Vol. 55, No. 10

Microbial Growth and Accumulation in IndustrialMetal-Working Fluids

INGER MATTSBY-BALTZER,l* MICHAEL SANDIN,1 BRITTA AHLSTROM,1 STIG ALLENMARK,1MARGARETA EDEBO,1 ENEVOLD FALSEN,1 KARSTEN PEDERSEN,2 NILS RODIN,

RICHARD A. THOMPSON,' AND LARS EDEBO'

Department of Clinical Bacteriology' and Department of Marine Microbiology,2University of Goteborg, and Goteborg SKF Sweden,3 Goteborg, Sweden

Received 7 April 1989/Accepted 11 July 1989

The dynamics of microbial growth in metal-working fluids (MWF) and the effect of the addition of biocideswere studied in large fluid systems, in this case, one central tank which holds 150 m3. In this system,populations of Pseudomonas pseudoalcaligenes (>108 CFU/ml) were sustained for a year, although largequantities of biocides were added. Quantitation of 3-OH lauric acid, a marker for many Pseudomonas spp., bygas chromatography indicated that the bacterial biomass exceeded the viable counts by approximately 15 times.Fungi were grown on several occasions, the dominating genera being Fusarium and Candida. Soon after the oldMWF was removed and the tank was provided with fresh MWF, which consisted of an emulsion of mineral oilin water, there was a massive growth of P. pseudoalcaligenes that reached levels of >108 bacteria per ml.Initially, only low concentrations of other species were found for some weeks. After this period, differententerobacteria and other gram-negative rods often appeared at high concentrations (107 and 108 bacteria per

ml, respectively). Bacteria identified as P. pseudoalcaligenes showed great variation with respect to colonymorphology and a certain heterogeneity with respect to biochemical characteristics. Certain bacterial speciesgrew as microcolonies on metal strips immersed in the circulating MWF, but P. pseudoalcaligenes was notrecovered from this habitat. The total bacterial count in the air surrounding the machines in the metal-workingshop showed an inverse relation to increasing distance from the machine. The concentration of bacteria in theair varied because of the number of machines in use, temperature, and humidity. Peak values of more than 105CFU/m3 of air were recorded. The workshop data combined with experimental studies indicated that thebiocide concentrations employed in the MWF were too low to prevent microbial growth of Pseudomonasspecies, in particular. Stable growth of Pseudomonas spp. facilitated the establishment of other bacteria, suchas enterobacteria. New strategies are in demand to prevent microbes from growing in MWF.

Large quantities of metal-working fluids (MWF) are usedin industry for cooling and lubrication during metal-workingprocesses such as turning, grinding, and cutting. The metal-working machines in a shop are often supported by MWFfrom a central tank system. Water-based MWF are, how-ever, susceptible to microbial degradation, which causesdeterioration of the fluid. Slime is produced, and malodormay arise. In order to control the microbial growth, biocidesare added. The presence of microbes as well as biocidescreates health problems.The most common genus cultured from MWF is Pseudo-

monas (16, 20). Potentially pathogenic species such asPseudomonas aeruginosa and Klebsiella pneumoniae arefrequently found (20). However, infections in the respiratorytract have not been reported to be more frequent in peopleexposed to MWF (10, 12). Skin, nose, and throat irritationsare the most commonly noted complaints (11). These prob-lems may be caused not only by microbial products but alsoby other components, such as biocides, in the MWF (13).At high concentrations, biocides are toxic for humans, and

at somewhat lower levels, they may be ineffective, particu-larly in heavily infected fluids. Thus, the range of usefulconcentrations is narrow, since the toxicity of the commonlyused biocides is usually nonselective. In order to developtechniques for controlling microbial growth, more informa-tion is needed about the dynamics of microbial growth in theMWF, particularly in relation to the addition of biocides.

* Corresponding author.

In the present investigation, the objective was to study thedynamics of microbial growth in the MWF of large centraltank systems in use and to follow the effects of the additionof biocides. The microbial growth was monitored for 1 yearin two types of MWF, a water-based mineral oil emulsionand a synthetic solution. Viable organisms, as well as bacte-rial mass, were measured in the MWF, and sessile bacteriawere estimated. Also, the concentration of airborne bacteriaaround the metal-working machines was determined.

MATERIALS AND METHODS

MWF. Samples ofMWF were collected once a week for 1year from two large central tank systems at a metal-workingfactory in Goteborg, Sweden. In one tank, the C tank,synthetic MWF (SF1 and SF2) were employed. Such MWFare composed of water-soluble organic substances dissolvedin water (working dilution, 2 to 2.5%). Water was suppliedfrom a nearby creek. Because of malfunctions, the systemwas emptied and cleaned, and fresh MWF was supplied inweeks 6 (SF1), 16 (SF2), and 38 (SF2) (Fig. 1). The volumeof the tank was 45 m3, and the fluid was recirculated 4 timesper h. The temperature of the fluid was kept at 25°C.The other MWF system, the D tank, contained a mineral

oil emulsion in water (OW 1; 4%). Water for mixing theMWF was supplied from the city water system. The D tankhad been cleaned and filled with OW 1 7 weeks before thestart of the investigation. In week 28, half of the tank volumewas replaced by fresh OW 1, and in week 37, the tank was

completely emptied, cleaned, and refilled with fresh OW 1.

