J. clin. Path., 1977, 30, 495-504

Factors affecting the value of a simple biochemicalscheme for identifying Enterobacteriaceae: thereproducible recognition of biotypesJ. G. BARR, REBECCA J. MAHOOD, AND K. P. W. CURRY

From the Department of Clinical Bacteriology, Royal Victoria Hospital, Belfast, Northern Ireland

SUMMARY A biochemical screening scheme on agar media for differentiating Enterobacteriaceae ina hospital laboratory is evaluated. Careful observation of test reactions within the scheme permittedthe recognition of 78 biotypic reaction patterns which could contribute to epidemiological surveil-lance. The limitations of the technique are described and discussed and methods of importance inensuring the reproducibility of reaction patterns emphasised.

Many workers have described biochemical schemesfor identifying Enterobacteriaceae (Carpenter et al.,1966; Rosner, 1970; Chadwick et al., 1974; Lassen,1975) and several commercially available systemshave been evaluated (Tomfohrde et al., 1973; Nordet al., 1975).The scheme described here is based on bio-

chemical tests which have been shown over a longperiod to give reproducible and clear results. Thesmall and specific number of tests chosen was basedon our experience of the relative occurrence of generaand species and designed to allow the simplestidentification scheme for the Enterobacteriaceaecommonly isolated in this laboratory. Otherevidence shows that this distribution of isolates isnot uncommon in other routine medical laboratories(Rosner, 1970; Berger and Einecke, 1973; Chadwicket al., 1974).The tests adopted included many of those con-

sidered by Chadwick et al. (1974) to be valuable inthe definition of the major taxa proposed by Edwardsand Ewing (1962). The value of aesculin hydrolysis(Wasilauskas, 1971; Barr and Mahood, 1977) andinositol fermentation (Donovan, 1966; Barr andMahood, 1976) in differentiating the Klebsiella-Enterobacter-Serratia group from other Entero-bacteriaceae has already been established, andProteus spp and Providencia spp are differentiatedfrom other Enterobacteriaceae and from one anotheron the basis of phenylalanine deaminase and ureaseactivity. Other tests, including tryptophanase activity,growth in presence of potassium cyanide, citrate

Received for publication 29 November 1976

utilisation, and colony colour and hydrogen sulphideproduction on xylose-lysine deoxycholate medium(XLD medium; Taylor, 1969), contributed to thedifferentiation of Citrobacter spp and Escherichiacoli and aided differentiation among other genera.Ornithine decarboxylase activity differentiatesEntero-bacter spp and Serratia spp from other Klebsielleaeand aids differentiation between Proteus spp. Thesetests have allowed a differentiation of the vastmajority of clinical isolates of Enterobacteriaceaeinto the major taxa described by Edwards andEwing (1962).

It should be noted that variations in either therate or degree of specific enzyme action may allowdifferentiation between genera or between species ina standardised system. Thus the degree of aesculinhydrolysis and inositol fermentation may indicatespecies or biotype differences among Klebsielleae(Barr and Mahood, 1976).

This system, by highlighting unusual biotypicreaction patterns of individual genera or species,may often allow recognition of patterns of bio-chemical results which may be of value in epidemio-logical studies. This may be recognised in theincidence of tryptophanase-positive isolates ofEnterobacter spp, Klebsiella aerogenes, and Proteusmirabilis and by variations in production ofhydrogensulphide by P mirabilis and Citrobacter spp on XLDand inositol-hydrogen sulphide-motility media.

In these laboratories we have tried to demonstratethe importance of inoculum size, incubation time,and reproducibility in the semi-quantitative inter-pretation of individual biochemical results onthe performance of the scheme. It should not be

495

copyright. on F

ebruary 16, 2020 by guest. Protected by

http://jcp.bmj.com

/J C

lin Pathol: first published as 10.1136/jcp.30.6.495 on 1 June 1977. D

ownloaded from

J. G. Barr, Rebecca J. Mahood, and K. P. W. Curry

assumed that strictly positive or strictly negativebiochemical results are the only viable and repro-ducible alternatives when standardised methods areadopted.

Material and methods

ORGANISMS

Clinical isolatesA total of 1659 clinical isolates were biotyped by thescheme described over a period of eight months.These comprised 1197 cultures derived from urinarytract infections and a further 462 isolations derivedfrom other sources in the routine bacteriologicallaboratory.

Stock culturesThe 30 stock cultures listed below were supplementedwith 25 cultures of Enterobacteriaceae previouslyisolated from clinical material, stored, and theiridentification authenticated by extensive biochemicaltesting and by other workers. The National Collec-tion of Type Cultures (NCTC) cultures examinedwere Klebsiella aerogenes (2 strains) NCTC 8172 and9668; K altantae NCTC 9496; K edwards&i NCTC5054; K pneumoniae (4 strains) NCTC 9633, 204,8532, and 10246; K ozaenae (4 strains) NCTC 5050,9659, 10313, and 8883; K rhinoscleromatis (2 strains)NCTC 5046 and 5047; Enterobacter aerogenesNCTC 10006; E cloacae NCTC 10005; Serratiamarcescens NCTC 9940; Escherichia coli (2 strains)NCTC 10418 and 9001; Proteus mirabilis NCTC5887; P morganiiNCTC 235; P vulgaris NCTC 4175;P rettgeri NCTC 7475; Providencia B subgroup BNCTC 10318; Providencia spp NCTC 1026; Citro-bacter freundii (2 strains) NCTC 9750 and 8090;Yersinia enterocolitica NCTC 10460; and Pseudo-monas aeruginosa NCTC 10332 and 10701.

CULTURE MEDIAAesculin bile agar (Difco), Simmon's citrate agar(Difco), phenylalanine agar (Difco), and xyloselysine deoxycholate agar (Difco) were utilised asdescribed by Chadwick et al. (1974).

