useful in rapid diagnostic assays for S. Typhi in mixed bacterial samples (338). For the isolation of S. Typhi from food or faeces, similar methods and media to those used for other Salmonella serovars are employed. However, S. Typhi will not grow at elevated temperatures (e.g. 43°C) or at the concentrations of brilliant green which are suitable for the growth of other salmonellae. Selenite F and selenite cysteine broth are not very selective, while Muller-Kauffman tetrathionate broth, Rappaport (RV) and Rappaport (RVS) broth are inhibitory to S. Typhi (310). Iveson and MackayScollay (208) and Chau and Forrest (63) found that strontium selenite was superior to selenite F for the isolation of S. Typhi. Tetrathionate broth will grow a wide range of serovars including S. Typhi. S. Typhi grows well on desoxycholate citrate agar (DCA) and on bismuth sulphide agar (BSA), but xylose lysine desoxycholate agar (XLD) is not very selective. Mannitol lysine crystal violet brilliant green agar is not suitable for S. Typhi.

\ I

t

~32

Standard reference methods The development of standardised reference methods for the detection of salmonellae in foods became necessary with the adoption of microbiological specifications for this microorganism in both state and national food law. Standard reference methods have also proved useful for arbitration purposes in the international trade of foods, for assisting many laboratories engaged in routine work to obtain more uniform results, and for interlaboratory comparison testing programs. I!js important to realise that standard reference methods are usually based on the best practical procedures available and accordingly may not be the most sensitive. Hence it is likely that some laboratories experienced in Salmonella analysis will obtain equivalent or better results using alterjj]ftive techniques. For non-routine or research purposes, such techniques may be more acceptable to some laboratories. In Australia, a committee of the Standards Association of Au.s.tJ:alia (SAA) has promulgated a iillIDdard rpfprpnpp method (AS 17~h ') ~ ~~ the isolatioILand identification of salmo.Dp]]aein.. f90ds (359). A flow diagram of this method is presented in Figure 8.4. The analytical techniques described in this method conform to the five procedural steps listed previously. The discussion to follow is centred on this published method. ,,Sample preparation It is generally al:rppd th3t for adeqllatB rBcovery of salmonellae from cont2minatBd food mMerials, sample amounts of at least 25 g must be tested. or some high risk food categories such as drIed milk powder, sample sizes of hetween 50 to 100 g qre frequently rppommended (19, 20, 134, 203). While the probability of detecting salmonellae in a contaminated product increases with sample size,

80 Q)

'" os E 60

-- pH 4.8 a. 0.88 -0- pH 4.8 a. 0.88 -.- pH 5.2 a. 0.90

c.>

~

tn

.3 40 ~ ... 0 ';;;

>os C

20 0 0

10

30

20 Temperature

.C

Figure 8.3. Interaction of temperature, awand pH on death of salmonellae

T i

!

"

.

Stephen Jay, Dianne Davos, Mark Dundas, Elizabeth Frankish and Diane Lightfoot

I' 11

RESUSCITATION

g

25 g sample + 225 mL Buffered peptone water 37"C/16-20h

~

0.1 mL

1 mL

+

+

SELECTIVE

RV broth

ENRICHMENT BROTHS

8-

-

8

42°C/18-24h

Mannitol selenlte cystine broth 31"C/18-24

!/

PI.!. o~

h

~!

SELECTIVE SOLID MEDIA

XLDagar

IL-..J

I

IL-..JI 31"C/24-48

bismuth sulphite agar

h

~

3 typical colonies/plate

~

~ peptone water 37°C until turbid ~ BIOCHEMICAL TESTS

n Q nutrient agar

lysine

g

t:j decarboxylase broth

IL-..JI ONPG broth

37°C/18-24 h

~~

[]

SEROLOGICAL TESTS

Figure 8.4. Flow diagram of method for detecting Salmonella -234

CLEDmedium

Salmonella

enrichment is given in the following sections. At this point, however, it is useful to consider specifically sample preparation in relation to enrichment procedures for detecting salmonellae in foods. The recommended ratio of inoculum to enrichment medium varies from 1:4 to 1:10, often according to the type of food and the size of the sample to be tested (19, 134, 171, 203, 359). For most purposes, a ratio of 1:10 is preferred. The higher dilution is particularly important where food is added directly to a selective enrichment broth. Silliker and Taylor (345) demonstrated that the reduction of selectivity of enrichment broths was directly proportional to the amount of water soluble components in a number of different foods including gelatin, albumen, egg yolk and dried beef. Since the addition of food samples may adversely effect the performance of selective media due to alteration in pH, salinity, nutrient composition, or other physical factors, it is important to evaluate the influence of each specific product on enrichment procedures. This will be discussed further in later sections. Mter the sample has been added to the appropriate volume of enrichment medium, it is important to ensure that salmonellae will be released from the food into the diluent without loss of viability. If the sample is powdered, ground or comminuted, it may be readily dispersed by gentle swirling or stirring with a sterile glass rod. Other products will require mechanical blending to obtain a homogeneous suspension. The risk of destroying salmonellae by localised overheating during this process must be minimised. Most food samples can be adequately homogenised using Stomacher machines which generate very little heat during mixing. Some food materials exhibit a remarkable resistance to dispersion when added to enrichment media. Casein and products containing casein, for example, form large sticky insoluble lumps which could prevent the recovery of salmonellae trapped within the non-dispersed casein. To obtain optimum dispersion of this material, the SAA (357) described a procedure for mixing the sample in chilled phosphate solution (acid casein) or chilled buffered peptone water (other caseins and casein products) before the addition of the pre-enrichment medium. In the examination of gelatin, the US FDA recommends the inclusion of a gelatinase solution in the pre-enrichment medium to prevent solidification during incubation (19). Dilution of gelatine 1:20 in pre-enrichment medium is also successful in this regard. Similar dispersion problems can occur with fatty foods. Addition of the emulsifying agents Tergitol No. 7 and Tween 80 to tetrathionate brilliant green broth has been

shown to enhance Salmonella recovery from pork sausages (144). The US FDA (19) recommends the addition of 2.25 mL Tergitol 7 or Triton X100 to 225 mL pre-enrichment medium used for recovery of salmonellae from coconut, meats, meat substitutes, meat by-products, animal substances, glandular products and fish, meat and bone meals. When used, a minimum amount of surfactant should be added. This is determined by titration and is simply achieved by adding just enough to initiate foaming. It is necessary to ensure that the combined effect of surfactant concentration, enrichment composition and incubation temperature is not too inhibitory for the recovery of injured salmonellae. Morris and Dunn (276), using Tergitol No. 7 at 6.0% in tetrathionate brilliant green broth incubated at both 37°C and 43°C, demonstrated that this enrichment combination with pork sausage samples may result in inhibition of salmonellae when incubated at the higher temperature. It is also important for any laboratory using a surfactant, to test adequately each batch of surfactant to ensure that no toxicity towards salmonellae is apparent at the concentrations used. Different manufactured batches of surfactant can vary significantly in regard to toxicity, irrespective as to whether they are recommended for use in isolation of salmonellae. D'Aoust et al. (94) studied the effects of nine surfactants in pre-enrichment media. Tween 20, Teepo1610 and Brij 35 were eliminated because of their toxicity to Salmonella. Tergitol 7, Tween 80, Triton X100, Myrj 525 and Arlacel 80 and Tween 60 did not inhibit the growth of salmonellae but also did not increase the isolation rate from 45 fatty foods studied when compared to nutrient broth controls. It is therefore not clear whether surfactants do enhance recovery of salmonellae to any significant degree. Non-selective pre-enrichment Various treatments related to food processing such as heating, drying, freezing, irradiation, changes in pH, or addition of preservatives may induce sublethal damage in bacterial cells (55). Since the presence of inhibitors or other selective agents can reduce markedly the ability of media to support the repair and growth of injured cells, most methods for isolating salmonellae from foods involve a pre-enrichment step using a nonselective medium. This phase of the isolation procedure, according to Flowers et al. (134), should provide: a. nutrients for multiplication to favour the ratio of Salmonella to non-Salmonella b.

. . mIcroorgamsms; repair of cell damage;

235

Stephen Jay, Dianne Davos, Mark Dundas, Elizabeth Frankish and Diane Ughtfoot

c. d.

rehydration; dilution of toxic or inhibitory substances. Many different pre-enrichment media have been recommended for various food materials. One of the most widely used media is lactose broth, originally developed by North (288) to improve the recovery of salmonellae from dried egg products. It was claimed that in the presence of a mixed flora, the fermentation of lactose resulted in a lower pH which enabled salmonellae to survive and grow while other types of microorganisms were inhibited. Silliker et al. (346), however, considered that where there was an unfavourable coliform to Salmonella ratio in a food sample, this broth would favour the growth of lactose fermenting competitors. Taylor et al. (377) and Gerichter and Sechter (150) on the other hand, found that the carbohydrate used in preenrichment media did not influence their effectiveness in the ultimate number of Salmonella isolations. These findings suggest that other resuscitation media such as nutrient broth and buffered peptone water are also suitable for the pre-enrichment of most foods. In addition, it is possible that highly nutritious foods such as dried eggs and milk may be simply reconstituted in either distilled water or physiological saline. Pre-enrichment in buffered peptone water has increased in popularity because it offers several important advantages. The inclusion of peptone aids resuscitation and the presence of phosphate buffer prevents excessive acidification. Baylis et al. (34) found that the recovery of heat injured cells of Salmonella can vary depending on the formulation of buffered peptone water. Moreover, phosphates have been shown to assist in the rapid repair of damaged cells (322, 323). Van Leusden et al. (390) recommended buffered peptone water for routine use as does Fricker (141) and this is the medium specified in the Australian Standard reference method AS 1766.2.5-1991 (359). Bailey and Cox (25) described a medium called universal pre-enrichment (UP) broth for the simultaneous recovery of Salmonella and Listeria. This medium was highly buffered and low in carbohydrate. Good recovery of low numbers (10 cfu) of heat-injured Salmonella and Listeria was demonstrated in both pure and mixed culture and several different foods. Lactose broth is widely used in the USA and has been adopted as the official pre-enrichment medium for most foods by the AOAC and US FDA (19,20). The SAA specifies buffered peptone water for all foods except cocoa and cocoa products. The ratio of food to medium is usually recommended as 1:10. However, although lactose broth and buffered peptone water are often interchangeable

and the inoculation ratio of 1:10 is almost universal, these are not suitable for all food types. Table 8.12 summarises the recommendations of the US FDA and SAA. There is general agreement that the incubation temperature for pre-enrichment should be around the optimum for growth of salmonellae, i.e. 35°C to 37°C. The SAA recommends 37°C:!:: 1°C. The time of incubation has been well studied and recommendations have varied from 6-48 h. More recently, the time recommended has shortened to 16-24 h. AS 1766.2.5-1991 states 16-20 h while the US FDA states 24:!::2 h (19). D'Aoust and Maishment (93) found that short incubation times of 6 h were associated with reduced recovery of salmonellae. On the other hand, prolonged incubation can also lead to reduced recovery due to overgrowth. Fricker (141) reviews the issue at length and concludes that incubation of pre-enrichment cultures for 24 h before subculture probably provides the best results. Components of some food samples may impair resuscitation and adversely affect the recovery of salmonellae. Cocoa powder, for example, is known to contain naturally occurring substances which are bactericidal to some strains of salmonellae (56). Zapatka et al. (421) showed that this bactericidal activity could be diminished by the addition of casein to the pre-enrichment media (reconstituted non-fat dry milk or nutrient broth containing 5% (w/v) casein). In the examination of chocolate and products containing chocolate, the addition of skim milk to the resuscitation medium has been recommended (19, 134, 359). D'Aoust and Sewell (92) concluded that reconstituted skim milk powder with 0.002% w/v brilliant green was marginally more successful in detecting salmonellae compared with a method of the International Office of Cocoa and Chocolate and International Sugar Confectionery Manufacturers' Association which used mannitol broth for pre-enrichment. The recovery of salmonellae from other foods containing natural inhibitors or artificial preservatives could also be improved by the inclusion of non-toxic neutralisers in pre-enrichment media. The US FDA (19) for example, recommend the use of 0.5% K2S0a in the examination of dried onion and garlic, and dilution beyond their toxic levels of allspice, cinnamon and oregano. There is even now, limited information in the literature relating to neutralisers for use in resuscitation procedures for different foods. Repair of cell damage or resuscitation is an important function of the pre-enrichment phase and has been extensively reviewed by Andrews (7). Numerous studies on reconstitution of dried