2681

APPLIED AND ENVIRONMENTAL MICROBIOLOGY, OCt. 1989, p. 2681-26890099-2240/89/102681-09$02.00/0Copyright © 1989, American Society for Microbiology

on March 20, 2020 by guest

http://aem.asm

.org/D

ownloaded from

2682 MATTSBY-BALTZER ETAL.APLENIO.McBo.

G PG

i 44 4P C p

10 20 25" 35 40 45 50

Week, 1984

FIG. 1. Microbial growth in the C tank system containing synthetic MWF (SF1 and SF2) and monitored for 1 year. Chiorocresol (C),

orthophenylphenol (P), and the formaldehyde-liberating compound Grotan BK (G) were added as indicated. Symbols (lower panel): 0, total

count; * (and dotted line), P. pseudoalcaligenes; Q, gram-negative, oxidase-positive rods not tested by API 20 NE, probably P.

pseudoalcaligenes (also indicated by the dotted line) or gram-negative, oxidase-positive bacteria other than P. pseudoalcaligenes; U,

Corynebacterium spp.; A, gram-labile diplococci or streptococci; V, Flavobacterium spp.; 0, enterobacteria. Gram-negative, oxidase-

positive bacteria identified as not being P. pseudoalcaligenes and found in lower numbers than P. pseudoalcaligenes have not been included

in the figure.

Beginning in week 46, the daily loss of MWF was replacedwith another mineral oil emulsion (OW 2). In the last week of

that year, the tank was emptied and cleaned thoroughly, and

the same type of fluid that was used in the C tank (SF) was

added to the D tank (Fig. 2). This tank held 150 m' of MWF,

which was recirculated 3 times per h. The temperature was

not regulated, but it held at approximately 250C.

C C

I I i I I-VW

voco.a9i

0 420

V

9

I a

p%J 71cpb0-j

C3.

yeast

The biocides 3-chloro-2-cresol and orthophenylphenolwere present in the stock fluids of the mineral oil emulsions

(OW 1 and OW 2) and the synthetic solutions (SF1 and SF2).

Chlorocresol was maintained at a concentration of 700 to

1,500 ppm (700 to 1,500 pRgIml) and orthophenylphenol was

maintained at 100 to 250 ppm in the working dilutions of the

MWF. The presence of high numbers of enterobacteria was

c c

owl owl *Olwl 0W2 SF2

PPs~~~~Itotal total

Enterobacteraggi.AclensfcAcinetobacter -4f,~

enterobacteri tAerococcus viridans A //Cr ebacterium

10 I5 20 25

Week. 1984

35 4.0 45 so 5

Week, 1985

10 is

FIG. 2. Microbial growth during 1 year in the D tank system, which contained mineral oil emulsion (OW 1 and OW 2) or synthetic MWF

(SF2). Orthophenylphenol (P), chlorocresol (C), and formaldehyde-liberating biocide Glokill (G) were added as indicated.

C P10000.

I-i

cm i00~-J

5? '20

1:]

U-

40I0-J

8.

7-

6

5

4

APPL. ENVIRON. MICROBIOL.

moui /"-N

5

16

Maui ii

on March 20, 2020 by guest

http://aem.asm

.org/D

ownloaded from

MICROBIAL GROWTH AND ACCUMULATION IN INDUSTRIAL MWF

regarded as especially deleterious to the MWF because ofthe ability of the enterobacteria to lower the pH. A low pHincreased the risk of corrosion of the metals. Therefore, highnumbers of enterobacteria as measured with a dipslide(Hygicult E; Orion Diagnostica, Espoo, Finland), a fall inpH, malodor, or a combination of these factors were anindication for adding a third biocide, the formaldehyde-liberating biocide N,N',N"-tris-(,-hydroxyethyl)-hexahy-drotriazin (trade name, Glokill or Grotan BK) at a concen-tration of 500 ppm.The MWF were circulated during weekends and holidays.

The pH was kept constant within the range of 8 to 9 by theaddition of amines. The daily loss of fluid was replaced oncea week or when necessary with fresh MWF diluted with citywater. In addition, all other coolant-lubrication systems atthe factory (25 tanks) were analyzed once for microbialgrowth.

Quantification and identification of microorganisms. Sam-ples of the MWF were diluted and seeded onto agar plateswith a spiral plater (model D; Spiral Systems, Inc., Cincin-nati, Ohio). The total bacterial viable counts were estimatedas CFU on blood and brain heart infusion agar plates afteraerobic incubation at 30°C for 2 days. Blood agar plates werealso incubated anaerobically for 5 to 7 days at 37°C. Acid-producing glucose-fermenting bacteria were counted on aplate containing glucose (1%), yeast extract (0.1%), nutrientagar (3.9%), and a pH indicator. The latter plate was usedonly in the beginning of the investigation, since it was foundthat anaerobic incubation of blood agar plates yielded bettergrowth of fermenting bacteria by suppressing the growth ofPseudomonas spp. All morphologically different colonies onthis plate were further identified.For the cultivation of fungi, undiluted samples were incu-

bated on Sabouraud plates with or without antibiotics (chlor-amphenicol [50 mg/liter] and gentamicin [100 mg/liter]).Biochemical and immunological tests were used for spe-

cies identification of the bacteria (6, 7). Gram-negative andoxidase-positive bacterial strains were screened by API 20NE (API System S.A., Montalieu, Vercieu, France). Aspecial test scheme was used for identifying gram-positivebacteria (7).