Ornithine decarboxylase agar, derived from thefluid medium of Fay and Barry (1972), was preparedin 1-5% agar (Oxoid LI 1). The inositol-hydrogensulphide-motility medium and potassium cyanidemedium were prepared as described by Barr andMahood (1977). Urease medium was that proposedby Christensen (1946) with the modificationsdescribed by Cowan (1974).

Replidishes (Sterilin Ltd) were used throughout,3 ml of each medium, except inositol-hydrogen

sulphide-motility medium (4 ml) being dispensed ineach compartment.For standard cultures and 50 clinical isolates

biochemical tests were also carried out by classicalmethods as described by Cowan (1974).

Bacterial isolates not identified on screening mediawere identified after further tests-namely, arabinosefermentation (1 4 peptone water); lysine decarboxy-lase activity (Decarboxylase base (Difco) accordingto Moller, 1955), and methyl-red and Voges-Proskauer tests (Cowan, 1974, using the methoddescribed by Barritt, 1936).The quality of all screening media was controlled

by regular testing with standard cultures of knownreaction.

PREPARATION OF INOCULAInocula for screening media were prepared by trans-ferring at least one colony from a primary isolationplate into peptone water and incubating at least4 hours at 37°C to obtain a culture of optical density0-2-0-5 at 700 ,um as measured by a Hilger spectro-photometer. A practised eye can quickly judge aturbid culture to be within or significantly below thisrange of optical density.

Sequential tenfold dilutions of this inoculum wereprepared in 9 ml distilled water and estimates oftotal viable cells calculated by the method of Milesand Misra (1938).

Inositol-hydrogen sulphide-motility medium wasinoculated using a sterile straight platinum wire.Other media were inoculated with one drop, about0-02 ml, of peptone water culture or a deriveddilution from a sterile Pasteur pipette.

CONDITIONS OF INCUBATIONThe peptone water culture used for inoculation ofagar media was reincubated for a further 16 hoursand used for detecting tryptophanase activity.

All media were incubated at 37°C for 16 hours withlids of replidishes placed in the vented position.Before incubation ornithine decarboxylase agar wasoverlaid with sterile liquid paraffin (0-5 ml percompartment). Urease agar plates were sealed withone-inch wide adhesive cellophane tape strips(Sellotape) and placed in parallel in such a manner

that two strips overlaid each compartment junction.Contact between sealing tape and replidish was

obtained by gentle finger pressure, preventingdiffusion of volatile metabolites from one compart-ment to another.

RECORDING BIOCHEMICAL REACTIONSCitrate utilisation was recorded by indicator colourchange or growth, or both; ornithine decarboxylase

496

copyright. on F

ebruary 16, 2020 by guest. Protected by

http://jcp.bmj.com

/J C

lin Pathol: first published as 10.1136/jcp.30.6.495 on 1 June 1977. D

ownloaded from

Factors affecting the value of a simple biochemical scheme for identifying Enterobacteriaceae

activity by indicator colour change; resistance topotassium cyanide by presence or absence of growth;and indole production detected by use of Kovac'sreagent. Colony colour on XLD medium wasrecorded and the presence or absence of mediumblackening due to hydrogen sulphide produictionnoted. In inositol-hydrogen sulphide-motility mediumhydrogen sulphide production was recorded bysulphide precipitation after 16 hours. Motility, madevisible by reduction of triphenyltetrazolium chlorideto formazan, was recorded after 16 hours.

Urease activity was recorded according to indicatorcolour change as strong (3 +), characteristic ofProteus spp; moderate (2+), characteristic ofKlebsiella spp; weak (1+), characteristic of Entero-bacter spp and Yersinia enterocolitica; and absent(-)-

Aesculin hydrolysis and inositol fermentationwere recorded in a semi-quantitative manner aspreviously described (Barr and Mahood, 1976, 1977).The degree of aesculin hydrolysis was recorded asstrong (2 +), complete colour change within inocula-tion compartment; weak (1+), hydrolysis in theimmediate area of bacterial growth; and (-) nohydrolysis. Inositol fermentation was recorded as(3 +), complete indicator colour change withininoculation compartment; (2+), distinct zone ofindicator colour change along line of stab inocula-tion, and (-) in the absence of fermentation.,To ascertain the effect of inoculum size on test

results on screening media the response of tenfoldserial dilutions of bacteria representative of themajor genera and species were evaluated by themethods described above.

IDENTIFICATION OF CLINICAL ISOLATESBacteria were identified as far as possible on thebasis of results obtained on screening media.However, in an exploratory study with culturecollection bacteria and 50 clinical isolates of Entero-bacteriaceae biochemical tests were also carried outin parallel in classical fluid media. When these testsfailed to give species identification (orwith Citrobactergenus definition) bacteria were further tested andidentified according to the methods described byCowan (1974).

In the subsequent study isolates which onscreening media alone failed to allow speciesidentification were further tested by methyl-redVoges-Proskauer, arabinose fermentation, and lysinedecarboxylase tests. Routinely no attempt was madeby further biochemical testing to differentiatebetween Klebsiella aerogenes, K edwardsii, andK atlantae; to delineate species among Citrobacterisolates; or to give species definition to atypicalProteus isolates.

Results

COMPARISON OF BIOCHEMICAL TESTIN{1 BYSCREENING MEDIA AND BY CLASSICALMETHODSThe biochemical reactions of stock cultures and 50clinical isolates were compared on screening mediaand classical test media.

Reactions obtained for ornithine decarboxylaseactivity, inositol fermentation, and aesculin hydro-lysis were identical by both methods. It was possiblealso, as in the screening media, to distinguish betweenthe aesculin reaction of Pseudomonas spp andKlebsielleae on agar slopes. Similarly those bacteriashowing late fermentation (2-6 days) in inositolpeptone water corresponded with those showingonly trace fermentation along the line of stabinoculation on the screening medium.

