Vet. Res. 38 (2007) 685–696 Available online at:c© INRA, EDP Sciences, 2007 www.vetres.orgDOI: 10.1051/vetres:2007026

Original article

Staphylococcus aureus leucotoxin LukM/F’ is secretedand stimulates neutralising antibody response

in the course of intramammary infection

Pascal R*

INRA, UR1282 IASP, 37380 Nouzilly, France

(Received 8 December 2006; accepted 30 March 2007)

Abstract – Leucotoxins are regarded as virulence factors of Staphylococcus aureus, but data sup-porting their importance in infection are scarce. Experimental infections of the mammary glandsof six goats with a leucotoxin-producing strain were used to investigate in vivo production of leu-cotoxin. Leucotoxin M/F’ was monitored in milk as well as antibodies to LukM/F’ in serum andmilk. Leucotoxin antigen was detected by ELISA in milk samples of two goats. The appearance ofneutrophils in these samples showed similarity with the altered appearance of neutrophils incubatedwith purified LukM/F’ leucotoxin. Leucotoxic activity was found in the milk samples of these twogoats, in particular those of the only goat which developed gangrenous (necrotic) mastitis, and thisactivity was inhibited by antibodies to LukM/F’. Increases in specific antibody titers occurred inthe serum and milk of infected goats, and neutralizing titers increased in serum. The results indicatethat the lukM and lukF’ genes were functional in vivo, and actively transcribed and translated inthe course of mammary infection. These results suggest that LukM/F’ can interfere with neutrophildefense activities during infection, and they prompt further studies to investigate the contribution ofleucotoxins to staphylococcal mastitis.

Staphylococcus aureus / leucotoxin /mastitis / ruminants

1. INTRODUCTION

Leucotoxins are considered to be vir-ulence factors of Staphylococcus aureus,owing to their toxic activities on sev-eral leucocytic cell types. Although director indirect contributions to pathogenic-ity are likely, data supporting the im-portance of staphylococcal leucotoxins ininfection are scarce. Leucotoxins are bi-component pore-forming toxins that targetleucocytes, in particular neutrophilic gran-ulocytes [15]. They constitute a family of

* Corresponding author: rainard@tours.inra.fr

toxins usually designated by the first com-ponent (class S component) that binds tothe target cell, followed by the designa-tion of the second component (class Fcomponent) that associates to the cell-bound first component to finally form anoctameric pore: the possible homologousassociations for strains of ruminant ori-gins are the gamma hemolysins HlgA/Band HlgC/B, and the leucotoxins LukE/Dand LukM/F’. The Panton-Valentine leuco-toxin (LukPVL) is rarely found in S. au-reus strains associated with ruminant mas-titis. The phagocyte-specific activity ofleucotoxins would enable S. aureus to

Article available at http://www.vetres.org or http://dx.doi.org/10.1051/vetres:2007026

686 P. Rainard

interfere with the phagocytic defenses ofthe host they colonize. A prerequisite fora role of leucotoxins in pathogenesis ofinfection is synthesis and secretion at in-fection foci, but evidence of in vivo leuco-toxin production are scant: expression ofthe genes encoding two leucotoxins (luk-PV and hlgCB) was found by RT-PCR ina rabbit model of S. aureus endophtalmi-tis using a strain of human origin [3], and aleucotoxin was detected in a few milk sam-ples from cows infected by S. aureus [11],although its identity was not specified.

Mammary infections are a suitablemodel for investigating the contributionof leucotoxins to pathogenesis. The mostfrequently isolated pathogens in masti-tis of ruminants are staphylococci, andS. aureus is considered a major pathogenresponsible for considerable increases inmilk leucocyte concentration and clini-cal mastitis [2, 4, 18]. Since phagocytosisby neutrophils is an important defense ofthe mammary gland of ruminants againstmastitis-causing bacteria [12, 17], the pro-duction of leucotoxins may confer onS. aureus an increased capacity to resisthost defenses and to settle in the mam-mary gland. Cattle mastitis field isolatesare almost all equipped with genes of theleucotoxins LukE/D and hemolysin γ, anda proportion (50 to 80%) possess the genesencoding the LukM/F’ leucotoxin [6, 21].Data concerning mastitis isolates of goatorigin are few, but indicate that LukM/F’is also well represented [16]. The most ac-tive leucotoxin on bovine neutrophils isLukM/F’ [1]. Considering that this leu-cotoxin is widely found among ruminantmastitis isolates but not among isolates ofhuman or other animal origins [20, 21], itcan be put forward that LukM/LukF’ couldplay a distinctive role in the pathogene-sis of ruminant mastitis. Owing to its highactivity on neutrophils, this leucotoxinmay be the counterpart in ruminants ofthe Panton-Valentine leucotoxin of humanstrains of S. aureus. Although the in vitro

effects of several staphylococcal leucotox-ins have been reported [7, 9], the in vivoproduction of leucotoxins by S. aureus dur-ing infection is much less documented. Inparticular, their production during experi-mentally controlled intramammary infec-tions has never been documented.

The aim of this study was to testwhether a LukM/F’-producing strain ofS. aureus could induce mastitis andwhether leucotoxin could be detected inmilk. An indirect means to assess the invivo production of leucotoxins is to mon-itor the appearance of antibodies in thecourse of infection. That is why antibod-ies to LukM/F’ were recorded in serum andmilk following infection. The neutralizingactivities of normal milk and serum, i.e. be-fore infection, and at different times afterinfection, were also investigated.

