Evaluation of an adenosine 5' triphosphate assay as a screening method to detect significant bacteriuria

JOURNAL OF CLINICAL MICROBIOLOGY, Jan. 1976, p.42-46

Copyright© 1976 AmericanSocietyforMicrobiology PrintedinU.S.A.

Evalution

of

an

Adenosine

5'-Triphosphate

Assay as

a

Screening

Method

to

Detect

Significant

Bacteriuria

DAVID N.ALEXANDER, GRACE M. EDERER, ANDJOHN M.MATSENI*

DepartmentsofLaboratory Medicine andPathology,Pediatrics, and Microbiology, University of Minnesota

SchoolofMedicine,Minneapolis, Minnesota55455

Received forpublication 18August1975

Thebioluminescent reaction of adenosine 5'-triphosphate (ATP) withluciferin and luciferase has beenused in conjunction witha sensitive

photometer

(Lab-Line's ATP photometer) to detect significant bacteriuria in urine. This rapid method of screeningurine specimensforbacteriuria wasevaluated byusing 348 urinespecimenssubmitted to the clinicalmicrobiologylaboratory at the Univer-sity ofMinnesota

Hospitals

for routine cultureusingthe calibrated

loop-streak

plate method. Therewas 89.4%agreementbetween theculturemethod andthe ATP assay, with 7.0% false positive and 27.0% false negative results fromthe ATP assay using 105organisms/mlof urine orgreateraspositive forsignificant bacteriuria and less than 105 organisms/ml asnegative forsignificant bacteriu-ria.

Quantitativeurineculturesareroutinely per-formed on all urine specimens submitted for culture todistinguish between urines with sig-nificant bacteriuria (100,000 bacteria/ml or greater)(8, 10, 12) and urines thataresterileor

have lowercounts. Culturemethodsrequire 18

to 24 h for incubation before the number of organismscanbeassessed. Immediate approxi-mation ofthe numberoforganismscan be

ob-tained by Gram staining freshlyvoided urineor

by using phase-contrast microscopy(3, 19), but the evaluation ofnegative specimens by these methods is time consuming. Since more than 60% ofthe urines received for culture at the University of Minnesota Hospitals do not

con-tain significant numbers of organisms, it seemed important to evaluate arapid and

im-mediatemeansof determiningsignificant

bacte-riuria: the enzymatic bioluminescent reaction ofbacterial adenosine 5'-triphosophate (ATP) withluciferin-luciferase.

This assay was originally developed by the

National Aeronautics and Space Administra-tionforquickly determiningthe purity of

drink-ing water supplies during long space flights

(11), butcanbemade quantitative for bacterial

ATP intheurineby using three basic

prepara-tory steps. (i) Nonbacterial ATP must be

released and destroyed. (ii) The chemical that

destroys the ATP in step (i) must be removed

priorto releasing the bacterial ATP. And (iii)

thebacterialATPmustbe releasedinto its free

soluble stateready toreact with luciferin and

'Presentaddress:DepartmentofPathology, University ofUtahMedical Center, Salt Lake City, Utah84132.

42

luciferase to

produce

light

and other

products

(5). The purpose of this paper is to report the results of our evaluation of the ATP assay method for determining

significant

bacteriuria.

MATERIALS AND METHODS

Reagents.Octyl phenoxy polyethoxyethanolwas

obtainedfrom Rohm andHaas Corp. and isa

non-ionic detergent used for the selective lysing ofthe

erythrocytes, leukocytes, and epithelial cells found

in urine. Dimethyl sulfoxide (Me2SO) of reagent

grade quality was obtained from the J. T. Baker Chemical Co. Trizma-7.7, a tris(hydroxymethyl)-aminomethane andhydrochloride buffer of reagent

grade, wasobtained from theSigma Chemical Co.

(no.T-4378)and dilutedto aconcentrationof 0.05 M,

giving a pH of7.7at 25C. Potato apyrase, an

ATP-ase, was also obtained from Sigma Chemical Co.

(grade I, A6132).