11

"

.1

Salmonella I~

foodsin pre-enrichment media have resulted in acceptance of slow rehydration methods, which have been shown to increase Salmonella recovery from some foods. Presumably, slow rehydration reduces osmotic injury which occurs by rapid rehydration from a near dry state. Ray et al. (324) found that for dried milk products, reconstitution for 1 h in a high osmotic environment (solid:liquid ratio 1:2.5) followed by normal pre-enrichment (solid:liquid ratio 1:10) gave higher recoveries of salmonellae. This was later confirmed by van Schothorst et al. (394) who found reconstitution of dried milk powder in buffered peptone water (1:2) for 30 minutes at ambient temperature followed by dilution to a final ratio of 1:9 substantially increased recovery of salmonellae. In a variation of the reconstitution procedure, similar results were obtained for dried instant non-fat milk powder using a slow rehydration soak method (9). In this method the sample was poured onto the surface of the pre-enrichment medium and allowed to dissolve slowly at ambient temperature. This method has also been demonstrated to enhance recovery of Salmonella from dry whole milk, lactic casein, non-instant non-fat milk and rennet casein, but not from sodium casinate (308). In a comparison between rapid hydration and the soak method, Wilson et al. (416) found that the soak method gave higher recoveries from soya flour but not from brewers' yeast, dried active yeast or onion powder. A refrigerated pre-enrichment procedure for dry foods was described by D'Aoust et al. (87) in 1993. Pre-enrichment cultures were incubated at 4°C for 72 h prior to selective enrichment. Pre-

enrichment media were those appropriate to the sample type as specified in the BAM 1984 6th edition, Chapter 7. This procedure did not claim greater recovery of Salmonella from dry foods but rather was designed to validate a method of refrigeration of pre-enrichment cultures over a weekend. There has been considerable conjecture regarding the optimum equilibration time for rehydration. According to van Schothorst et al. (394) the time of sample equilibration at the initial 1:2 sample/medium ratio is of little consequence. A further investigation by D'Aoust and Sewell (89) showed no advantage between long (4 h) and short (15 min) equilibration at ambient temperature. This study was carried out with high sample to medium ratio m\xtures of feeds and feed ingredients followed by dilution to normal enrichment culture levels. While it appears that the control of aw during rehydration of dry materials may be important for optimal recovery of damaged salmonellae, slow rehydration methods have only been demonstrated to be significant for certain dairy products and soya flour. The significance of aerobic or anaerobic preenrichment is not clear. Alford and Knight (1) described a shortened pre-enrichment using continuous shaking for 4 h at 37°C which was followed by addition of selenite and further shaking for 20 h. Improved recovery rates obtained with this procedure were attributed to actively multiplying cells being less sensitive to exposure to toxic substances such as selenite in subsequent pre-enrichment media. It was conceded, however, that the value of aerated pre-

Table 8.12. Pre-enrichment media for the isolation of salmonellae Medium (volume, mL) Lactose broth (225)

from foods Reference

Food (mass, g) Eggs and egg products (25); prepared powdered mixes, e.g. cake

AOAC(19)

cookie, doughnut,etc., cheese, dough and prepared salads (25) infant formula (25), fruits and nutmeats (25); crusteans and fish (25); food dyes, pH >6.0 (25); frog legs, food snails and shellfish (25) Lactose broth (225) + Tergitol No 7 (2.2) or Triton X100 (2.2)

Coconut (25); heat processed and dried products (25), e.g. meats, animal substances and meals

AOAC (19)

Lactose broth (225) + 5% aqueous gelatinase (5)

Gelatin (25)

AOAC (19)

Sterile distilled water (250) + 1% aqueous brilliant green (0.45)

Non-fat dry milk (25); dry whole milk (25)

AOAC (19)

Trypticase soy broth (225)

Dried yeast (25); spices (25) except for allspice, which are diluted 1 :100 and cloves 1: 1000

Trypticase

Onion flakes, onion powder, garlic flakes and garlic powder (25)

AOAC (19) AOAC (19)

soy broth with 1.25 g K2SO,

cinnamon

and oregano

AOAC (19)

(225) Nutrient broth (225)

Sugar frosting and topping mixes (25)

Sterile reconstituted non-fat dried milk (225)

Cocoa

Buffered peptone water (225)

All foods unless otherwise specified

and cocoa

products

including chocolate

AOAC(19) and SAA (359)

(25)

SAA (359)

231 ;;;;;;..

~

I

11

ill

i!

:1

I

Stephen Jay, Dianne Davos, Mark Dundas, Elizabeth Frankish and Diane Lightfoot

enrichment may be limited by the competition with other microorganisms, the composition of the food, and the availability of shaker-incubation space. In a more recent study, pre-enrichment under anaerobic rather than aerobic conditions, Kafel (217) provided higher Salmonella isolation rates from fish. It has been suggested that the atmosphere of pre-enrichment may play an important role in the repair of damaged cells. Van Schothorst et al. (395) demonstrated that pre-enrichment of foods may be particularly important where selective enrichment involves the use of tetrathionate brilliant green broth incubated at the elevated temperature of 43°C. Their investigation was undertaken because of the conflicting evidence in the literature on the usefulness of the direct enrichment of foods in this medium at 43°C. From their experimental data, it was postulated that pre-enrichment provides a large population of salmonellae where there are most probably cells present which are less sensitive to tetrathionate, and these will multiply especially when an equally large competitive flora is able to 'detoxify' the medium. The compositing or pooling of samples prior to pre-enrichment has been discussed previously as a practical and economical approach to Salmonella testing. Wet compositing of preenrichment cultures is an alternative system which has been frequently employed by the food industry in the USA (342). This method involves the transfer of 1 mL aliquots of a number of individually pre-enriched sub-samples into an appropriate volume (1:10 v/v) of a single selective enrichment broth. Price et al. (311) and Silliker and Gabis (342) showed that this method, as compared to individual sample analysis, could be used for a variety of food products without loss of sensitivity. Furthermore, Price et al. (311) demonstrated that as many as 25 pre-enrichment broth cultures could be pooled, and that the enrichment ratio may be varied from 1/10 to 1/50 and 1/100, with no apparent reduction of efficiency in the recovery of salmonellae. The relative advantages of wet compositing include: a. an increase in the capacity for Salmonella testing of foods; b. retention of pre-enrichment broth cultures for subsequent individual analysis if positive tests are found; and c. the avoidance ofthe hazards and inconvenience of handling large flasks of culture. Selective enrichment Selective enrichment media are employed in the examination of foods to encourage the multiplication of salmonellae whilst reducing or inhibiting the growth of competitive 238

microorganisms such as coliforms, Proteus and Pseudomonas. Raw and unprocessed foods are sometimes inoculated directly into selective enrichment media because salmonellae in these products are unlikely to be sub-lethally injured. Moreover, pre-enrichment could be detrimental with fresh foods where overgrowth by other microorganisms can occur. According to Litchfield (248), this is particularly the case with lactose broth which is a good growth medium for the Enterobacteriaceae. Nevertheless, the ICMSF has recommended that all food samples be preenriched in a non-selective medium prior to selective enrichment (203). Their recommendation was based on the findings of Edel and Kampelmacher (112) and Gabis and Silliker (143) that pre-enrichment gave higher isolation rates than direct selective enrichment in the a~alysis of frozen meat, poultry and liquid eggs. The SAA has also specified a pre-enrichment step in its method for the detection of salmonellae in foods (359). There appears to be insufficient comparative data in the literature to justifiably exclude direct selective enrichment for all raw and non-frozen foods. In the examination of fresh or chilled poultry by the whole bird rinse technique, preenrichment is unnecessary and the rinse fluid may be transferred directly to the selective enrichment broths. Cox et al. (72), for example, found that direct selective enrichment of the total volume of rinse fluid was effective in detecting low levels of salmonellae on non-frozen broiler carcases. They used distilled water as the rinse fluid and added a concentrated solution of selective enrichment broth to give the resultant solution a single strength concentration. By comparison, Thomason and Dodd (378) examined 208 naturally contaminated samples of raw meat and poultry and concluded that while direct selective enrichment gave slightly higher recoveries than did pre-enrichment followed by selective enrichment, both procedures should be used to obtain maximal recoveries of salmonellae. There are numerous modifications of many selective enrichment media reported in the literature. Ideally, these media should support the multiplication of salmonellae to detectable levels, be sufficiently selective to prevent over growth by competitors, maintain selectivity after addition of the sample and allow recovery of all serovars. The different types of selective enrichment media which have been used for the isolation of salmonellae from foods are presented in Table 8.13. These media are based on inhibitors such as tetrathionate, selenite, magnesium chloride, strontium chloride, the dyes brilliant green and malachite green, and antibiotics including

.