Identical results were also derived from phenyl-alanine deaminase and urease activity, although itwas impossible on slope media to distinguish betweenthe intensity of Proteus spp and Klebsiella spp ureaseactivity. Contrary to the observations of Chadwicket al. (1974), phenylalanine deaminase activity onthe screening medium was demonstrated for allisolates of Proteus morganii.Some variation in results obtained in 16-hour

incubations were, however, observed with potassiumcyanide (KCN) and Simmon's citrate media withstock cultures. Errors in determination of KCNresistance were often observed in our use of KCNbroth (Rogers and Taylor, 1961) and we had muchless difficulty in observing growth on an agar base.With Simmons's citrate medium the divergence inresults was due to difficulties in discerning, after16 hours' incubation, growth or indicator colourchange in citrate agar distributed in Bijou bottles.Those cultures which, in conflict with screeningmedium results, were recorded as negative after16 hours' incubation were always recorded as positiveafter 24-72 hours. No positive citrate result wasencountered in Bijou bottles which was not accom-panied by a similar result on the screening medium.

In summary, we may say that divergence of resultsresided mainly in the easier observation of reactionsin media distributed in replidishes (Sterilin).Repeated screening of stock cultures demonstrated

the reproducibility of test medium results andtherefore the biotype reaction patterns recorded.Reproducibility was further established whenmultiple isolates from one source were shown to givethe same biotypic reaction pattern.

EFFECT OF INOCULUM SIZE ON RESPONSEPRODUCED ON SPECIFIC SCREENING MEDIAThe primary aim of this investigation was to establish

497

copyright. on F

ebruary 16, 2020 by guest. Protected by

http://jcp.bmj.com

/J C

lin Pathol: first published as 10.1136/jcp.30.6.495 on 1 June 1977. D

ownloaded from

498

conditions under which reproducible reaction patterns

for bacterial isolates would be obtained fromrepeated isolations of the same bacterium or fromrepeated biochemical testing of the same isolate.

Conditions of medium preparation, mediumdispensing, and test incubation conditions were

readily standardised. However, since it was more

difficult to standardise inoculum size, the effect ofvariations in inoculum size on specific test resultsduring 16-hour incubations was evaluated. Each testmedium was evaluated in turn in the presence ofdilutions of organisms which gave known responses

on those media (Table 1).A number of screening media, XLD agar,

phenylalanine deaminase agar, and peptone water-used for detecting tryptophanase activity-are notincluded. In all cases these 3 tests gave the same

result irrespective of inoculum size. Other media,clearly influenced by inoculum size, are consideredin turn.The rate of aesculin hydrolysis by Klebsielleae

has already been shown to be more rapid than thatof non-Klebsielleae (Barr and Mahood, 1977). Theinvestigations described here show that while largevariations in inoculum size have little effect on thereaction demonstrated by Klebsielleae reduction ininoculum size of aesculin hydrolysing isolates ofE coli may result in the recording of erroneous

negative results after 16-hour incubations. Nodifferentiation between aesculin hydrolysis of Kaerogenes and E aerogenes was recorded.The detection of citrate utilisation as a sole

carbon source was, as described earlier, easier inscreening media than by conventional test methods.Only with Proteus mirabilis was the effect of inoculumsize considered to be significant in screening media.As shown in Table 1, sequential reductions ininoculum size resulted in increasingly reduceddetection of citrate utilisation in 16-hour incubations.This observation should not result in difficulties indifferentiation between Proteus vulgaris and Pmirabilis but may certainly be an important factor inthe differentiation of some intermediate biotypes ofP vulgaris and P mirabilis.

In the scheme adopted here considerable depen-dence has been placed on accurate demonstration ofornithine decarboxylase activity in differentiatingKlebsiella spp from Enterobacter spp and Serratia sppand in species differentiation among Proteus isolates.It is clear from the results given in Table 1 that a

reduction in inoculum size of Enterobacter aerogenesfrom 108 to 106 viable cells/ml would significantlyreduce the detection of this enzyme and could rendersome isolates indistinguishable from Klebsiella spp.While detection of ornithine decarboxylase activityby other isolates is unlikely, for the most part, to

J. G. Barr, Rebecca J. Mahood, and K. P. W. CurrY

influence species identification, variation in inoculumsize over a narrow range would play a considerablepart in differentiating biotypes of E coli and Pro teas-morganii.

This limited investigation showed that for mostgenera reduction in inoculum size over a consider-able range did not influence the recording of sensi-tivity or resistance to potassium cyanide. The mostimportant differentiation of Citrobacter spp andE coli was not influenced by inoculum size withinnormal ranges. The variation due to inoculum sizeamong the Klebsielleae is unlikely in the screeningregimen employed to influence Klebsielleae recog-nition, although possible error on recognition ofKpneumoniae is noted in the sensitivity ofK aerogeniesto KCN when low inocula were used.

Motility was not influenced by inoculum sizeexcept among isolates ofE coli. Although non-motileisolates of E coli are not infrequent their pre-dominance among those screened in this study(Table 2) is more likely to be due to slight differencesin inoculum size among those screened than to apreponderance of non-motile isolates. Although wehave not checked this observation the results shownin Table 1 demonstrate that recognition of motilityin the screening media by E coli is more sensitive toinoculum size than other genera investigated. Inview of the sensitivity to inoculum size of ornithinedecarboxylase activity among Enterobacter isolatesit is of value to note here that motility of Entero-bacter aerogenes was normally recorded in thismedium and should allow adequate differentiationfrom Klebsiella spp in the screening scheme.Among Citrobacter spp and Proteus spp examined

hydrogen sulphide production by Proteus morganialone was shown to be sensitive to inoculum size(Table 1). Since biotypes have been encounteredamong Citrobacter spp and Proteus spp (Table 2)which failed to produce hydrogen sulphide under theexperimental conditions employed, the insensitivityof isolates reported here to reduction in inoculumsize is considered to be valuable in the distinction ofbiotypes within these genera.Proteus spp always demonstrated strong (3+)

urease activity and could therefore be readilydifferentiated from Providencia stuartii, irrespectiveof inoculum size. With high inocula Klebsiellaaerogenes/oxytoca isolates all gave a strong ureasereaction, sufficient to differentiate them from isolatesof Enterobacter spp. The medium and experimentalconditions here may also, as suggested by Chadwicket al. (1974), differentiate isolates of Yersiniaenterocolitica from E coli. The single isolate ofY. enterocolitica evaluated here demonstrated ureaseactivity over a wide range of inoculum size.The rate of inositol fermentation by non-Kleb-

copyright. on F

ebruary 16, 2020 by guest. Protected by

http://jcp.bmj.com

/J C

lin Pathol: first published as 10.1136/jcp.30.6.495 on 1 June 1977. D

ownloaded from

Factors affecting the value of a simple biochemical scheme for identifying Enterobacteriaceac 499

