SALMONELLAE FROM FOOD SAMPLES

SMITH, W. E., HILLIER, J., AND MUDD, S. 1948 Electronmicrograph studies of two strains of pleuropneumonialike(L) organisms of human derivation. J. Bacteriol., 56,589-601.

STACEY, M. AND WEBB, M. 1947 Studies on the antibac-terial properties of the bile acids and some compoundderived from cholanic acid. Proc. Royal Soc. (London),Ser. B, 134, 523-537.

SWITZER, W. P. 1954 A suspected PPLO in Iowa swine.Iowa Vet., 25, 9-11.

WARREN, J. AND SABIN, A. B. 1942 Some biologic and im-

munologic characteristics of a pleuropneumonia-like mi-croorganism of human origin. Proc. Soc. Exptl. Biol.Med., 51, 24-26.

WINSLOW, C.-E. A. AND FALK, I. S. 1923 Studies on saltaction. IX. The additive and antagonistic effects of sodiumand calcium chlorides upon the viability of Bact. coli.J. Bacteriol., 8, 237-244.

WOGLOM, W. H. AND WARREN, J. 1938a A pyogenic filterableagent in the albino rat. J. Exptl. Med., 68, 513-528.

WOGLOM, W. H. AND WARREN, J. 1938b A pyogenic virus inthe rat. Science, 87, 370-371.

Isolation of Salmonellae from Food Samples

I. Factors Affecting the Choice of Media for the Detection and Enumeration ofSalmonella

W. I. TAYLOR, J. H. SILLIKER, AND H. P. ANDREWS

Research Laboratories, Swift & Company, Chicago, Illinois

Received for publication October 29, 1957

At a recent symposium on the problems of detectionand enumeration of Salmonella in foods (Dack, 1955),shortcomings of the methodologies in current use werediscussed by both food analysts and public health mi-crobiologists. It has become increasingly apparent tothe food microbiologist, who finds himself handicappedby inadequate methods, that the analytical problem ofsalmonellae in foods possesses certain peculiarities notcommonly encountered in the clinical laboratory. First,the organisms have often been subjected to physio-logically debilitating processes such as freezing, desic-cation, curing ingredients, and extremes of pH, heat,and osmotic pressures during the manufacture or stor-age of the product. Second, salmonellae in foods usuallycomprise an exceedingly small component of the totalmicrobial population, and in addition, are almost in-variably outnumbered by physiologically similar coli-form bacteria. Examination of the methods used for theenumeration of the salmonellae and the problems posedevoke the conclusion that the food bacteriologist hasappropriated in toto the media and methods of theclinical laboratory in an attempt to solve problems forwhich they were not designed, with little considerationbeing given to their applicabilitv. The results have beenunsatisfactory.The paucity of Salmonella prevents their enumera-

tion by direct plating methods. The most probablenumber (MPN) technique (Hoskins, 1934) is used rou-

tinely for that purpose and involves quantitativeinoculation of enrichment broth with a food samplewith subsequent identification of salmonellae on a dif-

ferential medium. Improvements in the enrichmentbroth are, therefore, of primary importance in thisschema. Selenite F enrichment broth has been estab-lished as the medium of choice over tetrathionate broth.This does not preclude the possibility that there arechanges in formulae which would enhance the abilityof selenite F to detect salmonellae. Similarly, evaluationof the abilities of differential and selective media tofacilitate recovery of Salmonella sp., specifically, com-pletes the selection of the tools with which to continuethe examination of changes in methodology. The au-thors intend to present in this and subsequent reportsthe results of some of the investigations which examinefactors affecting the choice of media, and to discusschanges in methodology which have proved efficaciousin detection of Salmonella sp. and their enumerationfrom foods.

MATERIALS AND METHODS

Selenite F medium (Leifson, 1936) modified by theaddition of cystine (North and Bartram, 1953) was usedas the control for experiments conducted on enrichmentmedia. Dulcitol or mannitol were substituted for lactosein the appropriate formulae tested.