Stephen Jay, Dianne Davos, Mark Dundas, Elizabeth Frankish and Diane Lightfoot

shown by North and Bartram (287) to greatly improve the growth of salmonellae in the presence of large amounts of organic material. They also demonstrated that the quantity and nutritional quality of peptones and phosphates used in selenite media affected the recovery efficiency of the broth. Occupational health and safety issues for laboratory workers are beginning to impact on the use of selenite in media. Although the risk is small and the medium for Salmonella isolation quite productive, it is possible that the use of selenite containing media will be restricted or cease altogether. Other recommended modifications to selenite media include the addition of brilliant green and/or sulphapyridine to prevent the growth of strains of Proteus, Escherichia, Enterobacter and Pseudomonas which are not inhibited by high concentrations of selenite. Stokes and Osborne (362) showed that the incorporation of 0.0005% brilliant green in a mannitol taurocholate selenite medium prevented the development of Escherichia and Proteus while supporting the luxuriant growth of salmonellae from very small inocula (one cell per mL of medium). In a subsequent investigation, Osborne and Stokes (299) found that egg products considerably reduced the selective properties of the selenite brilliant green medium. They also demonstrated that the neutralising effect of egg products could be eliminated by the addition to the medium of sulphapyridine (0.05%). The latter modification was considered suitable for the isolation of salmonellae from a wide range of samples. In the choice of enrichment media, it should be noted that broths containing a number of different selective agents may lead to the inhibition of strains of salmonellae sensitive to their combined activities. Iveson and MacKay-Scollay (208,209) showed, with their studies in Australia, that strontium chloride broth gave equivalent or better results than either Rappaport's or strontium selenite media in recovery of salmonellae from human, animal and environmental samples. The selective property of this medium is based on the inhibition by the strontium ion of non-pathogenic Gram negative bacteria. In particular, it was claimed that in strontium chloride broth Proteus was suppressed to a greater extent than in selenite or tetrathionate broths. Rappaport medium, introduced in 1956, used the selective agents magnesium chloride, malachite green and low pH (321). This medium was not used as widely as tetrathionate or selenite media, but interest was renewed by Vassiliadis et al. (402) who modified the medium by reducing the malachite green (Rappaport R25 240 ...

medium) and used it as a secondary selective enrichment medium. Later the medium was again modified by reducing the malachite green concentration still further to 0.004% (401, 404). This medium was known as Rappaport RIO broth, later as Rappaport-Vassiliadis (RV) medium, with a recommended incubation temperature of 43°C. The R25 modification of Rappaport broth has been shown to be superior to Muller-Kauffmann (MK) tetrathionate and selenite F broths for isolation of salmonellae from chicken giblets (173) and superior to strontium chloride B for isolation of salmonellae from polluted water (174). It was suggested that this medium, incubated at 37°C, should be used routinely as a single enrichment medium. Van Schothorst and Renaud (392) studied the growth of Salmonella in RV medium and suggested that changing one of the ingredients from tryptone to soya peptone would enhance its performance. The RV modification has gained favour over the R25 modification and MK tetrathionate for analysis of foods, following reports by Vassiliadis (398), Tong-pim et al. (386), Northolt et al. (289) and Vassiliadis et al. (399). RV medium was found to be superior to both tetrathionate and selenite cystine broths by AlIen and co-workers (2) in the analysis of frozen shrimp. In the study conducted by Northholt and co-workers, it was noted that MK tetrathionate medium, when prepared in the laboratory from base ingredients, gave equivalent results to RV medium. However, two commercially available dehydrated MK preparations compared poorly to RV. This highlights another reported advantage of RV medium, that is, it is easy to prepare standardised and reproducible batches. In addition, prepared RV medium is stable for at least one month and this is considered an advantage over other media (398). Since various serovars may have different sensitivities to inhibitory substances, it is generally advised that two dissimilar selective enrichments be employed for each test (203). Litchfield (248) concluded in his review that optimum recovery of salmonellae could be obtained by using both selenite cystine and tetrathionate broths in the examination of food samples. Provision for the parallel use of these media is specified in the official methods of many food microbiology laboratories in the USA (19, 134). In Australia, the SAA altered its Salmonella method (AS 1766.2.5) in 1989 and substituted RV medium for tetrathionate broth. In this revised method the RV medium is incubated at 42°C in combination with mannitol selenite cystine broth incubated at 37°C. Considering the sensitivity of some strains of salmonellae to brilliant green dye, it is considered unwise for laboratories to use only

r 1

Stephen Jay, Dianne Davos, Mark Dundas, Elizabeth Frankish and Diane Lightfoot

and open-ended tube (Craigie tube). The method involves the inoculation of mixed cultures into Craigie tubes placed in larger tubes containing the same medium. Motile microorganisms move through twice the depth of medium to reach the outer surface. Since salmonellae travel fastest through the media, they are readily cultured from the surface of the medium in the outer tube. This technique has been recommended (171) as a method of secondary enrichment for salmonellae after culture in selenite F broth and on selective agar media. In the examination of animal feedingstuffs the method was shown to more than double the number of Salmonella isolations. Harper and Shortridge (167) developed a selective and differential motility medium in Craigie tubes for the routine examination of clinical specimens. The medium was a modified Salmonella-Shigella broth in 0.2% agar, and the method was shown to be more effective in isolating salmonellae other than S. Typhi from specimens heavily contaminated with microorganisms of faecal origin. The spread of a culture of salmonellae through the Craigie tubes and surrounding medium is usually detected by blackening caused by H2S production. Other motile microorganisms are either inhibited by bile salts in the medium or migrate more slowly through the tube. In an earlier investigation, Stuart and Pivnick (363) obtained similar results using a modified U tube (test tube with a small bore side-arm) containing a medium based on Rappaport's broth in 0.6% agar. In the examination of food samples, selective motility media in either Craigie tubes or U tubes could prove useful as a primary and/or secondary selective enrichment procedure. The method may not, however, be reliable in isolating strains of salmonellae with low motility rates, e.g. S. Choleraesuis var. Kunzendorf and S. Typhi. Banwart (28) described a refined U tube selective motility system for detecting salmonellae consisting of a glass flask with a central chamber and three side U tubes connected to it. The central chamber contained lactose broth and each side U tube contained different semisolid selective or differential agars overlaid with brain heart infusion broth. In subsequent studies, Banwart (29, 30), Banwart and Kreitzer (32), and Banwart et aI. (31) modified the selective media and demonstrated that Salmonella could be reliably detected in a variety of foods such as egg and poultry products, cake mixes and candies. Moreover, salmonellae could be isolated and identified in 48 h as opposed to 72 h with normal cultural procedures. Fung and Kraft (142) developed a less complex multi-layer motility agar system using a glass flask with a single straight 242 ....

side-arm. The side-arm contained successive layers of solid agar composed of different selective or biochemical media such as selenite cystine agar and triple sugar iron agar. The body of the flask contained lactose broth into which the sample was inoculated. This system was shown to detect small numbers of salmonellae in mixed cultures and poultry products in the presence of large numbers of competitive microorganisms. The major advantages claimed for both systems compared to conventional procedures are that they are rapid, sensitive, provide savings of time, labour and material, and allow the examination of relatively large sample amounts. The disadvantage is that special bulky apparatus is required. In the routine examination of large numbers of samples, the use of such apparatus would be impractical and incur additional incubation costs. A more recent method by De Smedt et aI. (99) appears to alleviate these short comings. Samples undergo a traditional preenrichment procedure after which 0.1 mL aliquots are inoculated onto the surface of a semi-solid, modified RV medium (MSRV) in Petri dishes. Motile bacteria which have migrated over the surface of the medium are confirmed as salmonellae by slide agglutination. This method is considered to be a rapid alternative to standard cultural detection of salmonellae and is reviewed in more detail in the rapid method section of this paper. Differential selective agar media Numerous selective plating media have been recommended for the culture of salmonellae from liquid enrichment broths. These media generally consist of a basic nutritional medium with added dyes, bile salts, antibiotics, and/or other chemicals to inhibit the growth of competitors. In addition, they usually contain an indicator system to differentiate Salmonella from other microorganisms. Indicator systems are often based on H2S production and/or fermentation of a particular carbohydrate such as lactose, sucrose or xylose. Commonly employed agar media and their modifications have been described and reviewed by Litchfield (248), Fagerberg and Averts (123) and Fricker (141). Selective plating media can be differentiated into categories according to their relative inhibitory action on Gram negative microorganisms. Fagerberg and Averts (123) listed in order of increasing selectivity and differentiation eight agar media commonly used (Table 8.14). It is now almost universal practice to use multiple selective agars to optimise the isolation of salmonellae from selective enrichment culture. This approach is used by the FDA, where bismuth sulphite (BS), hektoen enteric (HE) and xylose

! \ j 1 jI J I I! I

I

I

Salmonella

I1

,I I

lysine desoxycholate (XLD) agars are used (20) and by the SAA where XLD and BS agars are used (359). As discussed previously it is a widely held opinion in Australia that the selective agars chosen should not both contain brilliant green dye; e.g. it is considered inappropriate to use both BS and brilliant green (BG) agars. An authoritative report by Andrews et al. (6) describes the comparative efficiency of BG, BS, HE, XLD, Salmonella-Shigella (SS) and desoxycholate (DC) agars for the recovery of salmonellae from foods. BG agar has been reported to give excellent suppression of non-Salmonella enterics, and with the addition of either sulphadiazine or sulphapyridine (BGS agar), Proteus spp. and pseudomonads are greatly inhibited (145, 299). Read and Reyes (326) showed that the quality of the brilliant green dye used in either laboratory prepared or commercial dehydrated media had a significant effect on the performance of BGS agar. In an investigation to determine factors affecting the selectivity of BG agar, Moats and Kinner (271) concluded that more reproducible results could be obtained when brilliant green dye is added after sterilisation of a commercial base medium. Other modifications of BG agar which have been demonstrated to improve isolation of salmonellae include the incorporation of an H2S indicator (272), the addition of novobiocin (270), and the combined antibiotic enrichment with sulphacetamide and mandelic acid (410). It has been also suggested that increasing the incubation temperature from 37-41.5°C or 43°C may further enhance the selectivity of BG agars (326, 410). The BS agar of Wilson and Blair is generally considered to be the best medium to use in combination with other media. In a Canadian study published in 1994 (408), the authors concluded that BS agar should be compulsory for all Salmonella isolation methods. The advantage of BS agar lies in its ability to detect rarer strains of salmonellae which have atypical carbohydrate reactions, for example, those that ferment lactose or sucrose. BS agar does not rely on carbohydrate fermentation to differentiate salmonellae from

other bacteria. Blackburn and Ellis (41), for example, showed that BS agar was especially useful in the isolation of lactose positive salmonellae from dried milk and milk drying plants. Their investigation revealed that 86 (15.6%) of 552 cultures examined were positive in lactose fermentation tests. Almost all salmonellae appear on this medium as black colonies with a characteristic jet black centre due to H2S production. The inhibitory action to coliforms is attributed to the combined effect of bismuth sulphite precipitate and sodium sulphite solution. Although the medium can be used directly after pouring for the isolation of S. Typhi, Cook (67) found that freshly poured plates were inhibitory to other serovars. This inhibition may be reduced by 'ageing' the medium in the refrigerator for 3-4 d after pouring. The 'ageing' of the medium is considered necessary for the production of the characteristic Salmonella colony (387). One of the disadvantages of BSA is that it has a very short shelf life and should be used shortly after preparation. AlIen et al. (3) described a stabilised BSA medium that was comparable with freshly prepared and overnight aged BSA after 3-4 months of storage. Other more recent media which aid the identification of atypical biochemical strains are novobiocin brilliant green glucose (NBG) agar (102) and mannitol lysine crystal violet brilliant green (MLCB) agar. MLCB was first described by Inoue in 1968 at a meeting of microbiologists in Japan and further studied and reported on by van Schothorst et al. (393). These media will differentiate lactose positive salmonellae although the false positive rate'{or MLCB has been considered unacceptably

high

~

is claimeg

.

that

NBG

agar provides a more con~iSten[ differentiation reaction of all salmonellae than does BS agar. Certainly, the reaction on BS agar may vary with Salmonella strain, medium batch, age and manufacturer. In 1990, Rambach (320) described an agar medium (RAM) which was formulated to provide a marked colour differentiation of salmonellae

Table 8.14. Selective agar plating media for salmonellae Category

Medium

Slightly selective

MacConkey agar

Moderately selective and differential

Salmonella-Shigella (SS) agar Desoxycholate citrate(DS) agar Hectoen enteric (HE) agar Xylose lysine desoxycholate (XLD) agar

Highly selective and differential

Bismuth sulphite (BS) agar Brilliant green (BG) agar Brilliant green sulphonamide(BGS) agar

243

I

il

'11

~~

,:r

~i

I

11:

.