Bacterial growth in MWF in vitro. Pseudomonas pseudoal-caligenes (strain 1, Culture Collection, University of Gote-borg [CCUG] 15284; strain 2, CCUG 15286), Corynebacte-rium spp., (CCUG 15429), and Klebsiella oxytoca (CCUG15788), which were isolated from the MWF, were used in theexperiments. In addition, Salmonella lomita (CCUG 12637),Salmonella irumu (CCUG 12638), Shigella sonnei, Staphy-lococcus aureus, and Candida albicans were obtained fromour laboratory. The incubations were performed in 50-mlflasks on a shaker at room temperature or at 30°C. When OW1 (4.5%) was incubated with P. pseudoalcaligenes for vari-ous days prior to being inoculated with other bacterialstrains, the P. pseudoalcaligenes count had reached at least108 CFU/ml. The MWF used in the experiments containedthe biocides chlorocresol and orthophenylphenol, alreadyadded by the manufacturer to the stock solutions. A biocide-free solution of OW 1 was used on two occasions, as statedin the legends to Fig. 3 and 4.Adherence of bacteria to steel or glass plates in contact with

MWF. The biofilm reactor used in our study has beendescribed in detail by Pedersen et al. (15). In short, thereactor consisted of a rectangular box with a sealed lidcontaining a test pile and two diffusors. The test pile was

composed of a stock of microscopic glass slides or metallicplates of the same size. A laminar flow was created over the

e6-

5.

4 -

3-1

EI.

a0

Rps

21N<1.3o o

r . .o 2 4 6

Days of incubation

FIG. 3. Growth of K. oxytoca in OW 1, which had been prein-cubated with P. pseudoalcaligenes for 2, 4, or 6 days (d) prior to theinoculation of K. oxytoca. P.ps, P. pseudoalcaligenes. No biocidewas present in the MWF.

surfaces by the circulating MWF with the aid of the diffu-sors. After appropriate incubation periods in the MWF, thetest surfaces were rinsed by a water stream for 15 s in orderto remove unattached bacteria from the surfaces. Micro-scopic analysis was performed with epifluorescence micros-copy after the surfaces were stained.Chemical determination of Pseudomonas spp. by gas chro-

matographic analysis of 3-hydroxylauric acid. The MWFsample (1.5 ml) was diluted with 9.5 ml 1-butanol andcentrifuged at 2,500 x g and 5°C for 30 min. The supernatantwas carefully withdrawn and discarded. The pellet was

9-a

8-

6-eIL,

0 -

- 4.0

3

2

'1.3

Grotan BKb9

8

7

6

5

4

3

2

<1.3u-

70 2 7

P.ps +Grotan BK

K~~~0

0

U

0 7

O 2 7

Days of incubationFIG. 4. (a) The effect of the biocide Grotan BK on the growth of

subsequent inoculated S. lomita (-), S. irumu (-), Shigella sonnei(O), or C. albicans (0) in OW 1. (b) The MWF was preincubatedwith P. pseudoalcaligenes (A) prior to the addition of Grotan BKand inoculation of the bacteria. No biocides were present in thestock solution of OW 1. Other symbols are as in panel a.

VOL. 55, 1989 2683

I

on March 20, 2020 by guest

http://aem.asm

.org/D

ownloaded from

2684 MATTSBY-BALTZER ET AL.

suspended in 2 ml of 6 M HCl together with 1 ml of theinternal standard solution (2-hydroxymyristic acid in meth-anol [1 mg/ml]). The sample was then hydrolyzed at 100°Cfor 1 h. The fatty acids were extracted twice from thesolution with 2 ml of diethyl ether in hexane (1:1). Thecombined organic phases were evaporated. The methylatedfatty acids were redissolved in 0.5 ml of 0.1 M trimethyl-amine (purity, >98%; Fluka) in benzene and 10 [lI ofheptafluorobutyric anhydride (purity, >99%; Fluka). Thederivatization was complete after 15 min at 50°C. Excessreagent was removed by washing the benzene solution with0.5 ml of 1 M phosphate buffer, pH 6.0. After the solutionwas centrifuged, 3 RI was injected on the gas chromato-graph.The gas chromatograph used was a Sigma 3B (The Perkin-

Elmer Corp., Norwalk, Conn.) equipped with a 63Ni (10mCi) electron capture detector. A fused silica column (0.22mm by 25 m) coated with BP-I (SGE, Victoria, Australia)was used. Runs were performed with a linear gradient of10°C/min from 180 to 280°C. A 0.2-,u sample of the benzenesolution was injected with a splitless injection.

Quantification of total bacterial mass was done by per-forming the same procedure with known quantities of P.pseudoalcaligenes. The ratio of the 3-hydroxylauric acidpeak area to that of 2-hydroxymyristic acid was used forquantification. Washing of the mineral oil-based MWF with1-butanol was necessary to obtain usable chromatograms. Aratio of 1.5 ml of MWF to 9.5 ml of 1-butanol was found togive an optimal cleanup without any measurable loss of thehydroxy-fatty acid.

Airborne bacteria. Measurements of CFU per cubic meterof air were performed in the machine shop, which wassupplied with MWF from the D tank with the Biotest Reutercentrifugal air sampler (Biotest Diagnostics, Frankfurt amMain, Federal Republic of Germany) and with the Andersensampler (1). With the Andersen sampler, airborne particlesare size differentiated into six groups. The particles in four ofthese groups are small enough to be inhaled. The group withthe smallest particles (particles of 0.65 to 1.1 ,um) reaches allparts of the lung (1). The air sampling was performed onseveral occasions at approximately the same sites and thesame times during the day. The amount of MWF aerosolgenerated by the metal-working machines in the machineshop varied because of the number of machines that were inoperation at that particular time. Also, the temperature andhumidity varied on the different sampling occasions. One tothree measurements were done at each site.