Firefly lantern extract. Firefly lantern extract

(FLE) was purchased from Worthington

Biochemi-cal Corp.,inFreehold, N.J. Each vial of this

lyophi-lizedwater extract of 50 mgoffirefly lanterns was

reconstituted with 5.0 ml of sterile distilled water,

resulting in asolution containingluciferin,

lucifer-ase, 0.05M KAsO4, and 0.02 M MgSO4, at pH 7.4. Each vial was mixed thoroughly in a Vortex mixer,

and, depending on the number of samples to run,

either two or three vials were reconstituted,

vor-texed, mixed together, and vortexed again,

result-inginahomogenous FLE solution (containing both

theenzymeand the substrate) for all the samples to

be analyzed in that run. The FLE was stored in a

dark container at -20 Cuntil it wasreconstituted.

Whenreconstituted, it was stored either on ice or in

a 4C refrigeratorandkept away from bright light.

ATP standards. ATP wag obtained from Sigma

Chemical Co.(stock no. FF-ATP) as a disodium salt.

on February 7, 2020 by guest

http://jcm.asm.org/

ATP BACTERIURIA ASSAY 43

Each vialcontained 1 mg of ATP, disodium, and 40 mg ofmagnesium sulfate and was stored at -20 C awayfrom the light. It was reconstituted by adding 10 ml of sterile trizma buffer to each vial, mixing, and then in a sterile manner dispensing the 0.1-mg/ml solution in 1-ml aliquots into 10 separate vials, which were stored at -20 C.

Equipment and supplies. The Lab-Line ATP pho-tometer, no.9140, was obtained from Scientific Prod-ucts,Inc.,ona rental basis.Disposable liquid scintil-lation vials were from Wheaton Laboratory Prod-ucts and were the Vitro "180" brand. Biopette, a semiautomatic pipette from the Schwarz/MannCo., was used fordispensing1-ml aliquots of urine. The reusable tips wereautoclaved inaphysiological (0.9

N NaCI) saline solution before use. Membrane fil-ters (Millipore Corp.) of 25 mm in diameter and 0.45-,um in pore size were used. The 100-pu Eppendorf pipette was used, and the tips were sterilized

be-tweenuses.

Standard curve using ATP standard. Standard ATP calibration curves were prepared twice daily with serialdilutions of ATP prepared immediately beforeusefrom a stock 0.1-mg/ml solution. The FLE usedinthe assay was reconstituted 1 to 2 h prior to the assaytimeand was dispensed into the scintilla-tion vials 5 to 20min before assay. It was found that the enzyme solution achieved more stability, evi-dencedby the smoother, more reproducible standard curves, ifreconstituted more than an hour before assay.

Theprocedure for assay with the ATP photometer waskept uniform for both the test samples and the standard solutions of ATP. Using a 100-pd Eppen-dorf pipetteand sterile tips, 0.1mlof the test sample

or ATP standard was injected into a scintillation

vialcontaining0.1 mlof FLEand 0.4 ml of trizma

buffer. This was swirled and then placed in the photometer, and the light emission value in counts per minute was recorded. This value is the count takenduring the minute after an exactly timed 15-s delay from the injection time (the delay being auto-maticallytimed by the instrument). Urinesassayed by the photometer were interpreted as positive or

negative for bacteriuria bycomparing the light

emis-sion unitsvaluewith theATPstandardcurve.

Milli-grams of ATP in100,000organisms wascalculated

tobe3x 10-8by using the value3 x 10-1

iLg

of ATP percell, whichisthemeanATP content ofa bacte-rial cell (5).

Standardcurve using bacteria. Several strains

of Escherichia coliand Klebsiella specieswere

inocu-latedintotryptic soy broth(GIBCO)and incubated

for 18 to 24 h. Assuming therewereapproximately

109 organisms/ml by the end ofthe incubation pe-riod,the broth cultures were thendilutedwith

ster-ile water so that the numbers of organisms per

milliliter rangedfrom 109to 103. One milliliter of

eachof these dilutions wasextracted with Me2SO

andassayedwith thephotometer.