.., Stephen Jay, Dianne Davos, Mark Dundas, Elizabeth Frankish and Diane Lightfoot

from other Gram negative bacteria. Salmonellae were red in colour while E. coli were blue, Proteus colourless and Citrobacter violet. S. Typhi was also colourless. Differentiation of salmonellae was achieved by their ability to produce acid from propylene glycol and the incorporation in the medium of the chromogenic substrate X-Gal (5-bromo-4-chloro-3-indolyl-[3-D-galactopyranoside) to allow different coloured colonies to be produced by other members of the Enterobacteriaceae due to their [3-D-galactosidase activity. Due to the presence of [3-D-galactosidase activity in some Salmonella strains, users of RAM must be cautious in assuming all blue-green colonies are non-salmonellae (227). Similarly, Garrick and Smith (146) reported atypical reactions with salmonellae on RAM and also Citrobacter freundii as mimicking Salmonella on RAM. Another approach to selective, differential isolation was reported by Cox and Stallard (71) and later expanded on by Cox in 1993 (70). The agar medium described contained glycerol to differentiate Salmonella from Citrobacter, mannitol, lysine and a H2S detection system. The medium was called lysine mannitol glycerol agar (LMG) and was reported to be capable of the detection of S. Typhi, distinguishing lactose and sucrose positive salmonellae, although not capable of detecting H2S or lysine decarboxylase negative salmonellae. A number of chromogenic agars have also been developed and commercially marketed. Selective agar plates are usually incubated at 37°C for 24 h. Because BS agar is highly inhibitory, incubation for 48 h is usually necessary. It has been suggested that Salmonella isolations can be increased by the reincubation, for an additional 24 h, of BGS agar plates which fail to show growth or have atypical colonies. It is also relevant to note that Salmonella colonies develop a wrinkled appearance on BG or BGS agars when incubated plates are left at ambient temperature for several days. This phenomenon is extremely useful in that it often allows salmonellae to be isolated from a plate which, after initial incubation at 37°C, may be regarded as negative. Recognition of typical colony appearance can be particularly useful where plates are . crowded with lactose fermenting . mIcroorganIsms. It is often recommended that two to three typical colonies should be selected from each plate for biochemical and serological confirmation (19, 20, 134, 203, 359). When examining samples for multiple serovars, at least 20 colonies per plate should be examined. Maximum yields of salmonellae may also be obtained by increasing the number of different selective enrichments and plating media. Harvey and Price (171) claimed

I I I 1

I

r,

L ....

244

that the isolation rate of salmonellae could be dramatically increased by subculturing swabs taken from apparently negative plates into modified Craigie motility tubes containing semisolid nutrient agar. The improved recovery rates were attributed to secondary enrichment in media independent of selective toxic chemicals. Biochemical confirmation There is an extensive number of biochemical tests which can be used to characterise presumptive Salmonella isolates. The reactions which are of diagnostic importance in their identification have been listed previously in Table 8.1. A detailed description of these tests is provided by Edwards and Ewing (122) and Cowan and Steel (69). In the routine examination of foods, identification of cultures can usually be accomplished ~ith a limited number of biochemical tests followed by agglutination tests with polyvalent Salmonella antisera. Some of the more commonly employed confirmatory tests are shown in Table 8.15. Many non-selective differential agar media have been developed as biochemical screening procedures to eliminate false positive isolates. The advantage of multiple test media is that they reduce the number of inoculations and the equipment needed to identify salmonellae rapidly and with reasonable accuracy. The more frequently used media have been reviewed by Litchfield (248) and details are presented in Table 8.16. According to the methods prescribed by the SAA (359), suspect colonies are sub-cultured to peptone water and incubated at 37°C until turbid. This culture is used to inoculate two nutrient agar slopes which are used for serological confirmation and the two biochemical tests recommended, ONPG and lysine decarboxylase (LD). Generally both tests are carried out separately in broth cultures. Proudford (315) developed a simple procedure for performing these tests in a single test tube containing a differential agar medium composed of lysine decarboxylase medium in the butt and ONPG medium in the slant (OL medium). This medium subsequently has been demonstrated to provide a useful method for screening cultures of suspected salmonellae isolated from a wide range offoods (314) and from poultry (64, 96). Workers using the SAA method (359) must recognise that atypical Salmonella may give a positive ONPG or a negative LD reaction. All microorganisms showing a negative ONPG reaction irrespective of the LD reaction or a positive LD reaction irrespective of the ONPG reaction, should be investigated further before concluding they are not salmonellae. Most laboratories use additional biochemical tests as a matter of course. For laboratories using the SAA

Salmonella

method (359) it is easy to extend the range of biochemical tests by: (a) performing an indole test on the peptone water culture used to inoculate the ONPG and lysine decarboxylase tests after continuing its incubation up to 18 to 24 h, (b) performing an oxidase test on the nutrient agar slope. Substituting tryptone water for peptone water will enhance the sensitivity of the indole test. These two additional tests will assist in screening out some microorganisms which mimic Salmonella. Other tests which are useful to include are the Voges-Proskauer and urease tests. Where serovars with atypical biochemical characteristics are suspected, a wider range of confirmation tests should be performed as a matter of routine. There have been a number of combination reaction systems developed commercially to simplify the biochemical identification of Enterobacteriaceae, particularly for laboratories with limited facilities for media preparation. Many of these systems are miniaturised and allow up to 24 biochemical tests to be performed simultaneously. For some laboratories, test kits Table 8.15. Biochemical tests used for confirmation of salmonellae Test

Typical reactions of salmonellae

KCN tolerance Urease production Lysine decarboxylase production

+

+

H,S production

+

Sucrose fermentation Lactose fermentation Dulcitol fermentation

Serological

,

confirmation

Once an isolate has been identified as a Salmonella sp. by biochemical and morphological methods, laboratories must confirm this identification serologically. In most cases commercially available polyvalent 0 and polyvalent H antisera are used. If positive results are obtained, the cultures are forwarded to the reference centre for complete identification. Some laboratories may wish to proceed further with identification and the range of commercial antisera available makes possible the identification of many commonly occurring serovars. Laboratories using commercially prepared antisera should precisely follow the manufacturers' instructions. Cultures should, however, always be forwarded to a Salmonella Reference Laboratory for verification and inclusion in the surveillance reports prepared by that laboratory. With the common serovars which are found from a wide variety of sources, e.g. S. Typhimurium, there is a need to further characterise strains within the particular serovar. With phage typing using standard sets of typing bacteriophages it is possible to subdivide strains of the serovar into phage types which allow for more detailed epidemiological tracing.

I

Enumeration The numbers of salmonellae in foods may be estimated by direct plating onto selective agar or by the most probable number (MPN) technique. Direct plating procedures are considered impractical because the development of typical colonies may be obscured by the presence of food

p-D-Galactosidase (ONPG) production Citrate utilisation

such as the API system (Biomerieux), Microbact biochemical test strips (Oxoid), Minitek systems (Becton, Dickinson and Company), and the Micro ID system (General Diagnostics) offer a practical and economical method for the rapid identification of suspected Salmonella isolates.

+

Indole formation

Table 8.16. Non-selective differential agar media for the biochemical characterisation

of salmonellae

Medium Triple sugar iron agar

H,S production;

Triple sugar iron urea agar

As for TSI agar + urease activity.

fermentation

(acid + gas) of dextrose,

Lysine iron agar

H,S production;

decarboxylation

Kligler's iron agar

H,S production;

fermentation

Gillies' two tube media

i) fermentation

of glucose and mannitol;

ii) fermentation

of salicin and sucrose; H,S production;

of lysine; fermentation

lactose and sucrose.

of dextrose.

of dextrose and lactose. urease activity motility.

Kohn's two tube media

i) fermentation of dextrose and mannitol; urease activity ii) fermentation of sucrose and salicin; H,S production; indole formation;

Dulcitol sucrose salicin iron urea agar

fermentation

motility

of dulcitol, sucrose and salicin; urease activity; H,S production

245

.

., Stephen Jay, Dianne Davos, Mark Dundas, Elizabeth Frankish and Diane Ughtfoot

material and/or the growth of large numbers of the other organisms usually present. By comparison, the MPN procedure is useful in enumerating salmonellae in food provided that consideration is given to the nature of the product, proper preparation of the sample to eliminate clumps and to ensure dispersion of salmonellae, and proper selection of preenrichment and selective enrichment broths and selective agar media (248). The ability to use an MPN method is limited in most routine laboratories, however, because of increased costs, time and demand on incubation space. Rapid methods Due to the time involved (4-5 days) for clearance of foods for Salmonella by conventional cultural methods, the development of more rapid detection techniques generated considerable interest and development of such methods has and will continue into the foreseeable future. According to Swaminathan et al. (367) the possible advantages of rapid detection of salmonellae are: (a) reduction of warehousing costs for the food industry, (b) ability to respond quickly to contamination problems, (c) increased testing of food products for the presence of Salmonella, thus increasing food safety, and (d) an opportunity to buy raw meats and poultry and perishable food ingredients that have been tested and found free of Salmonella. Rapid methods for Salmonella detection from food have included fluorescent antibody (FA) stains, enzyme immunoassay (EIA), enrichment serology, immuno-sensors, fluorogenic staining, bacteriophage methods, hydrophobic grid membrane filtration, electrical measurements of metabolic by-products, shortened liquid enrichment, geneprobes and polymerase chain reaction (PCR). These techniques have been reviewed by Ibrahim and Fleet (197), Feng (126), Blackburn (42) and in Chapter 5 of this book. Enrichment serology. Semi-solid media have been used for many decades for isolation of Salmonella from mixed cultures based on their ability to migrate faster than most competitive bacteria. Semi-solid media and serological detection (enrichment serology) were at first shown to have unacceptable levels of false negatives. Attempts to rectify this led to extended cultural enrichment, giving the technique little advantage over traditional cultural methods (47). During the late 1980s however, significant advances in enrichment serology techniques occurred. Modified semi-solid Rappaport- Vassiliadis medium. A selective motility technique using