Table 1 Effect of inoculum size on results recorded by different species or genera on various screening media

Number of isolates giving positive results in screening media

Screening test and organism (total number) t Inoculum: viable cells/ml

10' 108 10 10J' 105 10'

Aesculin hydrolysisKlebsiella aerogenes/oxytoca (10)Enterobacter aerogenes (7)Escherichia coli (10)Proteus vulgaris (4)Proteus rettgeri (3)

Citrate utilisationK aerogenes/oxytoca (10)E aerogenes (7)P rettgeri (3)Proteus mirabilis (5)Citrobacter spp (6)Pseudomonas spp (5)

Ornithine decarboxylationK aerogenes/oxytoca (10)E aerogenes (7)Esch coli (10)P rettgeri (3)P mirabilis (5)Proteus morganii (3)Providencia stuartii (2)Citrobacter spp (6)

Resistance to KCNK aerogenes/oxytoca (10)K pneumoniae (2)E aerogenes (7)Esch coli (10)Proteus spp (I 5)Citrobacter spp (6)Pseudomonas spp (5)

MotilityK aerogenes/oxytoca (10)E aerogenes (7)Esch coli (10)Proteus spp (15)Citrobacter spp (6)Pseudomonas spp (5)

H2S productionP ulgaris (4)P rettgeri (3)P mirabillis (5)P morganii (3)Citrobacter spp (6)

Urease activity §K aerogenesloxytoca (10)

E aerogenes (7)Enterobacter clocae (3)Proteus spp (15)Providencia stuartii (3)Yersinia enterocolitica (1)

Inositol fermentation §K aerogenes/oxytoca (10)

E aerogenes (7)

Esch coli (10)P rettgeri (3)Proi'idencia stuartii (3)

3 + (10)$ 3 + (10) 3 + (10) 3 + (10) 3 + (10) NTNT 3 + (7) 3 + (7) 3 + (7) 3 + (7) 3 + (7)NT 2 + (1) 1 + (1) + (0) + (0) (0)+ (0) -(0) + (0) + (0) + (0) NT

+ (0) -(0) t (0) + (0) - (0) NT

+ (10) + (10) + (6) + (5) + (I) NTNT + (7) + (7) + (7) + (6) + (6)+ (3) + (3) + (3) + (3) + (3) NT+ (5) + (3) + (2) + (1) NT NTNT + (6) --(6) + (6) + (6) + (6)NT -(5) t-(5) +-(5) -(5) + (5)

+ (0) + (0) + (0) + (0) -(0) NTNT + (7) + (7) t (4) + (4) + (4)NT + (8) + (8) + (0) -(0) + (0)+ (1) +( 1) -(l) + (1) + (1) NT+ (5) + (5) + (5) + (5) NT NT+ (3) -(3) + (1) + (1) -- (1) NTNT + (2) + (0) +(0) + (0) + (0)NT -(6) (6) -(6) -(6) + (5)

+ (10) + (10) + (7) + (7) + (5) NTNT + (0) + (0) + (0) + (0) + (0)NT - (7) + (7) + (7) + (7) + (7)NT + (0) + (0) + (0) + (0) +(0)+ (15) + (15) (I5) (15) NT NTNT + (6) + (6) (6) t (0) + (0)NT + (5) + (5) + (0) + (0) + (0)

+ (0) + (0) + (0) + (0) + (0) NTNT + (7) + (7) + (7) + (7) + (7)NT + (9) + (5) + (2) + (2) - (2)+ (15) + (15) + (15) + (15) NT NTNT +-(6) - (6) + (6) + (6) T (6)NT (3) (3) + (3) + (3) (3)

+ (4) + (4) (4) -(4) -(2) NT+ (0) + (0) +(0) (0) -(0) NT

+ (5) + (5) t (5) -(4) NT NT+ ((3) (3) + (1) -(1) -(1) NTNT -(5) -(5) (5) -(4) -'(4)

3 + (3) \3 + (1) \2 + (9) \2 + (3) \2 + (3) lNT2 + (7) f2 + (9) + (1) l + (7) + (7)NT + (0) -(0) + (0) T (0) + (0)NT I + (3) 1 (3) + (0) + (0) +(0)3 + (15) 3+ (I5) 3 + (15) 3 + (15) NT NTNT + (0) +-(0) - (0) (0) -(0)NT I - (1) 1l-(1) I + (1) + (O) +(O)

3 + (10) 3 + (10) 3 + (8) 3 (4) \2 + (8) \

2 --(2) 2 + (5) Jl + (2) JNTI +(1)

NT 3 + (7) 3-(7) 3 + (7) 3- +(1) 2 + (5)2+(6) fl +(2)

NT + (0) + (0) + (0) + (0) + (0)3 (3) 3+ (3) 3 + (3) 3 + (3) 2 + (3) NTNT 3 (3) 3 T(3) 3 + (3) 3 + (3) 3+ (3)