Evaluation ofenzyme activity. Changes inthe

enzyme activityof the FLEthrough time, and

varia-tionsfromonebatch of FLEtoanother, have been

encounteredby manyresearchers(1, 4, 6, 13, 14, 16).

To offset the deleterious effects of the decreasing

activity of the FLE with time, onestandard curve was prepared prior to the processing of a group of test samples, and another standard curve was pre-pared after completing the assay of thetestsamples. The testsamples, however, were assayed when the enzyme activity was intermediate to these ex-tremes. Therefore, the difference between the "be-fore" and "after" standard curves was dividedinto

several sections, and several "new" standardcurves

were drawn. The test sample could then be assigned

to its appropriate standard curve by knowing the

timewhen the test sample wasassayed. Conversion from a light emission value of the test sample to a readinginorganisms per milliliter was then accom-plished from the standard curve. Since the

differ-encebetween thestandard curves prepared with the

sameenzymesolution at twotimes, 3 to 4 h apart, wassignificant, these measures were necessary. Sec-tioningthe standard curve was done under the as-sumption that enzyme activity decays approxi-mately linearly with respecttotime.

ATP assay of urine specimens. Releasing and destroying the nonbacterial ATP wasaccomplished by adding 0.3 ml of a 2:1 mixture of 0.5% octyl phenoxy polyethoxyethanol and a

10-,ug/ml

potato apyrase solution to a test tube (13 by 100 mm), to which a 2.0-mlaliquot of thewell-mixed urine speci-men was thenadded (using the Biopette). This mix-turewasallowedtostand15 to 30min,during which

time the octyl phenoxy polyethoxyethanol selec-tively lysed any cell that might be in the urine

(leukocytes, erythrocytes, or epithelial cells), but not the bacterial cells (15).

Removing the ATPase beforereleasing the bacte-rialATPwas doneby filteringthe urine through a membrane filter (Millipore Corp.). The filter was

washedwith 2 ml of buffer(trizma)to remove any remainingATP,ATPase, and otherinterfering sub-stances,leavingonly the bacteriaonthefilter.

Extractingbacterial ATPwithMe2SOwaschosen

asthe method forreleasing bacterial ATP because of

itsconvenience. A2-mlamountofMe2SOwasplaced

inaplastic disposable weighing boatto which the filter membrane from the membrane filterwasthen added. The membranewasremovedasepticallyfrom

the filterapparatusby usingaforceps flamed three

timeswithethanol. The filter membrane stood5to

15min in theMe2SO before pouringtheMe2SO-ATP

lysateintoasteriletesttubeandfreezingat -20C until thetimeof assay withthephotometer.

Clinicalspecimens and reference method. The calibrated 0.001-ml loop method was used as the reference methodinevaluatingthe ATP assay(2-7).

Atotal of348urinespecimens receivedatthe Uni-versity of Minnesota Hospitals were tested in the ATPphotometer immediatelyafter the clinical

mi-crobiological laboratory had cultured them by the

calibrated loop method. Carewastaken to process

the urines assoon aspossible,or torefrigeratethem

until the time whenthey couldbe

assayed by

the photometer. Samples werecollected atall timesof

thedayornight, andnospecialpreparationor

selec-tion of the urine specimens was made. The

orga-nisms were identified according to the

procedures

usedinthisclinical

microbiological

laboratory(2). 3, 1976

on February 7, 2020 by guest

http://jcm.asm.org/

RESULTS

Figure 1 is an example ofa standard curve made by plotting arbitrary light emission val-uesagainst concentrations ofATPfrom10-;to

10-9 mg/ml. This figure illustrates the loss of sensitivity of the bioluminescent reactionin the critical range (equivalentto 105organisms/ml) using the materials and instrumentation de-scribed earlier. An emission curve made by extracting the ATP from various dilutions ofan overnight broth culture of E. coli is shown in Fig. 2. Similar curves were obtained when other strains ofE. coli andKlebsiella species were assayed.