serological detection was reported in 1987 by De Smedt et al. (99), allowing presumptive detection of Salmonella within 48 h. Preenrichment was for 20 h, followed by inoculation of the surface of a MSRV agar plate and incubation for 24 h to allow for selective migration of motile salmonellae. Presumptive detection was performed by serological tests on migrating cultures. Significantly, this method required no additional equipment or expense above that of traditional cultural procedures. In a subsequent study, De Smedt and Bolderdijk (97) reported that as few as 60 salmonellae per mL of preenrichment culture, even in the presence of 107 competitive bacteria, were needed for detection of Salmonella using MSRV. A collaborative study by 15 laboratories where an MSRV method, was compared to a traditional cultural procedure for analysis of cocoa, chocolate and sugar confectionery was published in 1990 (98). The MSRV procedure was pre-enrichment for 20 h at 37°C followed by inoculation of three drops of the pre-enrichment culture onto the surface of a MRSV agar plate which was incubated at 42°C for 24 h. In addition, after 8 h of incubation, three drops from each of the two selective broths used in the traditional procedure were inoculated onto MSRV. These MSRV plates were incubated at 42°C for 16 h. Any migrated cultures were confirmed as Salmonella by biochemical and serological means. In this study it was discovered that some strains of S. Typhimurium did not fully develop flagella and therefore did not migrate. Each participating laboratory was requested to check all MSRV cultures, whether migrating or non-migrating. No significant difference between the traditional cultural procedure and the MSRV procedure was noted after all Salmonella positive results (migratory and non-migratory) from the MSRV technique were taken into account. In another collaborative study (101), similar methods were compared and a variety of foods used. Again no statistical difference between MSRV and a traditional cultural procedure was noted. These two collaborative studies tested all MSRV cultures, whether migratory or nonmigratory for the presence of Salmonella. This negated the advantage of the original MSRV procedure, that is, all samples without migratory zones are considered negative. Work by O'Donoghue and Winn (294) published in 1993, showed that a MSRV method was equivalent to an in-house conventional culture procedure for the detection of Salmonella in low and high moisture foods. In this study only MSRV agar plates exhibiting migratory growth zones were tested for the presence of Salmonella. The AOAC has adopted a MSRV procedure for

246 ~ "'---

Salmonella

."

1

analysis of cocoa and chocolate as first action status in 1994 (20, 100). The method adopted involves pre-enrichment for 20 :!: 2 h at 35°C followedby inoculation of this culture onto MSRV and into tetrathionate broth. After 24 h of incubation at 35°C the MSRV agar plate is checked for a migratory zone which, if present, is tested for Salmonella. The tetrathionate broth is inoculated onto MSRV after 8 h of incubation at 35°C.Any migratory zone is likewise checked for the presence of Salmonella after 16 h incubation at 42°C. It should be noted that this procedure does not include testing of non-migratory MSRV cultures. Presumptive results are obtained within 48 h. A brief Australian report published in 1996 by Dziedziczak and Kabilafkas indicated excellent results when analysing milk powder samples using a method based on MSRV. Negative results were obtained after pre-enrichment and 20 h of incubation of MSRV (105). Results from Perales and Erkiaga (306) suggest that semi-solid Rappaport broth is superior to MSRV when used at 35°C. Test methods based on the use of MRSV have AOAC final action status (20). 1-2 TeseM. A product from BioControl Systems Inc. USA to detect salmonellae in foods is called the 1-2 TestTM. The test apparatus consists of a two chambered plastic vial. One chamber (enrichment chamber) is used for inoculation and can be supplied containing tetrathionate-brilliant green broth. The other chamber (motility chamber) contains a semi-solid motility agar. The original method employed pre-enrichment of samples for 18 h at 35°C after which 0.1 mL of pre-enrichment culture was added to the enrichment chamber. Polyclonal flagellar antiserum was added to the top of the motility chamber and the test incubated at 35°C to 37°C. Salmonella detection is apparent from a white line of immobilised salmonellae in the motility chamber. The test was checked after 8 and 24 h of incubation. A study by D'Aoust and Sewell (91) in which 186 foods were analysed by a standard cultural method and the 1-2 TestTM revealed that the 1-2 TestTM detected 25 Salmonella positive samples while the culture method detected 43. However, two raw poultry samples found to be Salmonella positive by the 1-2 TestTM were found to be negative by the cultural method. It is interesting to note that although the 1-2 TestTM detected salmonellae in only 25 samples, salmonellae were isolated from 38 samples following subculture from the 1-2 TestTM enrichment chamber. D'Aoust and Sewell made some suggestions for improving the performance of the 1-2 TestTM and recommended further study. Although the 1-2 TestTM received AOAC first action status for foods in 1989 (133), research into

enrichment protocols which would improve its performance continued (186, 281, 296). These efforts lead to a modified procedure for raw flesh and highly contaminated foods which used a selective enrichment step performed outside the 1-2 TestTMinoculation chamber. In this procedure, after pre-enrichment, selective enrichment is performed in tetrathionate-brilliant green broth incubated at 42°C for 6-8 h. The inoculation chamber of the 1-2 TestTM is emptied of its supplied broth (or the 1-2 TestTM is supplied with an empty chamber from BioControl) and 1.5 mL of selective enrichment culture is inoculated into the chamber. The test is then incubated at 35°C for 14-30 h, although absence of an immobilised band of cells at 14 h is considered negative. This modification has received AOAC first action approval in 1994 (20, 409). The procedure for foods other than ra w flesh and highly contaminated foods remained unchanged and the selective enrichment stage of the test is performed with the tetrathionate brilliant green broth supplied in the 1-2 TestTM inoculation chamber. Oxoid Salmonella Rapid Test. Oxoid Ltd, UK have also developed a rapid selective motility enrichment serology method in kit form called the Oxoid Salmonella Rapid Test (OSRT). The kit detects motile salmonellae in raw ingredients, finished products and in factory environmental samples. The test is performed in a culture vessel consisting of a plastic culture jar containing two tubes. Each tube consists of two parts separated by a porous partition. For each tube the section below the partition contains a different selective medium and the section above the partition contains a different indicator medium. Samples are pre-enriched in the usual manner after which a 1 mL aliquot of the pre-enrichmentculture is used to inoculate a special Salmonella selective medium contained in a culture vessel. The culture vessel is then incubated at 41°C for 24 h. If salmonellae are present in the sample they migrate from the culture vessel into the tubes, first through the selective media and then into the indicator media. A positive reaction which is indicated by a colour change, must be further confirmed using a latex agglutination test (also supplied by Oxoid). Presumptive positive results can only be reported after completion of the agglutination test. The kit, like most traditional cultural methods, uses two dissimilar selective media (RV medium and a modified lysine iron desoxycholate medium) and differential indicator media (a modified lysine iron cystine neutral red medium and a modified Brilliant green medium). Presumptive results are provided in 48 hand confirmation is easily done by direct sub-culture of the indicator media. Performance results

247

'I j 11

.

I

Stephen Jay, Dianne Davos, Mark Dundas, Elizabeth Frankish and Diane Lightfoot

published by Holbrook et al. (178, 179) indicate that this technique has good selectivity and sensitivity when compared to traditional techniques and other rapid methods. In a precollaborative study the OSRT yielded an unacceptable false negative rate (8). Imrnunoassays Radioimmunoassay. Radioimmunoassay methods have been developed which are both selective and sensitive (199). One such method by Stew art et al. (361) employed the specific inhibition of dulcitol fermentation by Salmonella poly H agglutinating serum. Carbon 14 labelled dulcitol was used and salmonellae were detected by the absence of production of 14C02 from cu1tures with added agglutinating serum compared to control cultures. However, such assays employ radioisotopes which present difficulties with safety and disposal. In addition, the equipment required is expensive. Enzyme immunoassay (EIA). The EIA is a well established technique for assaying antigens or antibodies. It has the advantage over radioimmunoassays in that it does not rely on radioactivity as the detection system. Antibodies labelled with an enzyme detect antigen by enzymatic conversion of a substrate, usually resulting in a colour change which can be read visually or spectrophotometrically. Heterogeneous EIAs are used to detect bacterial antigens. In this type of assay, when the enzyme labelled antibody (ELA) binds in the test system, the enzyme label remains unchanged. The unbound ELA must be removed from the test system, leaving only bound ELA prior to the addition of the substrate. This is usually achieved by having one of the reactants fixed to a solid matrix, such as the surface of a micro titre plate well, a plastic bead or other solid matrix like titanous hydroxide (198), enabling unbound ELA to be simply washed from the system with a buffered detergent solution. Krysinski and Heimsch (226) first reported the use of an EIA for detection of salmonellae in foods in 1977. Since then there has been considerable development (262, 269, 348, 366) resulting in commercial EIA kits and excellent reviews of the subject have been published by Swaminathan et al. (367) and Ibrahim (196). For most laboratories, commercial kits offer a convenient way of introducing EIAs for routine use. The TECRA@ Salmonella visual immunoassay from TECRA@ Diagnostics Australia (TECRA@ International Pty Ltd) detects flagella antigens using the sandwich principle of immunoassay. Salmonellae are detected after pre-enrichment, selective enrichment and culture in M broth to ensure flagella development. Cultural procedures and the EIA can be completed in 48 h. The 248 ~---=

TECRA@ EIA is accepted by the Australian Department of Primary Industry and Energy for testing foods for compliance with export standards, and has AOAC Official First Action status (137). Studies of the TECRA@ EIA by Flowers et al. (137), Hughes et al. (184) and Jay and Comar (212) have shown it to be a sensitive and reliable method. Specificity is high, with false positive rates between 2% and 7% reported. Hughes et al. (184) and Jay and Comar (212) reported no false negatives while Flowers et al. (137) reported no significant difference between the TECRA@ EIA and the FDA cultural procedure (19). A positive result is a definite blue-green colour in the microtitre well. The assay can be read visually with no difficulty in interpretation. This feature of the TECRA@ EIA is &n advantage over assays which require spectrophotometric equipment. The TECRA@ EIA is supplied with a modification of the standard 96 well microtitre plate produced by Dynatech Laboratories Inc., USA. The micro titre plate consists of a base plate with the wells as separate strips of 12 units. Wells, used as complete or part strips, are inserted into the base plate to perform the EIA, thus making the EIA more flexible than with the standard microtitre plate format. TECRA@ have also released TECRA@ ULTlMATM which is a visual EIA that provides a result in less than 36 h, reduces selective enrichment to a single broth and eliminates the need to use M broth. Organon Teknika Corporation, USA, released the first commercially available Salmonella EIA in Australia, the Salmonella Bio-EnzaBeadTM Screening Kit. The assay detected flagella antigens, also by the sandwich principle and utilised magnetic force to transfer the solid matrix through each reagent. The solid matrix was a polycarbonate coated metal bead to which monoclonal antibodies were bound. The EIA was extensively studied (88, 90, 11O, 115, 131, 132, 135, 385) and received AOAC official first action approval (13, 132) but has since been replaced by the Salmonella- TePM ELISA test system (bioMerieux SA.). The Salmonella-TekTM detects flagella antigens monoclonal antibodies coated onto the inside of microtitre wells instead of onto magnetic beads. The Salmonella- Tek TM assay has been shown to consistently detect lower levels of Salmonella in mixed culture than the Bio-EnzaBeadTM and a 99.1 % agreement between the Salmonella- TePM method and the standard AOAC culture method was obtained (81). A study by van Poucke (391) showed Salmonella detection as low as 1 to 5 cfu per 25 g food sample by the Salmonella- TeFM kit method. When used with raw chicken samples, the