*Results recorded in semi-quantitative manner for aesculin hydrolysis, inositol fermentation, and urease activity, as described under Materialand Methods.tTotal number of each species or genus evaluated for a specific test.+Number of isolates (in parentheses) giving positive test reaction.§Different reactions given by isolates of Klebsiella aerogenes/oxytoca and Enterobacter aerogenes on urea and inositol media are recorded separ-ately at each inoculum level.NT = not tested.

copyright. on F

ebruary 16, 2020 by guest. Protected by

http://jcp.bmj.com

/J C

lin Pathol: first published as 10.1136/jcp.30.6.495 on 1 June 1977. D

ownloaded from

500 J. G. Barr, Rebecca J. Mahood, and K. P. W. Curry

Table 2 The forty m?ajor biotypic reaction patterns,* on agar screening media, allowing immediate identification of1529 clinical isolates ofEnterobacteriaceae and Pseudomonas aeruginosa

Biotype reaction patterns in screening mediat

Organismn No. of isolates A C PPA 0 I XLD KCN H,S INOS U M No. ofadditional

Total No. with biotypespecific reactionreaction patternspattern

Escherichia coli 827 530* Y - - - - 145 385235: Y - - - - 59 17619 _ R - - - 19 03: - - - R 3 07 - NG 0 7 09 N-G 0 9

10 2 Y 9 15 2- - Y 3 29 y _ 0 9

Klebsiella aerogenesl 191 141 2-Y 3 2 0 141edwardsii/atlantae 39 2- Y 3 0 39 3

1 § Y 3 2- 0 11Klebsiella oxytoca 185 92$ 2- - Y 3 2 0 92

81 2- - Y - 3 I 0 81 012 2- Y 3 0 12

Klebsiella pneumoniae 3 2 2- Y 0- 21 2- Y 2 I 0 1 0

Klebsiella ozaenae 2 - Y - +Enterobacter aerogenes 81 77+ 2 Y 3 70 7

4 2 R 3 4 0 5Serratia spp 2 29 2- Y 3 I 2 0 0Citrobacter spp 74 42: - Y - - 7 35

25+ - Y - - 7 83+ - - - - Y - - 0 3 52 - - - - Y 2 02 B 0 2

Proteus mirabilis 89 48$ B 3 48 029 - B 3- 29 0 512 Y- - 3- 12 0

Proteus vulgaris 12 6$ Y 3 6 04$ - - Y 3 4 0 32 1 - - Y 3 2 0

Proteus morganii 10 10o - R 3 9 1 1Proteus retigeri 21 13+ Y 3 3 9 4

6 1 Y 3 3- 2 4 42 2 Y 3 3- 2 0

Providencia stuartii 5 5 - - R 3 - O 5 2Providencia alcalifaciens 4 4 R 3 1 2Yersinia enterocolitica 3 3+ R 3 0 0Salmonella spp BY - -Shigella spp -Pseudonomas spp 21 16+ - R 9 7 3

5 1'- R - 5 0

*Thirty-three other observed reaction patterns included 15 represented by a single isolate and 18 represented by more than one isolate but lessthan one per cent of the total isolates of species concerned. A further five reaction patterns could be resolved only after two further biochemicaltests (Tables 3, 4).1Biochemical patterns were established on the basis of aesculin hydrolysis (A); citrate utilisation (C); phenylalanine deaminase activity (PPA);ornithine decarboxylase activity (0); indole production (I); colony colour (R = red, B = black, Y = yellow, NG = no growth) on xyloselysine deoxycholate agar (XLD); hydrogen sulphide production (H,S) and inositol fermentation (tNOS) in inositol-hydrogen sulphide-motilitymedium; and urease activity (U). Motility of isolates of each reaction pattern were recorded as positive (-) or negative (-) on inositol-hydrogensulphide-motility medium at 37°C. All other reactions were recorded as described in Materials and Methods.+The expected specific reaction patterns compiled from published reports on bacteria derived from culture collection and clinical sources.

§Of a total of 11 isolates 8 were from patients in closely adjacent wards and represented repeated isolations from three different patients.IPublished results for Salmonella spp, Shigella spp, and Klebsiella ozaenae not included in this study of isolates from clinical material.UThree cultures include two atypical isolates of Serratia spp.

sielleae was little influenced by inoculum size andisolates of Enterobacter aerogenes showed rapidfermentation over a wide dilution range. The effectof inoculum size was most pronounced and of mostimportance among Klebsiella isolates. A fall ininositol fermentation paralleling a fall in inoculum

size could be of importance in distinguishingKlebsiella aerogenes/oxytoca/edwardsii/atlantae, re-corded elsewhere as consistently rapid fermenters,from slow fermenters such as K ozaenae and Kpneumoniae (Barr and Mahood, 1976).Our results showed the important effect of

copyright. on F

ebruary 16, 2020 by guest. Protected by

http://jcp.bmj.com

/J C

lin Pathol: first published as 10.1136/jcp.30.6.495 on 1 June 1977. D

ownloaded from

Factors affecting the value ofa simple biochemical scheme for identifying Enterobacteriaceae

inoculum size on the results obtained in the bio-chemical tests described on screening media. So faras our experience goes, we do not know of any casein which the recording of a positive reaction at oneinoculum level was accompanied by negativereactions at higher inoculum levels. The intrinsicerror must be considered to be the recording, in theuse of low inocula, of false negative results in 16-hourincubations. Similarly, even at the highest inoculumlevels (109 viable cells/ml) insufficient turbidity isintroduced to influence the recording of false positiveresistance to potassium cyanide.

Indeed, within a wide range of inoculum size(109-107 viable cells/ml, which is accompanied by aclearly visible reduction in turbidity and reduction inoptimal density) the semi-quantitative recording ofaesculin hydrolysis, inositol fermentation, andurease activity was influenced in few cases.