Selenite brilliant green sulfapyridine medium (SBS)was prepared according to directions supplied by Dr.Stokes; when it became available commercially, thedehydrated medium was used (Osborne and Stokes,1955).Metabolite solutions ("metsol") (Heinmets et al.,

1954) consisted of the following metabolites in 0.2 per

1958] 189

on May 6, 2018 by guest

http://aem.asm

.org/D

ownloaded from

W. I. TAYLOR, J. H. SILLIKER, AND H. P. ANDREWS

cent concentration: sodium pyruvate, oxalacetic acid,

sodium acetate, sodium citrate, cis-trans aconitic acid,isocitric acid, lactic acid, malic acid, and sodium fuma-rate. These compounds were dissolved in 0.1 M, pH 7.0

phosphate buffer (KH2PO4/Na2HPO4).The inocula varied for the different phases of these

studies. In each case, however, dried egg albumenknown to contain salmonellae was used. These inoculaalways contained coliform organisms as well. In com-

parisons of this type, the value of using natural inoculacannot be overemphasized since such materials containthe organisms sought in their qualitative and quanti-tative relationships to the natural environment. Fur-thermore, such organisms have not been modified byprior exposure to artificial media. In some experiments,it was necessary to use pure cultures of either Escher-ichia coli or Salmonella sp. Multiple strains of newlyisolated organisms were employed in order to recreatethe inevitable mixed flora characteristic of naturallycontaminated foods.

Presumptive evidence of the presence of Salmonellasp. resulted from the appearance of red translucentcolonies on brilliant green agar plates. Three such colo-nies from each plate were subjected to biochemical andserological studies. Organisms which did not fermentlactose or sucrose but did ferment glucose and mannitol,and which were indol and urease negative, were typedserologically with polyvalent and group antisera. Ifthey agglutinated in both, they were considered to beSalmonella sp., and frequent spot checks of these posi-tives by Illinois State Public Health Laboratories al-ways confirmed their identity as serotypes of Sal-monella.

Plating media and methods employed in the quanti-tative detection of salmonellae were as follows: tryptoneglucose extract agar (TGE) and desoxycholate agar

(Des) were poured plates; brilliant green agar (BG),Levine's eosin methylene blue agar (EMB), MacCon-

key's agar (MacC), Salmonella-Shigella agar (SS), and

bismuth sulfite agar (BiS) were inoculated with a glassspreader bar or "hockey stick." All of the plating mediawere dehydrated media made in accordance with the

manufacturer's directions.TGE was chosen as the medium most likely to give

the maximum counts of both salmonellae and coliforms

since it contained no inhibitors. Each of the other mediawas conmpared with it for the enumeration of these or-

ganisms and statistical analysis was used to determinethe significance of the differences observed.

Since the survey of the quantitative abilities of mediaoffered the opportunity for comparisons of direct plat-ing methods with MPN determinations, the latter were

investigated. MPN's for salmonellae were obtained

using cystine-selenite broth (Sel) and mannitol broth

(Mann); MPN for coliform organisms compared lac-

tose broth (LB) with brilliant green bile, 2 per cent,

lactose broth (BGBLB). Verification of selenite brothpositives was made on BG agar; gas formation in theother MPN determinations was accepted without con-firmation since these were pure culture studies.

RESULTS AND DIscussIoNThe results of substituting dulcitol or mannitol for

lactose in selenite F broth may be seen in table 1. Therationale for these changes was simply that both ofthese carbohydrates are utilized by the salmonellaewhile lactose is not. It is logical to assume that a me-dium designed to enhance the growth of one group oforganisms over another would contain nutritive sub-stances utilizable only by the preferred group. Leifson(1936) claimed that he found no improvement whenusing carbohydrates other than lactose, and that thepurpose of the carbohydrate is primarily that of negat-ing overgrowth of coliforms or enterococci by affordinga source of acid to prevenit the pH of the medium fromrising during the growth of the organisms and, conse-quently, the loss of selenite toxicity associated with analkaline pH.

Six replicate three tube MPN's with mannitol sele-nite vs selenite F revealed no statistically significantdifferences in isolations of Salmonella sp. between thesemedia.

Dulcitol selenite was inoculated with a naturallycontaminated food sample (dried egg albumen) ofknown content of Salmonella. Two series of 50 replicateseach failed to produce a significantly greater number ofpositive isolations than selenite F produced.