Salmonella

,

Salmonella-TekTMgave a high incidence of false positive reactions (356). Modifications to the ElA method were introduced and included elevated temperature incubation (42°C) of the tetrathionate broths to inhibit growth of competing bacteria; addition of novobiocin into the M broth to reduce growth of Proteus species and the elimination of the microtitre plate agitation and centrifuging of the post enrichment M broths to simplify the assay. The modified Salmonella-TekTM was granted first action approval by the AOAC in 1994 (20, 111). BioMerieux S.A. have developed a Salmonella EIA to be used with the Vitek lmmuno Diagnostic Assay System (VIDA8") to detect Salmonella in food and environmental samples. The ElA is performed in the fully automated VIDAS@ instrument. Monoclonal capture antibodies coat the internal surface of a disposable pipette-tiplike device. This device acts as the solid phase and the pipette for the assay. Test sample from the reagent strip is drawn into the antibody coated device followed by washing and subsequent injections of ElA reagents. A fluorescent substrate is exposed to the enzyme conjugate remaining in the device and a relative fluorescence value is calculated automatically by the VIDAS@ computer. Assay results are obtained in 45 min, following a 42 h cultural procedure. The VIDAS@ Salmonella Assay was compared in a study published in 1994 with a conventional cultural method for Salmonella detection in naturally and artificially contaminated foods. A 92.9% overall agreement between the two methods and a sensitivity of 106 to 108 Salmonella per mL of M broth was obtained (43). A later study in 1997 compared the VIDA8" Salmonella Assay with a conventional cultural method for Salmonella detection in naturally and artificially contaminated foods with 99% overall agreement between the two methods (82). The VIDAS@ Salmonella has gained AOAC first action status (20, 82) and it is now a widely accepted method. Elisa Systems (Queensland Australia) produce an EIA for analysis of Salmonella in food and related products. Immuno-captureII mmuno-concentration. The ElAs so far discussed rely on the standard cultural procedures of pre-enrichment and selective enrichment to provide enough Salmonella cells which can then be detected. An ElA technology which would enable detection at the resuscitation culture stage, would provide even more rapid results (184). Two assays have been commercialised that take advantage of this approach. A dipstick-based assay (TECRA@ UniqueTM) has been developed by TECRA@ Diagnostics in

Australia to detect Salmonella in foods that utilises an antibody-coated dipstick which is initially used to capture Salmonella from a preenrichment sample. After washing, the dipstick is transferred to M broth for 4 h to allow replication of the captured Salmonella. The dipstick is then transferred through ElA reagents and presence of Salmonella detected via colour development on the dipstick. All reagents are supplied in a 6 tube test module and the assay is completed in 4-5 h after a 16 h pre-enrichment protocol. Total test time is 22 h. Evaluations of the TECRA@ UniqueTM in artificially contaminated foods have given a sensitivity of 1.5 x 104to 6.5 X 104 cfu per mL of enrichment culture (214) and a 97.5% agreement between the assay and conventional cultural methods (17). The TECRA@ UniqueTM has AOAC first action status (183). TECRA@ have also developed an automated system known as the Unique PlusTM which uses an automated immunoenrichment to eliminate further selective enrichment steps. The VIDAS@ lmmuno-Concentration Salmonella (lCS) has been developed by bioMerieux S.A. It is an immunological enrichment technique that relies on immunoconcentration and a result can be obtained within 24 h of sample enrichment. The sample is enriched overnight and then 0.8 mL is transferred to the lCS strip that performs the immunoconcentration. lmmuno-concentration takes 40 min and the concentrated salmonellae are then transferred by pipette into lCS broth (bioMerieux) and incubated for 5-6 h. One mL of the broth is then heat treated and assayed in the VIDAS@ using the VIDA8" Salmonella strip. The method has been recommended for AOAC first action status (238). Alternatively, the immuno-concentrated salmonellae can be transferred via a swab from the lCS strip onto specialised agar plates, the agar plates incubated and typical Salmonella colonies confirmed by the usual means. Two methods using this approach have been recommended for AOAC first action status (237, 239). DNA-DNA hybridisation - gene probes. DNA probes are single strands of DNA which are used to detect complementary DNA in a target. This technology has the potential to be used to detect any microorganism, although discovering appropriate DNA sequences for use is often a complex task. A paper by Fitts (128) describes the typical DNA hybridisation assay: a. The bacteria or other organisms are applied to a solid support such as a nitrocellulose membrane filter. This can be done by simply spotting a bacterial culture onto the filter, by 249 \

'I

Stephen Jay, Dianne Davos, Mark Dundas, Elizabeth Frankish and Diane Lightfoot

touching the filter to colonies on agar plates, or by actual filtration through the membrane. b. The bacteria are lysed to release their DNA, and the bacterial DNA is denatured into separate strands. c. The separated DNA strands are then fixed to the solid support so that they will not wash away later. d. The filters are soaked in a hybridisation cocktail that contains probe DNA. The probe DNA is most often a double-stranded DNA segment that is labelled by nick-translation. During nick-translation, a DNA strand is first nicked by a DNAase, then DNA synthesis is allowed to occur at the site of the nick. The nicked strand serves as a primer for DNA synthesis, while the opposite strand in the duplex molecule serves as the template. Nucleotides must be added for synthesis to occur. These nucleotides can be labelled with a radioisotope, or with a nonisotopic reporter molecule. The double-stranded probe molecule must be denatured before hybridisation. e. When a probe is mixed with test DNA on the nitrocellulose filter, the probe sequence finds its complementary sequence in the test DNA and forms hybrid molecules. These hybrids contain one radio-labelled strand that can then be detected by autoradiography. The first use of a DNA probe to detect salmonellae in foods was by Fitts et al. (129). A DNA probe of 10 unique sequences isolated from S. Typhimurium reacted with 23 Salmonella species and not with five other members of the Enterobacteriaceae. Further extensive testing, using hundreds of Salmonella reference strains from the Centers for Disease Control (Atlanta, GA, USA) collection demonstrated that two of the sequences reacted with all Salmonella, three with over 95% and five with under 50%. A probe comprising all sequences, has been developed for routine use as the Gene- Trak TM Salmonella Assay manufactured by Gene-Trak Systems (Framingham, MA, USA). The kit used radioactive labelled DNA which detected Salmonella at 1 x 105 organisms per mL of enrichment broth (106). Results were obtained in 46-53 h and the assay could analyse 96 samples simultaneously. Data showed the technique to be specific and sensitive (97, 130, 136) and the Gene-TrakTM Salmonella assay received AOAC official first action approval in 1988 followed by final action status in 1996 (20). A major advantage of a DNA probe is that a microorganism always carries its unique DNA. DNA probes are not dependent on the expression of antigens or enzymes which may be subject to environmental influences. The use of a combi250

nation of DNA sequences, as in the Gene-TrakTM system, provides great assurance that even mutated strains of Salmonella will not be missed. The use of radioactivity in DNA probes was a disadvantage, especially in the food industry. The expense of the equipment needed to quantify radioactivity is also a disadvantage. To this end Gene- TrakTM has replaced the isotopic kit with a DNA hybridisation assay that utilises an enzymatically labelled probe and a colorimetric end point. The colorimetric Gene- TrakTMSalmonella assay consists of a polydeoxyadenosine-tailed capture probe and a fluoresceinated detector probe which target regions of ribosonal RNA (rRNA) that are unique to Salmonella. If rRNA is present in the sample, hybridisation takes place. The probe target complexes are captured with a polydeocythynidine-coated dipstick and the entire complex is detected by an anti-fluoresceinhorseradish peroxidase conjugate and a colorimetric enzyme substrate. For food analysis, the assay can be completed in 2.5 h, following a 44 h cultural period (59). Comparative studies have demonstrated that the colorimetric hybridisation method was equivalent to standard cultural methods (59, 80, 337, 418) and close correlation to various rapid methods (28, 317, 356). The assay has been reported to be non-reactive with sub-genus v. Salmonellae (80) and was modified to include a probe to detect this sub-genus. The modified colorimetric hybridisation assay was granted official

first

action

method

by the

AOAC

in 1992

(80). D'Aoust et al. (86) reported in 1995 good results with the assay and suggested a modification that would, according to the authors, enhance the performance of the assay "to a level of unfailing sensitivity and specificity" (86). DNA or RNA probes provide rapid detection of Salmonella in foods, however, specificity is limited and a preenrichment step is still needed to improve sensitivity. Polymerase chain reaction (PCR). Polymerase chain reaction (PCR) is a method for creating multiple copies of a target DNA sequence. The PCR reaction relies on three significant steps; denaturation of the DNA strands, annealing of the primer to the template and the synthesis of new strands via DNA polymerase. The PCR reaction occurs in a series of cycles, resulting in duplication of each piece of DNA with each repetition of the cycle. PCR technology has advanced since the development of a heat stable DNA polymerase and an automated machine called a thermal cycler. An enzyme isolated from Thermus aquaticus, a thermophilic bacterium, led to the development of Taq DNA polymerase. This