The results of quantitative urine cultures compared with bacterial ATP assays on the

same 348 specimens showed atotal agreement of89.4%. Of the 37 urines which were not in agreement, the ATP assay gave 20 false posi-tives and 17false negatives, or 7.0 and 27.0%,

256

,000-cn

z

128,000-z °-

64,000-t 32,000-CD >

16,000-ir

8,000-°

4,000-165,000

19,500

:EQUIVALENTTO105

\ORGANISMS/ml*

7,300

5,650

respectively. A total of 63 urines, or 18.1%, werepositive for bacteriuria, asdeterminedby the culture method. Thesensitivityof the ATP assay was 73.0%, whereas the specificity was

93.0%.

Table 1 presents a moredetailed analysisof boththepositive urinespecimensthatwere in agreementandof 37 urinespecimensthatwere notinagreement. Inallthreeclassesofresults, whether false positive, false negative, or posi-tiveby both methods, thereseemstobe a

gen-eral agreement with the distribution of

orga-nismidentities given by theaverageof5years of urine culturesattheUniversityof Minnesota Hospitals.

Table2indicates the correlationbetween the culture method andthe ATP assaymethod at

512,000-

256,000-Z 128,000

-64,000

-cn

32,000

-I16,000

-E 8,000

CD o 4,000

-2,000

-10-5 10-6 107 10-8 10 9 MILLIGRAMS OF ATP ASSAYED

FIG. 1. An ATPstandardcurvegraphedas light

emission units versus concentration of ATP in the

standard solution. *The value of 105 organismslml

equals 3 x 10-8mgof ATPperml,using0.1-ml

as-saysample size.

1,000

-e 400,720

70,034

8 1,687

4 1,557

25X10 25X103

I8 I107 b6 105 104 103 1o2

ORGANISMS

FIG. 2. A light emission curveobtained from the

ATPassayof seriallydilutedE. coli.

TABLE 1. Identificationof the organisms in urine samples positive by the culture method

False negatives False positives Positive by both meth-ods Predominant species

No.of sam- No. ofsam- % No. of sam- %

ples ples ples

Escherichia coli 4 23.5 2 10.0 19 41.3

Klebsiella 3 17.6 1 5.0 5 10.9

Proteus species 1 2.2

Streptococcus 3 17.6 1 5.0 6 13.0

Proteus mirabilis 2 11.8 3 6.5

Pseudomonasaeruginosa 1 2.2

Staphylococcus 2 11.8 3 15.0 4 8.7

Candidaalbicans 2 10.0 3 6.5

Gram-negative rods 1 5.9 1 5.0 1 2.2

Other gram-negative rods 2 11.8 3 15.0 3 6.5

Nogrowth 7 35.0

on February 7, 2020 by guest

http://jcm.asm.org/

TABLE 2. Correlation of urine specimens with standard culture methods and ATP assay

No. of No. of No. of

sam-pies sam-

sam-Concn of

_{Concnofsor grwith}

or- with ples ples

Agree-poi nega- ment

ganbisms

growth tive by tive by

m(n

bycul- ATP as- ATP

as-ture

method say say

0to<102 178 7 171 96.1

102 to <103 4 0 4 100

103to<104 50 2 48 96

104 to <105 53 11 42 79.2

-105 63 46 17 73.0

the various levels of organism content. There was 96. 1%agreement betweenthe two methods when the number of organisms per milliliter being

analyzed

wasless than 10,000. However, agreement

dropped

to73.0%when greater than 10;organisms were present, due to the number of false negative results. Over

one-half

of the

false

positive results occurred inthe 104

orga-nisms range.

Figure 3 presents a

profile of the

46 urines designated positive by both quantitative cul-ture and by

the

ATP assay. Over 75% of the urine cultures designated as greater than or

equal

to 100,000organisms/ml by the

quantita-tive culturemethod are

quantified

as 107 orga-nisms/ml or greater by the ATP assay, and only 9% are

quantitated

as 100,000

organisms/ml by

the

ATP assay.

Data onthelight emission values of the false negatives show that several are borderline (quite near the

cut-off

point), but most are well within the "negative range." The urine speci-mensnegative

by

the ATP assaycould

unfortu-nately

notbe

quantitated

more

exactly

because

of the lack of

sensitivity

of the

standard

curve inthe concentrations below 10-9 mg of ATP per ml.