Prior to the publication of data on selenite brilliantgreen sulfapyridine medium, we were privileged to re-

TABLE 1

Comparisons of Isolations of Salmonella from selenite broths withlactose, mannitol, or dulcitol

StatisticalAnalysis

Media No. Trials Positive squre_er_enChial Proba-Ci bilitysquare per cent

Selenite F ................. 6 43/75 1.708 19Mannitol selenite .......... (MPN's) 34/75Selenite F ................. 2 39/100 0.513 47Dulcitol selenite .....

45/100Selenite F ................. 5 38/66 1.486 22SBS* ..................... (MPN's) 30/66Selenite F ................. 2 27/100 22.808 <1

SBSt... ... 0/100Selenite F ................. 5 37/60 1.216 27Metsol pretreatedt.. ...........(MPN 's) 30/60Mannitol selenite .......... 2 17/21 3.733 5Metsol pretreatedt. (MPN's) 10/21

* Selenite brilliant green sulfapyridine made in laboratory.t Selenite brilliant green sulfapyridine commercially pre-

pared dehydrated medium.+ Metabolite solutions-see text.

190 [VOL. 6

on May 6, 2018 by guest

http://aem.asm

.org/D

ownloaded from

SALMONELLAE FROM FOOD SAMPLES

ceive instructions from one of the authors for the prep-

aration of SBS broth. The results of five replicateMPN's on food samples contaminated with Salmonella,shown in table 1, were compared to those obtained withselenite F and showed no statistically significant ad-vantage over the latter. After a lapse of some months,an early lot of commercially prepared dehydrated SBSmedium was made available to us and experimentswith food samples of known content of Salmonellaproved the medium to be more disappointing than thealiquots we had made previously. Statistical analysisconfirmed the marked superiority of selenite F over theSBS medium observed in these trials.

Heinmets et al. (1954) had investigated rejuvenationof coliforms which had undergone debilitating exposures

to deleterious agents, physical and chemical, and foundthat they responded to various metabolic intermediatesof the tricarboxylic acid cycle. This could conceivablybe of advantage in detection of Salmonella from mixedflora in foods which had been subjected to the varietyof manufacturing processes designed to prevent or re-

duce bacterial contamination.Most probable number determinations of populations

of Salmonella on five naturally contaminated foodsamples were performed using metabolic solutions (met-sol) for pretreatment of the sample prior to enrichmentin selenite broth.

Similar trials were conducted with mannitol selenitebroth. The results shown in table 1 reveal that no sta-tistically significant enhancement of detection of Sal-monella accrued to metsol pretreatment. The picturepresented by the plates obtained from the enrichmentbroth was one of a heavy coliform overgrowth.

There is a large array of plating media now availableto the microbiologist, however, few, if any, are specifi-cally designed for quantitative analysis of salmonellaein foods. A representative selection of these media hasbeen examined for the ability to enumerate Salmonellaand coliform accurately to establish the plating mediumof choice for these organisms.The inocula for these experiments consisted of pure

cultures of Salmonella or coliform organisms newlyisolated from natural sources, grown in brain heart in-fusion broth for 18 hr, and pooled in 6 different groupsof 3 serotypes or strains. A total of 18 each of Salmonellaand coliforms were used in these experiments.The counts of Salmonella obtained from different

media and their significance are shown in table 2. Therewere no significant differences between the counts ofSalmonella observed on BG, Sel, BiS, EMB, and Des,and those observed on TGE.The counts obtained on MacC, SS, and Mann were

consistently lower than those obtained on TGE and,although the probabilities of 16 to 18 per cent are largerthan the 5 per cent normally cited in conventionalstatements of statistical significance, we believe they

constitute adequate evidence that the media are in factdifferent. With regard to these three media, the lowMPN's with mannitol broth as compared to selenitecomes as a surprise, since one assumes that a mediumcontaining inhibitors such as selenite would inhibitgrowth of some salmonellae and, therefore, producelower MPN's than a noninhibitory sugar broth.The lower counts obtained on SS plates are in sharp

contrast to the excellent showing of both BG and BiSagars, yet all are selective media. Results with the lattertwo give ample evidence that it is possible to obtaingood quantitative recovery of one group of organisms,the salmonellae, at the expense of another, the coli-forms.The coliform counts shown in table 3 reveal that

there were significant differences in the abilities ofseveral of these media to permit uninhibited growth of