EVALUATION OF SCREENING SCHEME IN

ROUTINE USE

Normally the colony characteristics of the bacteria

screened were observed on nutrient blood agar andMacConkey agar, the inoculum for screening mediabeing derived from the MacConkey agar plate. Thisallowed purity of cultures to be established andlactose fermentation to be recorded.A total of 1659 isolates were screened by the

methods described. All isolates except 21 of Pseudo-monas spp were oxidase-negative, Gram-negativebacilli. The reactions obtained gave a total of 78distinguishable patterns, the major 45 patterns beingshown in Tables 2, 3, and 4.The reaction patterns expected for different

species or genera on the basis of published reportsare shown in Table 2 and confirmed with stockcultures in this study. The majority of isolates ofeach species, species group (for example, Klebsiellaaerogenes/edwardsii/atlantae), or genus within the1659 isolates investigated were identified by thecharacteristic patterns described. Thus, for example,787 out of 827 isolates (95-1 %) of E coli, 180 out of191 isolates (94-2 %) of K aerogenes/edwardsii/atlantae, and 87 out of 132 isolates (65-9 %) of

Table 3 Resolution of biotype reaction patterns characteristic of species of Klebsielleae (Enterobacter aerogenes, Ecloacae, Hafnia alvei, and Serratia marcescens) but not allowing immediate identification: tests to completeidentification ofsome isolates

Biotype reaction patterns on screening media

Total A C PPA 0 I XLD KCN H,S INOS U M Additional tests for Identificationisolates organism identification

24 2+ + - + - Y + - - - 20 4 Enterobacter aerogenes(8)*E. cloacae (5)Serratia marcescens (1)Not identifiedt (10)

Arabinose fermenta-tion testLysine decarboxylasetest

2 2+ + - + - Y + - - I + 2 0 Enterobacter cloacae (2)

*Numbers in parentheses = number of isolates ascribed to each species.tFurther tests not carried out.

Table 4 Resolution of biotype reaction patterns characteristic of Enterobacter cloacae, Hafnia alvei, or Citrobacterspp but not allowing immediate identification: tests to complete identification ofsome isolates

Biotype reaction patterns on screening media Additional tests for Identificationorganism identification

Total A C PPA 0 I XLD KCN H,S INOS U Misolates + -

7 - + - + R - - - 5 2 Citrobacter spp (5)*Not identifiedt (2)

20 - - + + Y + - - - 16 4 Methyl-red test, Citrobacter spp (12)lysine decarboxylase Not identified (8)test

2 - + - + - y + _ _ _ 2 0 Enterobacter cloacae (I)Citrobacterspp (1)

*Numbers in parentheses = number of isolates ascribed to each species.tFurther tests not carried out.

501

copyright. on F

ebruary 16, 2020 by guest. Protected by

http://jcp.bmj.com

/J C

lin Pathol: first published as 10.1136/jcp.30.6.495 on 1 June 1977. D

ownloaded from

J. G. Barr, Rebecca J. Mahood, and K. P. W. Curry

Proteus spp gave patterns immediately recognisablefrom published reports of their biochemical activities.The 45 major patterns described comprised 1584

isolates, 95-3 % of the total number screened. Ofthese the 1529 isolates (921 %) listed in Table 2were identified to genus or species level withoutfurther tests. The identification of a further 55isolates to genus or species level was completed afterfurther tests (Tables 3, 4).

Three major biotypes typical of published reportsof biochemical activity and 6 minor biotypic reactionpatterns of E coli were recorded. The minor biotypesincluded those showing aesculin hydrolysis andinositol fermentation already described (Barr andMahood, 1977) and did not hinder identification.Within this scheme Yersinia enterocolitica is differ-entiated from E coli on the basis of colour change onXLD agar and weak urease activity. Pasteurellapneumotropica is differentiated from Y enterocoliticaon the basis of oxidase activity.Among the Klebsielleae the rapid inositol fermen-

ters K aerogenes, K oxytoca, and E aerogenes wereclearly differentiated from one another and fromother Klebsielleae the incidence of K oxytoca,(49-2%) of Klebsiella isolates, being considerablyhigher than in previous reports (Davis and Matsen,1974). Among other Klebsiella spp only Kpneumoniaecould otherwise be differentiated, although stockcultures of K ozaenae could also be presumptivelyidentified on the basis of slow inositol fermentation(Barr and Mahood, 1976). One novel biotype amongthe K aerogenes/edwardsii/atlantae group wascharacterised by the absence of aesculin hydrolysis.

Differentiation of Enterobacter spp and Serratiaspp was not achieved within the screening scheme.Some isolates of E aerogenes lacked the ability toferment inositol under the conditions described andcould be differentiated from Enterobacter cloacaeand Serratia marcescens only by further biochemicaltests (Table 3) (Chadwick et al., 1974; Barr andMahood, 1977).Two novel isolates of Klebsielleae were also

encountered in this investigation (Table 2). Thesecomprised two rapid inositol-fermenting, lactosenon-fermenting isolates of Serralia spp which weredifferentiated from other biotypes by the presence oftryptophanase activity.

Three biotype reaction patterns (Table 4) did notallow immediate differentiation between Entero-bacter cloacae and Citrobacter spp. Two reactionpatterns included indole positive reactions and wereincluded here to confirm the absence of tryptophanaseactivity among isolates of E cloacae. This wasconsidered necessary in later stages of the screenwhen the presumptive presence of tryptophanaseamong other Klebsielleae had been established. As

shown in Table 4, no indole-positive isolates ofEnterobacter cloacae were reported from this study.

This scheme should, however, allow the differenti-ation of most Enterobacter spp and Serratia sppwithin 48 hours of isolation. The presence or absenceof lactose fermentation on primary isolationMacConkey agar plates and the test results obtainedon basic screening media within 16 hours shouldallow selection of further biochemical tests (arabin-ose fermentation, lysine decarboxylation, and methyl-red test) to allow differentiation of genera and speciesafter a fturther 24 hours.