Salmonella

polymerase can withstand the high temperatures involved in the PCR process, therefore only has to be added to the sample once and not at every repetition of the PCR cycle. PCR denaturing, annealing and strand synthesis occurs at different temperatures, thus the process is carried out in a thermal cycler. The thermal cycler allows the denaturation of the two DNA strands to occur at 90-95°C, annealing of the primers to the DNA strands at 55°C and strand synthesis via Taq polymerase at around 75°C. As the polymerase chain reaction (PCR) targets the bacterial DNA directly, the degree of specificity, sensitivity and speed of detection of an organism within a sample can be increased. Recently, PCR primers that target specific regions of the Salmonella gene have been developed and investigated. A multiplex PCR technique for the detection of Salmonella on chicken skin was compared with standard cultural methods. The PCR technique was found to be more sensitive, 15% positives to 7.4% positives and quicker, 24 h compared to 3-4 d than the conventional method (253). Primer sets such as the oligonucleutide S18 and S19 from the omp gene have successfully detected 40 Salmonella serovars in vitro (229) and detection of Salmonella via these primers in biological samples is under investigation. Detection of Salmonella species by PCR in contaminated oyster samples has been investigated. Mter a pre-enrichment step of 3 h, 1-10 cells of Salmonella species were detected by the PCR (35). Unique PCR primers to detect Salmonella typhi have been developed based on the 5S-23S spacer region of a cloned DNA fragment. The PCR method showed a sensitivity of approximately 40 Salmonella Typhi cells in a spiked chicken rice sample (422). PCR will also target DNA from dead cells within a sample, however in practice there needs to be around 103 cfu of target bacteria per mL of pre-enrichment culture for detection to occur and this will almost certainly mean that live Salmonella cells exist in the sample tested. PCR also requires strict hygiene to minimise the possibility of contamination of the assay with genetic material from extraneous sources. The thermocycling step is usually performed in a separate room to other parts of the Salmonella assay procedure. Some laboratories use a type of cleanroom for this procedure. It has been demonstrated by Stefanovicova et al (360) that PCR can be utilised for the confirmation of presumptive Salmonella colonies, thus reducing the time required for confirmation to a maximum of six hours. A commercial PCR kit for the detection of Salmonella has been developed by QualiconTM a

subsidiary of E. 1. Du Pont de Nemours and Company, Wilmington, Delaware, USA. This is the BA.)('"Salmonella system. A result is available 28 h after commencing analysis (36). A comparative study of the BA.)('" gel-based assay and a traditional culture test method has been completed by Campden and Chorleywood Food Research Institute (37) and found that the BA.)('" assay was highly specific and sensitive with the BA.)('" system generating more positive detections than the cultural test method. Similar results were obtained by a study undertaken by Bailey in 1998 (24). DuPont Qualicon have developed an automated system for performing the assay that provides for many improvements. The automated instrument integrates the amplification and detection steps and a closed-tube system eliminates the possibility of cross contamination. Custom software, that is continually being upgraded to increase sensitivity, analyses the results of the PCR. The AOAC Research Institute awarded the gel-based assay performance tested method status in 1998 (275) and the automated system is currently undergoing evaluation. The agent for the BA.)('"in Australia is Oxoid Australia Pty Ltd. Electrical measurements - conductance. This technique relies on the change of conductance which occurs in a medium as microorganisms grow. Microbial growth in nutrient media results in an accumulation of small, highly charged molecules such as amino acids, fatty acids and organic acids. The electrical conductance of microbiological media alters with these changes in media constitution. Usually conductance and capacitance increase and impedance decreases. These changes can be detected by appropriate electronic detection systems. The three commercially available systems are the Bactometer (bioMerieux, Australia), Malthus-AT (Radiometer Ltd., Manor Royal, Crawley, West Sussex) and RABIT impediometer (Don Whitley Scientific, Australia). All are computer driven automated systems. Conductance detection of salmonellae was first reported by Easter et a1. in 1982 (108). The reduction of trimethylamine-N-oxide (TMAO) to trimethylamine (TMA) by salmonellae was used to produce a large change in medium conductance. Later, Easter and Gibson (107) proposed a method for Salmonella detection in food, using a pre-enrichment broth of buffered peptone water with 0.1 % TMAO and 0.5% dulcitol at pH 7.2 (BPW/T/D) and selective enrichment in selenite cystine broth with 0.5% of sodium biselenite, 0.5% dulcitol instead of lactose and 0.5% TMAO (SC/T/D). Since most Salmonella ferment dulcitol but not lactose, dulcitol was used in place of lactose in 251

Stephen Jay, Dianne Davos, Mark Dundas, Elizabeth Frankish and Diane Lightfoot the medium in order to enhance the conductance change. Conductance was measured by a prototype device, a Malthus instrument or a Bactometer. For Salmonella detection, the magnitude of conductance and the time for this to occur are important. Normal conductance responses for salmonellae in SC/T/D are approximately 500 pS and a maximum rate of conductance change of 15-25 3rS/l0 min (295). The conductance method was in complete agreement with conventional cultural methods for the naturally and artificially contaminated foods tested. However, some Citrobacter strains were found to produce positive conductance responses. The SC/T/D medium was modified by Gibson (152) and Ogden and Cann (295) chiefly by replacing dulcitol with mannitol (CS/T/M). In addition, the selenite concentration was reduced from 0.5% to 0.4%. The use of mannitol enabled dulcitol negative salmonellae to be detected. Gibson (152) recommended resuscitation in buffered peptone water with added mannitol and dimethylsulphoxide (DMSO). DMSO is less expensive than TMAO and successfully induces TMAO reductase. Some strains of Salmonella tested were not detected using SC/T/M and moreover some Citrobacter and E. coli strains gave positive responses (295). Accordingly it was recommended that both SC/T/D and SC/T/M should be used in conjunction with one another. Arnott et al. (21) in 1988 described the use of a modified lysine decarboxylase medium (MLD) used in combination with SC/T/M for the detection of salmonellae in confectionery products. This medium contained yeast extract, glucose, L-Iysine, ferrous ammonium sulphate, sodium thiosulphate and 0.08% sodium biselenite at pH 6.1. The two media detected all salmonellae tested except Salmonella Pullorum. Microorganisms which gave false positive reactions with the system were strains of C. freundii, E. coli, Klebsiella pneumoniae and Serratia marscescens. Pugh et al. (316) claim to have increased the confidence of conductance detection of salmonellae. Their approach was to use, after preenrichment, SC/T/D and MLD as previously described and as an addition include bacteriophages specific for Salmonella. The assay then comprised SC/T/D and MLD with and without bacteriophages. The bacteriophages used were Felix 0-1 and G47 of Gudel and Fey (163). Two conductance responses are produced if salmonellae are present; the normal response for Salmonella and an altered response for bacteriophage effected Salmonella. The combination of the two response curves provides a more sensitive detection criterion.

252

l

A comparison between conventional culture method and a conductance method using SC/T/D and a lysine medium demonstrated that the methods were equivalent for analysis of animal feeds (353). A conductance method using SC/T/D and MLD has been given official first action status by the AOAC (153) in 1992. Salmonellae can be recovered from the Bactometer, Malthus and RABIT systems by direct plating onto selective agar for confirmation and further identification. The conductance methods developed to date can screen for salmonellae within 48 h, 18-24 h for resuscitation and up to 24 h for detection of the conductance change in the detection medium. At present, the electronic equipment is expensive and this is likely to restrict its use. Despite a great initial interest and much published work, this methodology has not gained widespread acceptance. Hydrophobicgrid membrane filter method. The hydrophobic grid membrane filter (HGMF) allows the enumeration of a higher number of colonies than on a conventional filter. The hydrophobic grid restricts the lateral spread of most microbial colonies and this property makes possible the detection of specific microorganisms even in the presence of quite high levels of background microflora. The HGMF also provides a suitable subject for computer scanning since the colonies are of regular size and appear only in defined areas of the filter (between the grid lines). Detection of salmonellae in foods by HGMF was reported by Entis et al. (119). This method employed a pre-enrichment step of 18-24 h, 6 h selective enrichment in broth followed by filtration through a HGMF. The HGMF was placed onto selective agar and incubated for 24 h. Replicate plating was then performed using fresh HGMF transferred to a range of selective agars with incubation for a further 24 h. Suspect colonies were then sub-cultured and confirmed biochemically and serologically. This method achieved results in 2-3 d after sample receipt and was able to be automated. A manual version of this method was adopted as Official First Action by the AOAC in 1984 (116). The HGMF method was further improved by using immunological detection (58). An enzyme labelled antibody (ELA) stain was used to directly detect Salmonella colonies on the HGMF. This eliminated the need for replicate plating to a range of selective agars. The ELA comprised pooled Spicer-Edwards H antisera and conjugate (HRP-protein A). Mter the HGMF is stained it can be submitted to computer counting. Using ELA detection, results are available in 48 h. False positive results can occur due to strains of C.

Salmonella

[

freundii being found to cross-react with the ELA. This method was updated in 1990 and given official interim first action status for all foods (117). This update utilised EF-18 agar as a replacement for selective lysine agar (SLA). Entis and Boleszczuk (118) reported the equivalence of the HGMF method using EF-18 agar to the AOAC/BAM Salmonella reference method. The HGMF procedure using EF-18 agar currently has first status from the AOAC (20). Magnetic separation. Magnetic separation involves the separation and concentration of specific target bacterial cells from a mixed suspension. Immunomagnetic separation (IMS) utilises antibody-coated magnetisable beads to trap target cells and a high magnetic field to separate the target cells from a mixed suspension. IMS is a rapid selective enrichment technique. Subsequent selective plating, or rapid detection techniques, are required for the detection of Salmonella. Parmar et al. (305) used IMS and conductance measurements to detect Salmonella in milk powder. Salmonella Enteritidis was detected in 7.5 h, after a 6 h incubation period, from an initial inoculum size of 20 cells per mL of pre-enrichment culture. DynabeadsTM, from Dynal Biotech, Oslo, Norway are superparamagnetic, polystyrene beads coated with anti-salmonella serum. These can be used to detect Salmonella from 6 h preenrichments of minced meat (405). DynabeadsTM Anti-Salmonella (Dynal, Oslo, Norway) is a commercially available immunomagnetic separation procedure: 20 fll of the DynabeadsTM are added to 1 mL of an overnight pre-enrichment broth and immunocapture occurs for 10 minutes. A magnetic field is applied resulting in capture and isolation of the bead target bacteria complex, the beads are washed and plated onto selective agar. In a study to detect Salmonella from 180 poultry samples, the IMS technique proved superior (135 positives) to the MSRV method (33 positives) and the conventional Salmonella method (98 positives) (78). IMS used to detect Salmonella in 45 spiked skim milk powders showed no false negative results, compared with seven for selenite cysteine, two for RV and one for Muller-Kauffman tetrathionate enrichment broths (105). Further performance studies conducted by Cudjoe et al. (79) and Shaw et al. (340) demonstrated that Dynabeads showed superior recoveries to the standard cultural techniques. Rapid methodology overview Regardless of the methods and techniques employed, isolation and culturing of salmonellae from samples remains important for final

confirmation and complete identification. Identification of isolates is of particular importance for epidemiological data collection. Accordingly, it remains desirable that all samples positive by rapid methods should be confirmed by sub-culture of an appropriate portion of the test. This requires that selective enrichment and/or enrichment media are kept until completion of the assay. While the culture of salmonellae remains the accepted procedure for confirming the presence of Salmonella, some rapid techniques will always yield unconfirmed positive results. Indeed many investigators of rapid methods conclude that unconfirmed positive results, which cannot be satisfactorily explained as false positives, may in fact be true Salmonella detections, which are unable to be confirmed culturally. This is not surprising, as it is well understood that for a variety of reasons traditional cultural procedures will not always detect small numbers of salmonellae in certain samples. The factors that can influence recovery rates include the sensitivity of the methods, the susceptibility of the strains of salmonellae to inhibitors in the food or added to the media, over growth by competitors during incubation, etc. Despite these deficiencies, cultural confirmation will remain the criterion of Salmonella detection until another method or methods becomes widely accepted as equivalent or an improvement. There is no doubt about the benefits of rapid Salmonella detection (at a minimum it allows a faster response time to contamination). However, in order to gain acceptance, rapid methods must be at least as sensitive as cultural methods, easy to perform and relatively inexpensive. Automation is not essential but is an advantage. For a rapid method to be considered, a significant advance over traditional cultural methods, the detection time must be within 48 h and preferably within 24 h. Most rapid methods now meet this criterion. Introducing new techniques of any type into a laboratory should not be done without consideration of the costs and benefits. Traditional methods of Salmonella analysis are often claimed to be laborious, complicated and in some cases beyond the capabilities of small laboratories. However, for many food microbiology laboratories the cultural detection of salmonellae is a very mundane, routine procedure, which is accomplished economically and with minimal labour input per sample. This is not to suggest that some alternative techniques do not offer potential savings in labour, but such claims should be well studied. In the main, the major advantage of rapid methodology centres on shortening the duration of the test rather than reducing the labour per test.