DISCUSSION

A screening test for

significant

bacteriuria should have few false

positive

results

(when

compared

with an

accepted

culture

method)

and, ideally,

notanyfalse

negative

results. The

cause ofthe

significant

number offalse nega-tivesgivenbythe ATP assay doesnot seem to

berelatedtothe kindof

organisms,

sinceTable 1 shows no

striking

pattern in terms of the

organisms

involved.

Patients

whose

urine demonstrated a false negative result consistently gave results in agreement with the culture method on a later urine sample. Records were also kept of the antibiotics being used by the patients from which the specimens were taken and on the length of time the urinespecimens were refrig-erated. Neither factor appeared to influence results, with both groups being proportionately represented in the agreeing specimens and in those not in agreement.

In searching for a cause of the false nega-tives, it can be

suggested that all the

apyrase solution is not

rinsed

from

the filter

membrane by the 2-ml aliquot of buffer. The apyrase might then destroy the ATP released by the bacterial cells, thus giving rise to a false nega-tive value. However, there is evidence that this is not the case. Many urinespecimens had bac-teria so plentiful that the organisms clogged the filter. On two urines it was possible to filter

only

a few drops of buffer through; yet these urines gavepositiveresults by the ATP assay.

Nevertheless,

urineswith fewer bacteria might

be

affected

by the remaining apyrase,

though

this

is

merely speculation.

Profiles of the false positive results show no

striking unifying

features

either.

No one

orga-nism group seems responsible forthe false

posi-tive

results.

It is

significant that

35% of the

false

positive results are on cultures

that

are

sterile

by

thequantitative culture

method. This

indicates that it is unlikely that all the false positives were due simply to

quantifying

border-line cases (104 organisms/ml) as higher,

al-though

the latter

did contribute

atleast 50% of

the false positives.

Positive results from

the

ATP assay were

converted

tothe number of organisms per

milli-

20-Un

Z

15-U.)

0-c _

*5-32.6%

13.0%

8.70/

34.8%

10.9%

105

106

107 1O8 >109

ATP QUANTITATION IN ORGANISMS/mI

FIG. 3. Urine samples positive by both the ATP

assayand culture method.

on February 7, 2020 by guest

http://jcm.asm.org/

literby using theATPstandardcurve. Results are in agreementwithother published data in thatgenerally the ATP assay gaveaveryhigh number of organisms permilliliter for the ur-ineswithbacteriuria

compared

with the num-bergiven by the culturemethod. The concentra-tions givenby the ATP assaywere

primarily

in the range of 107to 10"organisms/ml. Chappelle and Picciolo obtained 17 positive bacteriuria results from their ATP assay, whereas only eight of these were positive by the pour plate method, and all 17 of the positivesby the ATP assay were in the range of 107 and 10" orga-nisms/ml (5).

Thore et al. (15) have evaluatedanATP bac-teriuria assay similartothe oneevaluatedinthis paper. The major difference is that

they

used boiling TRISS buffer

containing

ethylenedia-minetetraacetate toextractthe ATP from bacte-rial cells, whereas we used Me2SO.

They

ob-tained 4% false negatives and approximately 30% false positives at an ATP concentration limit of4 x 10-10 M ATP. Our

values

of27% false negatives and 7% false positives are not strictly

comparable,

because we used an ATP concentration limit of6 x 10-11 M ATP. How-ever, our

data

agreewith

that of

Thoreet

al.

in that: (i) the number of falsepositives and false negatives are

inversely

related; (ii)

these pro-portionsaredetermined

by

the ATP

concentra-tion

limit;

and (iii) most

importantly,

one

can-noteliminate the false negatives without

incur-ring an

unreasonably large

number of

false

positives.

Both limits (4x 10-'°for

Thore

etal. [15] and 6 x 10-11

for

our

studies)

correspond to

10a

orga-nisms/ml. There is a 10-fold difference between their estimation and our estimation as to the ATP contentof a single bacterial cell, leading

tothe

differences

intheactual ATP limits used

inthe two studies.

The primary advantage for considering the ATPassaymethod for screening for bacteriuria isthatquantitative results can be obtained

im-mediately.