Reaction patterns given by eight major biotypesof Citrobacter spp are shown in Tables 2 and 4, thebiotypes described in Table 4 requiring to be furthertested by methyl-red and lysine decarboxylase teststo ensure differentiation from E cloacae. The majorbiotypes recorded (Table 2) had the propertiescharacteristic of the two species, CitrobacterJreundii and C. intermedius, recognised by Sedlaket al. (1971). The two major biotypes recognised inTable 4 closely resembled C intermedius, C koseri,and the genus Levinea described by Young et al.(1971). However, other intergrading isolates are alsorecorded, which could be expected from the numberof species originally described by Werkman andGillen (1932). An additional five minor biotypesincluded other intergrading forms which producedindole and hydrogen sulphide either on XLD orinositol-hydrogen sulphide-motility medium. It hasbeen noted in Table 2 that Citrobacter spp andSalmoniella spp are differentiated on the basis ofsensitivity to potassium cyanide. Differentiation ofProteus spp and Providencia spp within the screeningregimen was clearly defined. The inositol fermentingbiotypes of Providenicia stuartii (three in number)were clearly differentiated on this test alone fromProvidencia alcalifaciens.Ten major and 13 minor biotype reaction patterns

were distinguished among isolates of Proteuts spp.All biotypes of Proteus rettgeri were readily differen-tiated from other Proteus spp on the basis of inositolfermentation, and were distinguished from Provi-dencia stuartii by virtue of colony colour on XLDagar and strong urease activity.

Proteus morganii isolates gave the most consistentreaction pattern and were immediately recognisable.Although no citrate positive isolates were reportedhere these would, should they arise, presumablyreadily be differentiated from indole-positive Proteusmirabilis isolates on the basis of colony colour onXLD agar and the absence of H.S production oninositol-H2S-motility agar.As would be expected from such a study, the

largest number of Proteus isolates were ascribed tothe species P mirabilis and to biotypes related to

502

copyright. on F

ebruary 16, 2020 by guest. Protected by

http://jcp.bmj.com

/J C

lin Pathol: first published as 10.1136/jcp.30.6.495 on 1 June 1977. D

ownloaded from

Factors affecting the value of a simple biochemical scheme for identifying Enterobacteriaceae

P vulgaris, species which have shown in numericaltaxonomic studies to be closely related (McKell andJones, 1976). Indeed, as shown in Tables 2, thepresence of indole production among large numbersof P mirabilis isolates (46%) which has beenpreviously noted by Matsen et al. (1972), variabilityin H2S production by P mirabilis isolates on XLDagar, and variability in citrate utilisation by isolatesofP vulgaris make it difficult on the basis of a narrowrange of biochemical tests to ascribe some isolatesin this group to particular species. However, althoughthe presence of eight biotypes ofP mirabilis and sixbiotypes of P vulgaris represent intergrading formswhich in this screening scheme may defy definition,the recording of these biotypes in themselves is avaluable aid in epidemiological surveillance withinthis group.

Isolates of Pseudomonas spp were also included inthe study because they constituted the mostcommonlyisolated oxidase-positive Gram-negative bacillus,which Chadwick et al. (1974) indicated were clearlydifferentiated from other Gram-negative bacilli by apositive citrate reaction, a red colony colour onXLD agar, and negative results in other screeningtests. We have confirmed these observations, includedweak aesculin hydrolysis as a further criterion foridentification in 16-hour incubations, and distin-guished Pseudomonas aeruginosa by virtue of greenpigmentation on phenylalanine deaminase agar.

Discussion

Forty-five major and 33 minor biotypic reactionpatterns were recorded within this biochemicalscreening regimen for the identification of Entero-bacteriaceae. The major patterns included thosewhich, from published reports of the biochemicalactivities of different genera and species, would havebeen expected to predominate. The scheme allowed92-1 % of isolates to be identified to species level or,in the case of many Klebsiella spp and Citrobacterspp, genus level on the basis of recognisable reactionpatterns. A further 3-2% were identified as Citro-bacter spp or as species of Klebsielleae after threefurther biochemical tests. The remaining 4 7°% ofisolates comprised those which have been ascribed tospecies or genera but whose reaction patterns wereatypical or poorly represented as a percentage ofisolates of the particular species or genera.No attempt has been made to ascribe species

names to isolates of Citrobacter or to introducebiochemical tests to further species definition amongisolates of Klebsiella. Similarly, the limitations indifferentiation between Proteus mirahilis and Proteusvulgaris in the presence of so many isolates givingatypical reactions has been noted and the value of

simple recording of biotypic reaction patterns,without attempting species definition, has beenrecognised.The reproducibility of the screening tests was

established by repeated testing with stock culturesand by the screening of multiple isolates of anorganism from the same source on successiveoccasions. Indeed, by recognising biotypes it hasoften been possible to predict the source of aparticular bacillus.We have emphasised, however, that careful

standardisation of techniques is required to obtainreproducible biotypic patterns. Of the possiblevaliables we have considered inoculum size to becritical, the importance of standardisation being, asreported, more important in some tests than inothers. But we do not know of any case in which apositive reaction at one inoculum level was accom-panied by a negative reaction with the use of a higherinoculum. We therefore suggest that using a heavyinoculum (at least 107 viable cells/ml), dispensed inthe volume we used, should prevent the recording offalse negative results in 16-hour incubations.The system described is clearly adaptable to use

with a multi-inoculator device, using replidishes asgrowth chambers, although it would then be neces-sary to ensure that a sufficiently high inoculum wasdelivered. By manual means, using Pasteur pipettesfor inoculation, 50 cultures can be put through thesystem, with a total of two hours required forinoculum preparation, inoculation, and recordingresults.We envisage that this system might provide a

means for epidemiological surveillance of Entero-bacteriaceae by biochemical typing. Result recordingcould easily be accommodated on a simple computerprogramme which could allow rapid retrieval ofspecific biotypic reaction patterns. The primarybiotyping reaction scheme can easily be supple-mented by biotyping schemes specific for particulargenera or species.

We thank Mrs W. Booth for her energy and care inpreparing the text.