RESULTS

The total viable counts, the concentration of the dominat-ing species, and the concentration of enterobacteria areshown for the D tank containing mineral oil emulsion (OW 1and OW 2) (Fig. 2). High total viable counts were recordedthroughout the year, ranging mostly from 108 to 109CFU/ml.Cleaning and refilling of the tank with fresh MWF at week 37only modestly decreased the total count for a short time. Atweek 46, the daily loss of fluid was replaced with anotherMWF, also based on mineral oil (OW 2). Throughout theperiod, the dominating species in the MWF was P. pseudoal-caligenes (Fig. 2). This bacterium usually made up 80 to 90%of the total viable counts. The morphological appearance ofthe P. pseudoalcaligenes colonies often varied markedlywithin one sample. Exchange of oil emulsion MWF had onlya small effect on the total counts but reduced the enterobac-terial counts.

Gram-positive cocci, identified as Aerococcus viridans,were found in low counts during weeks 8 to 15. Thereafter,enterobacteria were frequently observed in numbers of 104to 107 CFU/ml. As long as P. pseudoalcaligenes was thedominating species, the pH remained fairly stable. As en-terobacteria appeared, there was a tendency toward a dropin pH. During weeks 16 to 21, Enterobacter agglomeransappeared in high numbers (-107/ml). In later specimens, thespecies representing the highest enterobacterial count variedfrom sample to sample. A diphtheroid rod, Corynebacteriumspecies, was isolated on three occasions (week 23, 6.8 x108/ml, week 34, 2 x 106/ml, and week 39, 1.6 x 107/ml).Microorganisms isolated from the MWF of the D tank wereas follows. In mineral oil emulsion in water, gram-negativespecies were P. pseudoalcaligenes, Pseudomonas stutzeri,Shewanella putrefaciens (earlier called Alteromonas putre-faciens), Aeromonas spp., E. agglomerans, K. pneumoniae,K. oxytoca, Proteus vulgaris, Escherichia coli, Citrobacterdiversus, Citrobacter freundii, Serratia spp., and Mor-ganella morganii and gram-positive species were Strepto-coccus spp., A. viridans, and Corynebacterium spp. Regard-ing fungi, yeast (Candida species) was recovered during fourperiods (weeks 19 to 24, 32 to 36, 41 to 47, and 51). Mold,mostly Fusarium species, was found on three occasions,coinciding with the appearance of yeast.The MWF of the D tank was changed during the following

year to a synthetic fluid (SF2), the same fluid as the one usedin tank C for 8 months. In the first 5 weeks, bacterial growth(105 CFU/ml) was observed once. In week 6, 5 x 106 p.pseudoalcaligenes were recovered and then the bacterialcount increased to 107 to 108 CFU/ml and remained high. P.pseudoalcaligenes did not dominate to the same degree inthis MWF as in the mineral oil emulsion. Several otherspecies, such as Acinetobacter spp., Alcaligenes faecalis,M. morganii, and Corynebacterium spp., were found. En-terobacteria were found only once during the whole obser-vation period of 14 weeks. Peak recovery of Fusarium spp.coincided with the highest bacterial counts at weeks 8 and11. The addition of chlorocresol in week 10 reduced thebacterial as well as the Fusarium spp. counts, but in the nextweek, the counts had increased again.The microorganisms found in tank C, which contained

synthetic MWF (SF1 or SF2) varied more (Fig. 1). The totalnumber of bacteria fluctuated throughout the year (105 to1010 CFU/ml), as did the bacterial species, with P. pseudoal-caligenes and Corynebacterium spp. dominating. OtherPseudomonas species, such as P. putida, P. fluorescens,Flavobacterium odoratum, and Acinetobacter spp., werealso observed. Gram-positive species in the synthetic fluidwere, in addition to Corynebacterium spp., Bacillus, Staph-ylococcus, and Streptococcus spp. During weeks 6 to 20, P.pseudoalcaligenes or other gram-negative oxidase-positivenonfermenting bacteria dominated. Beginning in week 21,Corynebacterium spp. alone or together with gram-negativeoxidase-positive nonfermenting bacteria predominated, withthe exception of four occasions (weeks 31, 36, 41, and 50),when the latter group of bacteria showed the highest counts.Nearly no enterobacteria were found at the dilution tested,i.e., <2 x 103 CFU/ml. Cleaning and refilling of fresh MWFin weeks 16 and 38 reduced the bacterial count to ca. 105CFU/ml. Bacterial regrowth was rapid and a 10- to 100-foldincrease was seen after 1 week. Mold (Fusarium spp.) grewin five periods, weeks 7 to 11, 22 to 25, 32 to 33, 35 to 41, and46 to 49. Yeast was found in week 35 only.At the same factory, 25 different tanks were tested on one

occasion. Mineral oil emulsion was the coolant in 12 of the

APPL. ENVIRON. MICROBIOL.

on March 20, 2020 by guest

http://aem.asm

.org/D

ownloaded from

VOL. 55, 1989 MICROBIAL GROWTH AND ACCUMULATION IN INDUSTRIAL MWF 2685

9

E a-

o 7

Q2o

5

0)

0

9

10

9

! ; ~~~Dtank7~~~~~~~~~~~~~~~~~~~~~

r 10 Is 20 25 30 35 40 45 50

Week. 1984

FIG. 5. Estimation of the biomass of Pseudomonas spp. (0)present in the MWF of the D and C tanks by gas chromatographicquantification of 3-hydroxylauric acid. Total viable counts (1) are

also shown. Broken lines indicate the replacement of old MWF withfresh mixtures.

tanks, and synthetic fluid was the coolant in 13 of the tanks.All of the 20 tanks that yielded growth produced dominatinggrowth of oxidase-positive gram-negative bacteria. In one

culture from a tank that contained synthetic fluid, a diphthe-roid rod dominated together with the gram-negative isolate.Two samples from tanks with mineral oil emulsion and fivefrom tanks with synthetic fluid contained fungi.