This advantage is

offset

by two

prob-lems withtheassay at this state of its develop-ment. Firstly, approximately 15 min of a tech-nologist's time is required for processing a speci-men, and, secondly, the ATP assay method, eventhough now commercially available, does notappear to be

sufficiently

accurate for clini-cal use since false negative results can be ob-tained. Evenwhen the sensitivity of the proce-dure is improved and the possibility of false negative results areeliminated, the procedure

will be

costly

toconsider for routineusebecause of theinitialcostoftheinstrument, thecostof materialsper test,and the amountof

technolo-gist

time

required

for processing. This latter

problem might be resolved if the

procedure

wereautomated.

ACKNOWLEDGMENTS

TheinterestandmanuscriptreviewofPaulAlexanderis gratefully acknowledged, asisthe excellent secretarial

as-sistanceofDebbieGuevara.

LITERATURE CITED

1. Aledorf, L. M., R.I.Weed, andS. B.Troup.1966. Ionic effectsonfirefly bioluminescence assayof redblood cellATP. Anal. Biochem. 17:268-277.

2. Blazevic, D.J. 1974.Laboratoryprocedures in diagnos-tic microbiology.p.44-46. Telstar Productions,Inc., St.Paul, Minn.

3. Brody, L.,M. C.Webster,and R. M.Kark. 1968. Identi-ficationofelementsofurinary sedimentwith phase-contrastmicroscopy. J. Am. Med. Assoc. 206:177. 4. Chappelle, E. W., and G. V. Levin. 1968. Use ofthe

firefly bioluminescent reaction for rapid detection andcountingofbacteria. Biochem.Med.2:41-52. 5. Chappelle, E. W.,and G. W. Picciolo. 1971. Bacterial

adenosine triphosphate (ATP) as a measure of uri-narytract infection. NASApubl. TSP1-10051. Tech-nology UtilizationOffice, Goddard Space Flight Cen-ter.National AeronauticsandSpaceAdministration, Greenbelt, Md.

6. Johnson, R. A., J.G. Hardman,A. E.Boadus, andE. W.Sutherland. 1970.Analysisof adenosine 3',5'-mon-ophosphate with luciferase luminescence. Anal. Bio-chem. 35:91-97.

7. Kass,E. H. 1956. Asymptomaticinfections of the uri-narytract.Trans.Assoc. Am.Physicians 69:56-63. 8. Kass,E. H. 1957. Bacteriuria and thediagnosisof

infec-tions of the urinary tract. Arch. Intern. Med. 100:709-714.

9. Kunin, C. M. 1961. The quantitative significance of bacteria visualized in the unstained urinary sedi-ment. N. Engl. J. Med.265:589-590.

10. Kunin,C. M. 1972.Detection, prevention, and manage-ment of urinary tract infections, p. 1-99. Lea and Febiger, Philadelphia.

11. Levin, G. V.,E. Usdin,and A. R.Slonim. 1968.Rapid detection ofmicroorganismsinaerospacewater sys-tems. Aerosp.Med. 39:14.

12. Merritt,D. A., and J. P.Sanford. 1958. Sterile-voided urineculture. J. Lab. Clin. Med.52:473.

13. Rasmussen, H., and R. Nielsen. 1968. An improved analysisofadenosinetriphosphate bytheluciferase method. Acta Chem. Scand. 22:1745-1756.

14. Stanley, P. E., and S. G. Williams. 1969. Use ofthe liquid scintillation spectrometer for determining adenosine triphosphate by the luciferase enzyme. Anal. Biochem.29:381-392.

15. Thore, A., S. Ans6hn, A. Lundin, and S. Bergman. 1975. Detection ofbacteruria by luciferase assayof adenosinetriphosphate. J.Clin.Microbiol. 1:1-8. 16. VanDyke, K., R. Stitzel, T. McClellan,and C.

Szust-kiewicz. 1969.Anautomated procedure forthe sensi-tiveandspecificdeterminationofATP.Clin. Chem. 15:3-14.

on February 7, 2020 by guest

http://jcm.asm.org/