253 \

Stephen Jay, Dianne Davos, Mark Dundas, Elizabeth Frankish and Diane Lightfoot Control To reduce the incidence of salmonellosis, a multifaceted approach is needed. Epidemiological surveillance can define the incidence of salmonellosis, identify vehicles and factors that are important in the spread of Salmonella, and evaluate the effectiveness of preventative measures. Education and training are important because much foodborne disease results from ignorance. People handling foods should have the knowledge needed to control the hazard of Salmonella infection. This extends from the farm, where good animal husbandry and horticultural practices have an influence on the extent of contamination, to the consumer, where epidemiological data indicate that considerable mishandling occurs. Food service and home preparation are the major places where mishandling has caused foodborne illness (206). In food processing and food service establishments' foods undergo a number of processes which have significant impacts on the incidence and on the numbers of salmonellae. The risk of foodborne illness is increased when there is a lack of awareness of the microbiological hazards associated with failure to control these processes or treatments. Sampling foods for salmonellae as a means of assessing the safety of the food is neither practical nor cost-effective (206). Sampling and testing cannot detect low levels of contamination with a high degree of confidence. Failure to detect salmonellae implies that the extent of contamination is below a level set by the sampling plan. If 30 samples from one batch of food were tested and no salmonellae were found, we would have 95% confidence that the extent of contamination was below 11.6%. This would be true provided that the organisms were homogeneously distributed and that the methodology used was able to detect the salmonellae if they were present. To have the same confidence that salmonellae would be detected in a batch with an incidence of 1%, 300 samples would need to be tested (205). For ready to eat foods, manufacturers would aim to have contamination rates several orders of magnitude below 1%, and so it is clear that end product sampling has severe limitations in assuring food safety. The HACCP approach should be applied to control salmonellae in foods (206, 347). The establishment of an effective HACCP program requires microbiological expertise (a) to analyse the Salmonella hazards associated with different foods; (b) to identify the control points critical in 254

keeping growth, death and survival of salmonellae, and contamination with salmonellae, under control; (c) to establish monitoring procedures, acceptable limits at these critical control points (CCP) and the action to be taken when the limits are exceeded; and (d) to determine the nature and frequency of verification procedures to ensure that HACCP is working correctly. Though it is not possible to outline control steps for all the variety of foods in which salmonellae should be controlled, some have been outlined for a number of different foods (206, 347). For unprocessed agricultural commodities such as vegetables, fruit and spices, contamination may be from irrigation water and other water inputs, manures, some fertilisers, or from workers or equipment used in harvesting. The application of HACCP approaches can reduce the risk of Salmonella. Chemical treatment of minimally processed fruits and vegetables, seeds and their sprouts, has been extensively investigated to reduce contamination levels of pathogens including Salmonella. Whilst some treatments are effective in reducing Salmonella numbers on sprouted seeds, elimination of the pathogen is difficult (39, 413, 414). High concentrations of chlorine are required to achieve reductions in Salmonella of 1 to 2.5 loglO cfu/g. Chemical treatment on its own cannot be relied upon to render the products safe. Irradiation of alfalfa sprouts may eliminate Salmonella (319). Salmonella outbreaks in unpasteurised orange juice and apple ciders have elicited investigations for alternative controls. Combinations of intervention treatments using heat and freezethaw cycles can result in 5 10glOreductions in S. Typhimurium DT104 numbers (388). For intensively reared animals, a number of steps can be taken to reduce the entry and spread of salmonellae. A partial list includes: the use of Salmonella free breeding stock and feed including meat and bone meal, regular cleaning of premises and particularly between herds or flocks, cleaning of watering devices and feed troughs, replacement and disinfection of used litter, restricting access of insects, rodents, wild birds and people to the animals. Some countries appear to be able to produce animals with low contamination rates, but in general salmonellae are often present in intensively raised poultry and frequently found in pigs. At the farm level, the problem of excluding salmonellae is complex and a number of factors have to be simultaneously controlled. For free range or pasture raised animals, it is more difficult to exclude salmonellae since they may be present in the soil and on pasture from previous herds, in natural surface waters,

Salmonella

incoming stock, and wild animals. Good animal husbandry practices can minimise contamination ofareas of the farm and of the animals. Salmonellae, at present, cannot be eliminated from animals on farms, so that animals will be slaughtered that are carrying these organisms on their external surfaces and in their intestinal tracts. Skinning, dehairing, or defeathering, evisceration and chilling are critical control points in the abattoir and control at these steps can reduce the extent of carcass contamination with salmonellae. There is no system of slaughter used that will entirely prevent carcase contamination, and it should be assumed that raw meats will, from time to time, contain salmonellae. However the use of carcase surface treatment with organic acid solutions is a well accepted practice to assist in the control of contamination. Other approaches to reduce the incidence of carcase contamination with Salmonella include competitive exclusion treatment used in the poultry industry and feeding animals with compounds to reduce caecal concentrations prior to slaughter (5, 266). For processed foods (e.g. gelatine, cereals, yeast, chocolate and confectionery, peanut butter, desiccated coconut, dried milk and infant formulas, and cooked meats) cross contamination is the major problem. Recontamination may be from direct or indirect contact with raw product, or from the environment. As well as the heating or pasteurisation step, control of cross contamination and of environmental contamination are critical control points. Food preparation and handling procedures in the home can be a contributory cause of salmonellosis. Some of the critical control steps in the home are: obtain food from safe sources, store perishable foods either chilled or frozen, protect shelf stable food from moisture, wash vegetables and peel fruit if these are to be eaten raw, cook foods to at least 66°C and, if the cooked food is not eaten immediately, cool it rapidly to 8°C or less in a refrigerator. Cooked foods that are stored for later consumption may be recontaminated in the kitchen. Cross contamination can be from raw food prepared and served in the same area, or from hands, utensils and surfaces transferring contamination from raw to cooked food. Dish cloths can be a reservoir and a means of disseminating salmonellae. Reconstitution of foods such as dried milk, infant formulas, and dips results in a stable food becoming one that will readily support the growth of any contaminating salmonellae.

Control of Salmonella Typhi Basic sanitary and hygiene measures recommended for diarrhoeal diseases are

applicable for preventing typhoid fever. As S. Typhi in endemic areas is transmitted by food and water, it is important to use cleanliness in food preparation and handling, provide suitable hand washing facilities for personal hygiene and to boil milk and water before consumption. Latrines have to be organised at a site distant from a source of drinking water. Chlorination of suspected public and private water supplies is recommended. Travellers in countries with risk of typhoid should avoid consumption of raw vegetables, shellfish, or peeled fruits. They should only drink capsulated or closed bottled beverages. Vaccination is highly recommended (138).

I

'I

References 1. Alford, J. A. and N. L. Knight. 1969.Applicabilityof 2.

aeration and delayed addition of selenite to the isolation of salmonellae. Appl. Microbiol. 18: 1060-1064. Alien, G., V. R Bruce, W; H., Andrews, R B. Satchell and P. Stephenson. 1991. Recovery of Salmonella from frozen shrimp: evaluation of short-term enrichment, selective media, postenrichment, and a rapid immunodiffusion method. J. Food Prot. 54: 22-27.

3.

Alien, S. B., R. Firstenberg-Eden, D. A. Shingler, C. B. Bartley and N. M. Sullivan. 1993. Evaluation of stabilized bismuth sulfite agar for detection of Salmonella in foods. J. Food Prot. 56: 666-671.

4.

Altwegg, M., R W; Hickman-Brenner and J. J. Farmer Ill. 1989. Ribosomal RNA gene restriction patterns provide increased sensitivity for typing Salmonella typhi strains. J. Infect. Dis. 160: 145-149.

5. Anderson, R. C., S. A. Buckley, T. R. Callaway, K. J. Genovese, L. R Ubena, R. B. Harvey and D.J. Nisbet. 2000. Effect of sodium Typhimurium concentrations Food Prot. 64: 255-258. 6.

7. 8.

9.

10.

11.

12. 13. 14.

chlorate on Salmonella in the weaned pig gut. J.

Andrews, W; H, P. L. Poelma and C. R. Wilson. 1981. Comparative efficiency of brilliant green, bismuth sulfite, Salmonella-Shigella, hekteon enteric, and xylose lysine desoxycholate agars for the recovery of Salmonella from foods: collaborative study. J. Assoc. Off. Anal. Chem. 64: 899-928. Andrews, W;H. 1986. Resuscitation of injured Salmonella spp. and coliforms from foods. J. Food Prot. 49: 62-75. Andrews, W; H. 1991. General referee reports: food microbiology (nondairy). J. Assoc. Off. Anal. Chem. 74: 158-162.

Andrews, W;H., C. R. Wilson and P. L. Poelma. 1983. Improved Salmonella species recovery from nonfat dry milk pre-enriched under reduced rehydration. J. Food Sci. 48: 1162-1165. Angelotti, R., M. J. Foter and K. H. Lewis. 1961. Timetemperature effects on salmonellae and staphylococci in foods. 1. Behaviour in refrigerated foods. Am. J. Publ. Hlth 51: 76-83. Anonymous. 1976. Surveillance of typhoid and paratyphoid fevers 1973. WHO Wkly Epidem. Rec. 51 (9): 69-71. Anonymous. 1977. Salmonella surveillance in 1974. WHO Wkly Epidemiol. Rec. 52 (6): 53-61. Anonymous. 1986. Changes in methods. J. Assoc. Off. Anal. Chem. 69: 379-382. Anonymous. 1990. Typhoid fever - Skagit County, Washington. Morb. Mortal. Wkly Rep. 39: 749-751.

255 ~

I' II !I I1 I1 i. !I ~