Estimation of the bacterial biomass in the MWF by gas

chromatographic analysis of 3-hydroxylauric acid. Quantita-tion of 3-hydroxylauric acid, a marker for Pseudomonas spp.

and related gram-negative bacteria, was performed on theMWF samples (Fig. 5). The amounts of the 3-hydroxy-fattyacid in each sample corresponded to an average of ca. 10 to15 times more bacterial cells in both tanks than the viablecounts showed, excluding the values after the exchange ofthe MWF in the C tank. On all occasions except two, the gaschromatographic analyses showed higher values rangingfrom 3 to ca. 600 times. The biggest differences were seenafter the exchange of the MWF in the C tank, when theviable counts had decreased 400-fold while the 3-hydroxy-lauric acid concentration had decreased less than 10-fold.

Effects of biocides. The addition of biocides of the D tankin quantities calculated to be bactericidal and fungicidalusually did not greatly affect the total counts or the counts ofthe individual microorganisms (Fig. 2). Analysis of the

TABLE 1. Total number of bacteria in the hall that containedmetal-working machines supplied with MWF from the D tanka

Total number of bacteria (CFU/m3)Source of air found during week

10 15 19 23

0.5 m from MWMb 13,000 39,000 200,000 >C

1.5 m from MWM 4,000 16,000 150,000 >3 m from MWM 2,000 1,000 40,000 >Packaging area (10 m 230 20,000 >85,000from MWM)

Storing area 6,600 5,300a Sampling was performed with a Reuter centrifugal sampler.b MWM, Metal-working machine.c >, Bacterial count too high for proper measurement.

L-J

8-

7-

6-

C hlIoroc resol

-8

-8 -1 5 10 15 20 40 64Hours

FIG. 6. Viable counts of bacteria in the C tank after addition ofthe biocide chlorocresol, monitored for 3 days by aerobic (0) andanaerobic (0) cultures from the MWF.

chlorocresol concentrations during weeks 40 to 50 showedgreat variations within a range of 33 to 1,390 ppm. In the Ctank, reduced levels of bacteria were seen after the additionof chlorocresol or Grotan BK (weeks 11, 19, 42, and 47) (Fig.1). The decreases were, however, not markedly differentfrom the spontaneous fluctuations.The short-range effects of chlorocresol on the bacterial

counts of the C tank were studied by repeated sampling for3 days after the addition of the biocide (Fig. 6). No signifi-cant changes in the aerobic or anaerobic bacterial countswere observed.

Airborne bacteria. The number of viable bacteria per cubicmeter of air surrounding the metal-working machines wasmeasured on four occasions with the Reuter centrifugalsampler (Table 1). The bacterial count decreased with in-creasing distance from the machine. On the last samplingoccasion, week 23, when the highest bacterial counts wererecovered, the highest room temperature was registered(28°C). When the presence of P. pseudoalcaligenes and A.viridans in the aerosol was distinguished (week 15) (Table 2),the fraction of A. viridans increased with increasing distancefrom the machine.At week 19, the airborne particles were also analyzed for

size distribution with an Andersen sampler. The bacterialcount per cubic meter on this occasion was so high that onlyparticles equal to or less than 1.1 pLm were measurable, i.e.,

TABLE 2. Estimation of the distribution of P. pseudoalcaligenesand A. viridans in the air of the machine hall

with the D tank systema

U/M3~~~Proportion (o

Source CFU/m3of air P. pseudo- .

alcaligenes

0.5 m from MWMb 3.9 x 104 90 11.5 m from MWM 1.6 x 104 88 03 m from MWM 1.0 x 103 78 13Packaging area (10 m 2.3 x 102 28 43from MWM)

MWF 1.0 x 108c 81 0.2

a Sampling was performed during week 15 with a Reuter centrifugalsampler.

b MWM, Metal-working machine.c CFU per ml of MWF. Sample collected the day before analysis of airborne

bacteria.

on March 20, 2020 by guest

http://aem.asm

.org/D

ownloaded from

2686 MATTSBY-BALTZER ET AL.

FIG. 7. Colonies of bacteria on steel plates exposed to MWF of the C tank system in the biofilm reactor. The plates were exposed to thecirculating fluid for 1 (a) and 2 (b) weeks. Magnification, x 1,600.

particles small enough to reach the deepest part of the lung,the alveoli (Tables 1 and 3). Approximately a 100-folddecrease in viable bacteria was seen when counts from thepackaging area were compared with counts from an area 0.5m from the machine.

Bacterial growth on surfaces in contact with the MWF.Bacterial growth of 106 cells per cm2 was observed asmicrocolonies on the surface of steel or glass plates whichhad been exposed to circulating MWF from the C or D tanksystems in the biofilm reactor during a 2- or 3-week period(Fig. 7). The colony character of growth indicated multipli-cation in the adherent state. The main growth on the steeland glass plates in contact with the MWF of the D tank wasvarious enterobacteria and Shewanella putrefaciens (Table4). It was observed that in the extensive biochemical testsused for identification of isolates, nearly all characteristics ofsubsequent isolates of enterobacteria were identical (57 outof 59 biochemical markers). In one such example, 58 out of59 biochemical tests were indistinguishable, whereas re-

peated dulcitol fermentation tests distinguished one K. oxy-toca isolate from an E. agglomerans isolate (Table 4). Also,the K. pneumoniae isolate was very similar. P. pseudoalca-ligenes, which dominated in the MWF, was not isolated fromthe steel or glass surfaces. No enterobacteria were grown

TABLE 3. Total bacterial counts of the air in the machine shopwith the D tank systema

Bacterial countsSource (CFU/m3)c

0.5 m from MWMb ....................................... 40,0001.5 m from MWM....................................... 25,0003m fromMWM........................................_ dPackaging area (10 m from MWM) ............. .......... 400Storing area ....................................... 300

a Sampling was performed in week 19 by the Andersen sampler.b MWM, Metal-working machine.c Particle size was from 0.65 to 1.1 ,um.d _, Proteus spp. swarmed over the plate.