References

Barr, J. G. and Mahood, R. J. (1976). An assessment ofinositol fermentation by Klebsiellae and its implicationin screening media. Journal of Clinical Pathology, 29,827-832.

Barr, J. G. and Mahood, R. J. (1977). Complementaryuse of aesculin hydrolysis and inositol fermentation inthe characterisation of Klebsielleae. Journal of ClinicalPathology, 30, 4549.

Barritt, M. M. (1936). The intensification of the Voges-

503

copyright. on F

ebruary 16, 2020 by guest. Protected by

http://jcp.bmj.com

/J C

lin Pathol: first published as 10.1136/jcp.30.6.495 on 1 June 1977. D

ownloaded from

504

Proskauer reaction by the addition of a-naphthol.Journal ofPathology and Bacteriology, 42, 441-454.

Berger, U. and Einecke, H. (1973). Zur Unterscheidungder Gattungen Klebsiella-Enterobacter und ihier Artenin der Routinediagnostik. Zentralblatt fur Bakterio-logie, Parasitenkunde, Infektioniskrankleiter, undHygieneErst Abteilung: Originale, Reihe A: MedizinischeMikrobiologie und Parasitologie, 225, 471-486.

Carpenter, K. P., Lapage, S. P., and Steel, K. J. (1966).Biochemical identification of Enterobacteriaceae. InIdentificationt Methods for Microbiologists, Part A,edited by B. M. Gibbs and F. A. Skinner, p. 21.Academic Press, London.

Chadwick, P., Delisle, G. J., and Byer, M. (1974). Bio-chemical identification of hospital enterobacteria byreplica agar plating. Canadian Journal of Microbiology,20, 1653-1664.

Christensen, W. B. (1946). Urea decomposition as a

means of differentiating Proteus and paracolon culturesfrom each other and from Salmonella and Shigellatypes. Journal of Bacteriology, 52, 461-466.

Cowan, S. T. (1974). Cowan and Steel's Manual for theIdentification of Medical Bacteria, 2nd edition. Cam-bridge University Press, London.

Davis, T. J. and Matsen, J. M. (1974). Prevalence andcharacteristics of Klebsiella species: relation toassociation with a hospital environment. Journal ofInfectious Diseases, 130, 402-405.

Donovan, T. J. (1966). A Klebsiella screening medium.Journal of Medical amid Laboratory Technology, 23,194-197.

Edwards, P. R. and Ewing, W. H. (1962). Identificationiof Enterobacteriaceae, 2nd edition. Burgers PublishingCompany, Minneapolis.

Fay, G. D. and Barry, A. L. (1972). Rapid ornithinedecarboxylase test for the identification of Entero-bacteriaceae. Applied Microbiology, 23, 710-713.

Lassen, J. (1975). Rapid identification of Gram-negativerods using a three-tube method combined with a

dichotomic key. Acta Pathologica et MicrobiologicaScandinavica, 83, (6) Supplement, 525-533.

McKell, J. and Jones, D. (1976). A numerical taxonomicstudy of Proteus-Providence bacteria. Journal of AppliedBacteriology, 41, 143-161.

Matsen, J. M., Blazevic, D. J., Ryan, J. A., and Ewing,

J. G. Barr, Rebecca J. Mahood, and K. P. W. Curr.'W. H. (1972). Characterization of indole-positiveProteus mirabilis. Applied Microbiology, 23, 592-594.

Miles, A. A. and Misra, S. S. (1938). The estimation ofthe bactericidal power of the blood. Journal ofHygiene,38, 732-748.

M0ller, V. (1955). Simplified tests for some amino-aciddecarboxylases and for the arginine dihydrolase system.Acta Pathologica et Microbiologica Scandinavica, 36,158-172.

Nord, C. E., Wadstrom, T., and Dahlback, A. (1975).Evaluation of different diagnostic kits for Entero-bacteriaceae. In New Approaches to the Identificationof Micro-organisms, edited by C. G. Heden and T.Illeni, pp. 393-406. John Wiley.

Rogers, K. B. and Taylor, J. (1961). Laboratory diagnosisof gastroenteritis due to Escherichia coli. Bulletin ofthe World Health Organisation, 24, 59-71.

Rosner, R. F. (1970). Identification of enteric bacillidirectly from primary isolation media using reagent-impregnated paper strips. American Journal of ClinicalPathology, 54, 587-595.

Sedlak, J., Puchmayerova-Slajsova, M., Keleti, J., andLiideritz, 0. (1971). On the taxonomy, ecology andimmunochemistry of genus, Citrobacter. Jourlnil ofHygiene, Epidemiology, Microbiology and Immunolbgy,15, 366-374.

Taylor, W. 1. (1969). Isolation of Shigellae. 1. Xylose-lysine agars; new media for isolation of enteric patho-gens. Amzericani Joulrnal of Clintical Pathology, 44,471-475.

Tomfohrde, K. M., Rhoden, D. L., Smith, P. B., andBalows, A. (1973). Evaluation of the RedesignedEnterotube-a system for the identification of Entero-bacteriaceae. Applied Microbiology, 25, 301-304.

Wasilauskas, B. L. (1971). Preliminary observations onthe rapid differentiation of the Klebsiella-Enterobacter-Serratia group on bile-esculin-agar. Applied Mic ro-biology, 21, 162-163.

Werkman, C. H. and Gillen, G. F. (1932). Bacteriaproducing trimethylene glycol. Journal of Bacteriology,23, 167-182.

Young, V. M., Kenton, D. M., Hobbs, B. J., and Moody,M. R. (1971). Levinea, a new genus of the familyEnterobacteriaceae. Interniational Journal of SystematicBacteriology, 21, 58-63.

copyright. on F

ebruary 16, 2020 by guest. Protected by

http://jcp.bmj.com

/J C

lin Pathol: first published as 10.1136/jcp.30.6.495 on 1 June 1977. D

ownloaded from