TABLE 2Mean counts of Salmonella obtained on different media

Comparison ofSalmonella Counts values Observed

in Millions with TryptoneGlucose Ex-Media tract Agar

Mean of 95 per cent Per Proba-6 replica- confidence ent bility

tions interval per cent

Tryptone glucose extract agar. 738 506-1080 -

Brilliant green agar........... 927 635-1350 125.6 40Cystine-selenite broth. .. ... 711 488-1040 96.3 > 50Bismuth sulfite agar* ......... 693 475-1010 93.9 > 50Levine's eosin methylene blue

agar ....................... 652 447- 951 88.3 >50Desoxycholate agar........... 601 412- 877 81.4 48MacConkey's agar............ 511 350- 745 69.2 18Mannitol broth ............... 505 346- 732 68.4 16Salmonella-Shigella agar.........498 341- 726 67.5 16

*!Counts after 48 hr incubation.

TABLE 3Mean coliform counts obtained on different media

Comparison ofColiform Counts in Values Observed

Millions with TryptoneGlucose ExtractMedia Agar

Mean of 95 per cent Per Proba-6 repli- confidence cent bilitycations interval per cent

Tryptone glucose extract agar. 1850 413-8330 -

Levine's eosin methylene blueagar ..................... 1860 414-8340 100.5 >50

Desoxycholate agar ........... 1810 403-8110 97.8 > 50Lactose broth ................ 1570 350-7050 84.9 > 50MacConkey's agar ............ 1530 340-6860 82.7 > 50Brilliant green bile lactosebroth ....................... 1420 317-6380 76.8 >50

Brilliant green agar........... 215 48- 966 11.6 5Salmonella-Shigella agar .. 110 24- 494 5.9 2Bismuth sulfite agar*......... 91 20- 409 4.9 <1

* Counts after 48 hr incubation

1958] 191

on May 6, 2018 by guest

http://aem.asm

.org/D

ownloaded from

W. I. TAYLOR, J. H. SILLIKER, AND H. P. ANDREWS

these organisms. Counts obtained with BiS, SS, and BGagars were significantly lower in numbers detected thanthe other media.The information gained from these two tables is of

interest to those who must isolate salmonellae fromproducts in which the coliforms are the predominantgram negative flora, and who must quantify those re-sults. The inhibition of salmonellae by coliforms is anever-present factor (Levine and Tanimoto, 1954), andthe choice of a medium which is inimical to the growthof coliforms and permissive of unrestricted growth ofSalmonella is desirable. In addition, distinct differen-tiation between coliform and noncoliform, and an in-cubation period of not longer than 24 hr, are consideredsalutary characteristics. These criteria have resulted inestablishing BG agar as the medium of choice. Thisconclusion was arrived at empirically after years ofusing BiS and other media. It was found that the differ-ences between Salmonella and other enteric organismswere not as marked on SS or BiS as on BG. In addition,48 hr incubation was almost always necessary for thedevelopment of the salmonellae on BiS agar, at whichtime nonsalmonellae also appeared, more often thannot, with no obvious colonial characteristics to distin-guish them.

It is not the intent of the authors to imply that onemedium is completely adequate for the isolation of allsalmonellae, as in our experience and that of manyothers (Neter and Clark, 1944; Byrne et al., 1955; andShaughnessy and Jensen in Dack, 1955), it is incontro-vertibly shown that more than one enrichment brothand a larger number of plating media are necessary toobtain a maximum number of positive isolations. Thelimiting factors of numbers of samples and personnelavailable very often determine the practical applicationof this knowledge, however, and cystine selenite (SEL)followed by BG have most consistently outperformedany other combination with which we have had ex-perience. Other investigators have found BG superior(Broh-Kahn, 1946; Ayres, 1953), and others have foundBiS better (Hajna and Perry, 1938; Gunther and Taft,1939; Byrne, in Dack, 1955). With the growing realiza-tion of the shortcomings in detection of salmonellae infoods and the magnitude of the difference between theproblems posed by foods and those of the clinical labo-ratory specimens, additional workers have recentlybegun to devise methods and media which are adaptedto the specific needs of the food analyst (Ayres, 1949;Byrne et al., 1955; Osborne and Stokes, 1955).