APPL. ENVIRON. MICROBIOL.

on March 20, 2020 by guest

http://aem.asm

.org/D

ownloaded from

MICROBIAL GROWTH AND ACCUMULATION IN INDUSTRIAL MWF 2687

TABLE 4. Microbial growth on steel or glass plates in the biofilmreactor supplied by MWF from the D or C tank system

Time of Organisms found in:sampling D tank C tank

Week 23 K. oxytocaa P. putidaE. colia Corynebacterium spp.Shewanella putre- Streptococcus spp.faciens

Week 24 E. agglomeransa Pseudomonas spp.E. colia Flavobacterium odoratumProteus vulgarisa Flavobacterium spp.Shewanella putre- Acinetobacter calcoaceticusfaciens Streptococcus spp.

Staphylococcus spp.

Week 25 K. pneumoniaea P. putidaShewanella putre- F. odoratumfaciens Flavobacterium spp.

Streptococcus spp.

a Enterobacteria.

from the plates in the C tank (Table 4). This result was inagreement with the presence of bacteria in the fluid. Pseu-domonas species other than P. pseudoalcaligenes and otheroxidase-positive gram-negative and gram-positive bacteriawere found.

In vitro studies on survival and growth of bacteria in theMWF. The biocide-containing synthetic MWF, SF1 andSF2, were capable of killing two isolates of P. pseudoalca-ligenes from the D tank (Fig. 8a and b). In contrast, thesetwo bacterial strains survived in the mineral oil emulsion,OW 1 (Fig. 8a and b). A Corynebacterium isolate from the Ctank was, however, resistant to all MWF (Fig. 8b).The importance of the growth of P. pseudoalcaligenes for

the establishment of other bacteria in the mineral oil emul-sion is shown in Fig. 3. Only when growth of P. pseudoal-caligenes had preceded the inoculation with K. oxytoca didthe growth of the latter take place. The bacterial count of P.pseudoalcaligenes, approximately 108 CFU/ml, was stableduring the days after the inoculation of K. oxytoca.When a formaldehyde-liberating biocide, Grotan BK (200

ppm), was added to the mineral oil emulsion, none of theseparately incubated enterobacteria (S. lomita, S. irumu, or

Shigella sonnei) survived in the fluid (Fig. 4a). If P.pseudoalcaligenes was preincubated with the MWF for 6days prior to the addition of Grotan BK and the inoculation

9 a boV"tSF2N. - i SF1 2%

S SF2 2%4%OWI 2%

7'%SF02% 62X 4%

2%

SF2 2% 4. 2%

3 3- ~~~~~~~~~~S2 2%

2 SF2 4%

SF2 4% sF1 2%

0 6 24, 48 72 6 24 48 72

Incubation tim (hours) Iocs*tio tim (1 oisI

FIG. 8. P. pseudoalcaligenes 1 (a) or 2 (0) or Corynebacteriumspp. (0) (b) was incubated with fresh MWF of OW 1 or syntheticsolutions (SF1 and SF2). The MWF were diluted to the recom-

mended working dilutions before use.

of bacteria, the Salmonella strains survived (Fig. 4b). Thesurvival of Shigella strains was also increased. The yeast, C.albicans, grew well under both conditions (Fig. 4a and b).Staphylococcus aureus did not survive under any condi-tions.

DISCUSSION

The predominating bacteria found in the mineral oil emul-sion MWF have been identified as P. pseudoalcaligenes.Pseudomonas species are known for their capability ofutilizing a variety of hydrocarbons. Plasmids have beenshown to carry biodegradation abilities (3). Pseudomonasspecies have frequently been found in MWF consisting ofpetroleum emulsions (5, 16, 20, 22). P. pseudoalcaligenes isdifficult to distinguish from Pseudomonas oleovorans (7),which has been reported to be the most common finding inmineral oil emulsions (20). In the synthetic solutions, ahigher degree of variation ofPseudomonas species as well asof other nonfermenting gram-negative species was observed.However, P. pseudoalcaligenes together with Corynebacte-rium spp. were still the most frequent isolates. The increas-ing concentration of Corynebacterium spp. in the D tanksystem after the change from mineral oil emulsion to syn-thetic MWF indicated better growth conditions for thesebacteria in this type of fluid. According to Bennett (2),synthetic MWF are less resistant than mineral oil emulsionsto deterioration by coliform bacteria. In the C tank systemwith the synthetic MWF, however, enterobacteria werefound only once and at a modest concentration (104 CFU/ml), which indicates that the synthetic MWF has a greaterresistance than mineral oil emulsion to infection by entero-bacteria. This was also supported by the finding of entero-bacteria on only one occasion in the D tank system after thechange to the same synthetic MWF. A possible explanationfor the resistance may be that the chemical composition ofthe synthetic fluid inhibited the growth of enterobacteria, inparticular. Also, it is possible that coliform antagonists were

among the bacterial species that developed in the syntheticMWF. Breakdown products from Pseudomonas spp. were

present in both MWFs and may have served as nutrients forenterobacteria as well as other bacteria. However, it can notbe excluded that the breakdown products from the hydro-carbon in the mineral oil-based emulsion were importantgrowth factors for enterobacteria.The development of microbial growth in the D tank

system, which started with a stable growth of P. pseudoal-caligenes and after ca. 4 months continued with growth ofenterobacteria, may reflect a common sequence in mineraloil emulsions. The persistent growth of P. pseudoalcaligenesmay not only provide the fluid with a range of nutrients butalso neutralize the biocides present in the fluid and in thisway pave the way for the growth of enterobacteria, partic-ularly in an anaerobic environment. The in vitro experiments(Fig. 3, 4, and 8) supported this hypothesis, since growth ofP. pseudoalcaligenes was a prerequisite for the survival andmultiplication of several enterobacteria in mineral oil emul-sions with biocides.As judged from the measurements of 3-hydroxylauric acid

in MWF from both tank systems (Fig. 5a and b), at least 10times more bacterial biomass was present in the MWF thanthe viable counts showed. When the synthetic MWF was

replaced by a fresh load, the viable counts dropped close to1,000-fold, while the 3-hydroxylauric acid decrease was ca.

10-fold. These results indicate that 10% of the bacterial mass

in the MWF was not eliminated in the replacement of MWF.

VOL. 55, 1989

on March 20, 2020 by guest

http://aem.asm

.org/D

ownloaded from

2688 MATTSBY-BALTZER ET AL.

The addition of fresh MWF containing biocides killed mostof the remaining bacteria, which resulted in a higher ratio ofbacterial mass (3-hydroxylauric acid) to viable counts.Within a few weeks, however, the surviving bacteria hadmultiplied to levels even higher than those before the re-

placement of the MWF, which indicated favorable growthconditions in the fresh MWF and an underestimation of thebacterial biomass as found with the 3-hydroxylauric acidanalysis.An important source of recontamination of the MWF is

bacteria remaining in the tube connections and adhering tothe walls of the MWF system (Table 4). These adheringbacteria were usually also recovered from the MWF. How-ever, P. pseudoalcaligenes was never cultivated from theplates of the biofilm reactor, although it was the dominatingspecies in the MWF. Ruseska et al. (18) showed that bacteriacolonizing the walls were more resistant against biocides inthe MWF than bacteria suspended in the MWF, probablybecause the bacteria colonizing the walls were protected inthe microbial mass. Low efficiency of a biocide may also bedue to the development of bacterial resistance and break-down. Pseudomonas spp. are known for carrying plasmidsrelated to the breakdown of halogenated phenols (3).

Yeast (Candida species) and mold (Fusarium, Cephalo-sporium, and Cladosporium spp.) were found in the tankwith mineral oil emulsion, which is in agreement with earlierreports (16, 17). In the synthetic MWF, mold predominatedover yeast. Fungi were grown only from MWF samples,which simultaneously contained bacteria. The most frequentmold in both types ofMWF was Fusarium spp. Mycotoxinswith carcinogenic activity are produced by Fusarium spp.

(4). However, no health problems related to fungal growth inMWF have been established. Molds may cause serioustechnical problems in the MWF by growing in clumps whichmay block the flow in small pipes, filters, and orifices.Growth on dipslides or plates in sparse numbers may notreflect the danger of such clumps in the fluid.From our studies, we conclude that the biocide concen-

trations in MWF are often too low to prevent microbialgrowth, especially of Pseudomonas species. When biocidesare added to already infected fluids, it must be rememberedthat the biocides are degraded or otherwise neutralized bythe microbial biomass (2, 9). Thus, the limited effect of thebiocides and the fact that MWF systems often are under-dosed because of the toxicity of the biocides call for alter-native strategies in controlling the microbial growth. Indeed,a first step should be to monitor the biocide concentrations inthe MWF in order to keep the concentration in the intervalwhich allows antimicrobial action without toxic side effects.Such measures have already been taken at the factory whichwe studied.When the metal-working machines were running, a micro-

bial aerosol was generated in the machine hall. The bacterialspecies found in the air were the same as that found in thefluid. The differences in bacterial counts between the sam-

pling occasions could be explained by differences in thenumber of machines running, as well as variations in thetemperature and humidity. The viability of airborne gram-

negative bacteria is dependent on the temperature, humidity,and suspending fluid (21). The decreases in viable bacterialunits (CFU per cubic meter) with an increase in distancefrom the machine from 0.5 to 3 m were from 35,000 to 780CFU/m3 for P. pseudoalcaligenes and from 390 to 130CFU/m3 for the A. viridans (Table 2). Since the A. viridansgrows in aggregates, its lower rate of decrease in viablecounts may be a consequence of multiple-hit inactivation