SUMMARY

Substitution of the carbohydrates mannitol or dulci-tol for lactose in cystine selenite F enrichment brotheffected no increase in the numbers of Salmonella iso-lated from naturally contaminated dried egg albumen.

broth did not produce more isolations of Salmonella,and in some instances, was markedly inferior to Sele-nite F.The use of tricarboxylic acid cycle metabolites to

pretreat the physiologically damaged salmonellae didnot enhance their isolation.Enumeration of salmonellae on nine media disclosed

that counts on SS and MacConkey's agars and MPN'son mannitol broth were significantly lower than thoseobtained from brilliant green, bismuth sulfite, eosinmethylene blue, and desoxycholate agars or cystineselenite broth, these latter being quite comparable totryptone glucose extract agar counts.

Coliforms were greatly inhibited by brilliant green,SS, and bismuth sulfite agars as compared to tryptoneglucose extract agar counts. Counts on eosin methyleneblue, desoxycholate, and MacConkey's agars, andMPN's obtained with brilliant green bile 2 per cent andlactose broth were comparable with counts on tryptoneglucose extract agar.The combination of cystine selenite F enrichment

broth and subsequent streaking on brilliant green agarhas proved itself valuable through long experience inthe authors' laboratories. The combination has beendemonstrated in these experiments to be the method ofchoice for the detection and enumeration of Salmonellaorganisms in food analysis.

REFERENCES

AYRES, J. C. 1949 A procedure for the quantitative estima-tion of Salmonella in dried egg products. Food Technol.,3, 172-176.

AYRES, J. C. 1953 MIethodology for isolating Salmonellafrom dried egg products. Iowa State Coll. J. Sci., 27,479-489.

BROH-KAHN, R. H. 1946 The laboratory diagnosis of entericinfections caused bv the Salmonella-Shigella gI-oup. Mili-tary Surgeon, 99, 770-776.

BYRNE, A. F., RAYMAN, M. M., AND SCHNEIDER, M. D. 1955Methods for the detection and estimation of numbers ofSalmonella in dried eggs and other food products. Appl.Microbiol., 3, 368-372.

DACK, G. M. 1955 Symposium on methodology for Sal-monella in food. Bacterial. Revs. 19, 275-276.

GUNTHER, C. B. AND TAFT, L. 1939 A comparative study ofmedia employed in the isolation of typhoid bacilli fromfeces and urines. J. Lab. Clin. Med., 24, 461-471.

HAJNA, A. A. AND PERRY, C. A. 1938 A comparative studyof selective media for the isolation of typhoid bacilli fromstool specimens. J. Lab. Clin. Med., 23, 1185-1193.

HEINMETS, F., TAYLOR, W. W., LEHMAN, J. J. 1954 The useof metabolites in the restoration of the viability of heatand chemically inactivated Escherichia coli. J. Bacteriol.,67, 5-12.

HOSKINS, J. K. 1934 Most probable numbers for evaluationof coli-aerogenes tests by fermentation tube method. Pub-lic Health Repts., 49, 393-405.

LEIFSON, E. 1936 New selenite enrichment media for theisolation of typhoid and paratyphoid (Salmonella) bacilli.Am. J. Hyg., 24, 423-432.

LEVINE, M. AND TANIMOTO, R. H. 1954 Antagonisms among

192 [VOL. 6

Selenite brilliant green sulfapyridine enrichment

on May 6, 2018 by guest

http://aem.asm

.org/D

ownloaded from

PAPER DISCS FOR HYDROGEN SULFIDE TEST

enteric pathogens and coliform bacteria. J. Bacteriol.,67, 537-541.

NETER, E. R. AND CLARK, D. 1944 The effectiveness ofdifferent culture media in the isolation of enteric micro-organisms. Am. J. Digest. Diseases, 11, 229-233.