kinetics. A source of A. viridans other than the MWF is notlikely, since the relationship between A. viridans and P.pseudoalcaligenes in the MWF was most similar to that inthe air closest to the metal-working machine. Thus, P.pseudoalcaligenes in the workshop aerosol was subjected torapid killing, since gram-negative bacteria are more vulner-able in the air than gram-positive bacteria (8, 19, 21).Consequently, viable counts of the aerosol grossly underes-timated the microbial mass inhaled by the metal workers, inparticular, since the viable counts in the MWF representedless than 10% of the bacterial mass (Fig. 5). Analysis by theLimulus amoebocyte lysate assay (Coatest endotoxin, KabiDiagnostica, Molndal, Sweden) of air at a distance of 0.5 mfrom the machine (week 15) (Table 1) showed at least 100times more endotoxin than expected from the viable count.Thus, in most instances, the microbial mass, particularlyendotoxin (lipopolysaccharide), in the aerosol amounts to atleast 100-fold the viable counts. The airborne live bacteriawere recovered in aerosol particle sizes small enough (Table3) to reach the lung parenchyma. In a concurrent study onmetalworkers exposed to the MWF in the workshops, wefound significantly elevated levels in serum of antibodyagainst P. pseudoalcaligenes, particularly in the immuno-globulin Gl and G2 subclasses (14; I. Mattsby-Baltzer, L.Edebo, B. Jarvholm, B. Lavenius, and T. Soderstrom,manuscript in preparation). The present results demonstratethat large quantities of bacterial components have beenpresent in aerosols and available for antigenic stimulation viathe respiratory route.

ACKNOWLEDGMENTS

The excellent typing of Helene Eliasson is gratefully acknowl-edged.

This study was supported by grants from the Swedish WorkEnvironment Fund. I.M.-B. was supported by the Procordia Re-search Foundation.

LITERATURE CITED

1. Andersen, A. A. 1958. New sampler for the collection, sizing,and enumeration of viable airborne particles. J. Bacteriol.76:471-484.

2. Bennett, E. 0. 1972. The biology of metal working fluids.Lubrication Engineering 28:237-247.

3. Chakrabarty, A. M. 1976. Plasmids in Pseudomonas. Annu.Rev. Genet. 10:7-30.

4. Cole, R. J., J. W. Kirksey, H. G. Cutler, B. L. Coupnik, andJ. C. Peckham. 1973. Toxin from Fusarium moniliforms: effectson plants and animals. Science 179:1234-1236.

5. Fabian, F. W., and H. Pivnick. 1953. Growth of bacteria insoluble oil emulsions. Appl. Microbiol. 1:199-203.

6. Falsen, E. 1983. Immunodiffusion as an aid in routine identifi-cation of uncommon aerobic gram-negative bacteria, p. 447-483. In H. Ledere (ed.), gram-negative bacteria of medical andpublic health importance: taxonomy, identification, applica-tions. Institut National de la Sante et de la Recherche Medicale,Paris.

7. Falsen, E. 1988. Catalogue of strains, Culture Collection, 5th ed.University of Goteborg, Goteborg, Sweden.

8. Flynn, D., and L. J. Goldberg. 1971. Effect of relative humidityon aerosol persistence of Streptococcus salivarius. Arch. Envi-ron. Health 23:40-42.

9. Garrod, L. P. 1958. The effects of bacterial numbers on mini-mum bacteriostatic concentrations. J. Infect. Dis. 57:247-251.

10. Hill, E. C. 1983. Microbial aspects of health hazards from waterbased metal working fluids. Tribology Int. 16:136-140.

11. Hill, E. C., and T. Al-Zubaidy. 1979. Some health aspects ofinfections in oil and emulsions. Tribology Int. 12:161-164.

12. Jarvholm, B., and E. C. Hill. 1983. Bakterier i skarvatskor.

APPL. ENVIRON. MICROBIOL.

on March 20, 2020 by guest

http://aem.asm

.org/D

ownloaded from

MICROBIAL GROWTH AND ACCUMULATION IN INDUSTRIAL MWF 2689

Arbete, miljo, manniska. 2:160-163.13. Kipling, M. D. 1977. Health hazards from cutting fluids. Tribol-

ogy Intern. 10:41-46.14. Mattsby-Baltzer, I., B. Ahlstrom, L. Edebo, B. Jarvholm, and B.

Lavenius. 1988. Antibodies to Pseudomonas pseudoalcaligenesin metal workers exposed to infected metal working fluids. Int.Arch. Allergy Appl. Immunol. 88:304-311.

15. Pedersen, K., C. Holmstrom, A. K. Olsson, and A. Pedersen.1986. Statistic evaluation of the influence of species variation,culture conditions, surface metability and fluid shear on attach-ment and biofilm development of marine bacteria. Arch. Micro-biol. 145:1-8.

16. Rossmoore, H. W. 1981. Antimicrobial agents for water basedmetal working fluids. J. Occup. Med. 23:247-254.

17. Rossmoore, H. W., and G. H. Holtzman. 1974. Growth of fungiin cutting fluids. Dev. Ind. Microbiol. 15:273-280.

18. Ruseska, J., J. Robbins, and J. W. Costerton. 1982. Biocidetesting against corrosion-causing oil-field bacteria helps controlplugging. Oil Gas J. 80:253-264.

19. Strange, R. E., J. E. Benbough, P. Hambleton, and K. L. Martin.1972. Methods for the assessment of microbial populationsrecovered from enclosed aerosols. J. Gen. Microbiol. 72:117-125.

20. Tant, C. O., and E. 0. Bennett. 1956. The isolation of patho-genic bacteria from used emulsion oils. Appl. Microbiol. 4:332-338.

21. Wathes, C. M., K. Howard, and A. J. F. Webster. 1986. Thesurvival of Escherichia coli in an aerosol at air temperatures of15 and 30°C and a range of humidities. J. Hyg. 97:489-496.

22. Wort, M. D., G. I. LLoyd, and J. Schofield. 1976. Microbiolog-ical examination of six industrial soluble oil emulsion samples.Tribology Int. 9:35-37.

VOL. 55, 1989

on March 20, 2020 by guest

http://aem.asm

.org/D

ownloaded from