NORTH, W. R. AND BARTRAM, M. T. 1953 The efficiency of

selenite broth of different compositions in the isolation ofSalmonella. Appl. Microbiol., 1, 130-134.

OSBORNE, W. W. AND STOKES, J. L. 1955 A modified selenitebrilliant-green medium for the isolation of Salmonellafrom egg products. Appl. Microbiol., 3, 295-299.

Detection of Hydrogen Sulfide Production by Bacteria using Paper

Disc Methods12

T. M. COOK3 AND M. J. PELCZAR, JR.

Department of Microbiology, University of Maryland, College Park, Maryland

Received for publication October 30, 1957

There has been an increasing interest in rapid andmore expedient methods for the identification of bac-teria. One of the several approaches to this subject hasbeen the use of impregnated paper discs for the perfor-mance of routine biochemical determinations (Soto,1949; Knox, 1949; Snyder et al., 1951). Two reportshave described paper disc techniques for the detectionof hydrogen sulfide production by Enterobacteriaceae(Snyder, 1954; Sanders et al., 1957). Because sufficientquantitative information regarding disc methods wasnot available, further study to confirm and extend thework of previous investigators was indicated. This re-port presents the results of such studies of paper discmethods for hydrogen sulfide tests.

MATERIALS AND METHODS

Cultures. In this study a number of cultures, mainlyEnterobacteriaceae, were employed to evaluate the ex-perimental discs. Many of the Salmonella species werekindly supplied by Lt. Col. A. C. Sanders of the WalterReed Army Institute of Research. An attempt was

made to include representatives of each serologicalgroup of the genus Salmonella.

Preparation of discs. The paper discs (no. 740-E)4 usedin this study were 12.7 mm in diameter by about 1 mmthick. These were impregnated with various reagentsfor hydrogen sulfide tests. Experimental discs contain-ing sodium thiosulfate, ferric ammonium citrate, andglucose were prepared by aseptically pipetting Seitz-filtered solutions onto sterile dry discs contained in

1 From a thesis submitted to the Graduate School of theUniversity of Maryland in partial fulfillment of the require-ments for the degree of Master of Science.

2 This investigation was supported in part by a researchgrant from the Baltimore Biological Laboratory, Inc., Balti-more, Maryland.

3 Present address: Merck, Sharp & Dohme Research Lab-oratories, Rahway, New Jersey.

4Schleicher & Schuell Co., Keene, New Hampshire.

Petri dishes. The discs were satisfactory for immediateuse or could be used after drying. Discs containing so-dium thiosulfate and ferric ammonium citrate but noglucose were impregnated by touching to the surface ofthe solution with forceps and allowing them to becomesaturated by capillary action. After drying on alumi-num pans overnight at 37 C, the discs were placed inloosely capped vials and sterilized with dry heat at 140C for 3 hr. All solutions for impregnating discs wereadjusted to contain the desired amount per disc in 0.08ml, the absorptive capacity of a disc.To provide a simple experimental system, discs were

charged with substrate and indicator, only. Discs con-taining varying amounts of a substrate, such as sodiumthiosulfate or cysteine, and an indicator, such as ferricammonium citrate or lead acetate, were prepared andtested on plates and tubes inoculated with various testorganisms.For comparison, some discs were prepared with com-

plete Bacto-peptone iron agar (PIA)5 formula (minusagar) at 10 X normal strength according to Snyder'sdirections (Snyder, 1954) except that the Bacto-pro-teose peptone no. 3 of the PIA was replaced by Thio-tone.6

Discs containing 20 mg each of carbohydrate for usein the procedure of Sanders et al. (1957) were preparedand sterilized after drying in loosely capped bottles byexposure to ethylene oxide gas for 1 hr.Media. The following dehydrated commercial media

were prepared according to the manufacturers' direc-tions and employed in plates and tubes to test discs:phenol red agar base, phenol red broth base, trypticaseagar base, trypticase soy agar-BBL, and Bacto-heartinfusion broth. In addition, triple sugar iron agar (BBL)

5 Difco Laboratories, Inc., Detroit, Michigan.6 Baltimore Biological Laboratory, Inc., Baltimore, Mary-

land.

19581 193

on May 6, 2018 by guest

http://aem.asm

.org/D

ownloaded from