2. MATERIALS AND METHODS

2.1. Intramammary challenge withS. aureus

One strain of S. aureus (Ch122), iso-lated from the milk of an infected goat,was used to inoculate mammary glands.This strain was selected from our straincollection on the basis of high produc-tion of LukM/F’ leucotoxin [16]. It harborsthe genes encoding the LukM/F’, LukE/Dand Hlg leucotoxins, but produces essen-tially LukM and LukF’ when grown invitro in conventional bacteriological me-dia [16]. It also possesses the genes forhemolysins alpha and beta, is hemolytic onblood agar plates (strong β hemolysis) andproduces capsular polysaccharide type 8.The strain was kept at –70 ◦C in brain-heart infusion broth (BHI) supplementedwith 7% dimethylsulfoxide (Sigma Chem-ical Co., St. Louis, MO, USA). A tube ofBHI was inoculated overnight with frozenbacteria at 37 ◦C. Purity of the culture waschecked by plating on a blood agar plate.

Production of leucotoxin during mastitis 687

From the overnight BHI culture, a subcul-ture was made in a tube of BHI for 18 h at37 ◦C. The bacteria were pelleted by cen-trifugation (2000 × g, 20 ◦C, 15 min) andwashed twice with 20 mL of Hanks bal-anced salt solution (HBSS; Sigma). Afterthe last centrifugation, the pellet was resus-pended with 10 mL of HBSS, and adjustedto the required concentration on the ba-sis of the optical density of the suspension(OD 1.0 at 600 nm corresponds to about2.7 × 108 CFU/mL). The inoculum con-sisted of 0.5 mL of the bacterial suspensionadjusted to 20 000 cfu/mL, in a 1 mL dis-posable syringe fitted with a blunt-endedcannula. Extra-inocula were prepared and50 µL of their content plated on bloodagar to retrospectively check the size ofthe infectious dose by counting cfu afterovernight incubation at 37 ◦C. The inocu-lum was found to contain about 16 000 cfu.

Healthy lactating multiparous goat doesof Alpine breeds were selected on the basisof the absence of intramammary infec-tions and milk cell concentration less than200 000 cells/mL. They were housed in abarn on a straw bed in groups of aboutsix animals for several months, and weremaintained together with their kids whichwere allowed free suckling. The care anduse of goats conformed to the practicesin effect at the INRA Nouzilly ResearchCenter. Challenge exposure was carried outwithin 2 to 4 months after kidding. Bacte-riological examination of milk (half-glandforemilk) was done with a standard pro-cedure [13] one week before inoculation,on the day of inoculation, every other dayup to two weeks post-inoculation (pi), thentwice a week up to the end of the experi-ment (4 weeks).

To induce intramammary infections,goats were infused through the teat canalwith 0.5 mL of the bacterial suspensionwithin one hour of preparation. Beforeinfusion, teats were swabbed with 70%ethanol, and were gently massaged up-wards after infusion. One udder half was

inoculated, the other one serving as thecontrol gland. The goats were hand-milkedbefore infusion, and the kids were keptaway from the inoculated goats for at least3 h after inoculation.

2.2. Processing of milk samples

Milk samples were taken with asep-sis precautions for bacteriological exam-ination and determination of cell con-tent (number and microscopic appearance).Bacterial shedding in milk was semi-quantitatively assessed by recording thenumber of bacterial colonies in 50 µL milkplated on sheep blood esculine agar platesafter 24 h of incubation at 37 ◦C. Within2 h of sample collection, slides were pre-pared to microscopically assess the appear-ance of milk cells. Milk (100 µL + 30 µLfetal bovine serum) was cytocentrifuged(Shandon Southern centrifuge, ShandonInc. Pittsburgh, PA, USA) and the slideswere stained with May-Grünwald-Giemsabefore examination under the microscope.Another portion (500 µL) was used for cellcount which was performed with a Fosso-matic model 90 (Foss Electric, Hillerod,Denmark). The remainder of the milk wascentrifuged at 1500 × g for 30 min at4 ◦C. The cream was discarded, and theskimmed milk was stored in portions at–20 ◦C.

2.3. Quantitation of leucotoxincomponents in milk by ELISA

ELISA were performed in 96-well flat-bottomed plates (Nunc Immunoplate Max-isorp, Merck Eurolab, Strasbourg, France).Following each incubation step, threewashes were carried out with phosphatebuffered saline (PBS) supplemented with0.05% Tween 20, using a MW 96/384 mi-croplate washer (Beckman Coulter Inc.,Fullerton, CA, USA). Samples and im-mune reagents were diluted in PBS con-taining 0.5% gelatin and phenol red unless

688 P. Rainard

otherwise stated. Appropriate dilutions ofthe immune reagents were dispensed in100 µL samples and incubated at 37 ◦C for1 h unless otherwise stated.

Sandwich ELISA were used to mea-sure the concentrations of LukM andLukF’. The sequence of incubation stepswas as follows: (1) coating of the platewith affinity-purified bovine anti-LukMor anti-LukF’ (5 µg/mL) in carbonate-bicarbonate buffer 0.1 M pH 9.6 overnightat 4 ◦C; (2) blocking with 0.5% gelatinfor 30 min; (3) incubation with milksamples undiluted or diluted 1/4, alongwith serial dilutions of purified LukMor LukF’ to establish the standard curve;(4) incubation with rabbit anti-LukM oranti-LukF’ immune serum diluted 1/1000;(5) incubation with peroxidase-conjugatedgoat antibody to rabbit IgG (Jackson Im-munoresearch Laboratories, West Grove,PA, USA) diluted 1/10000; (6) incubationwith a 2% dilution of 2,2’-azino-bis(3-ethylbenzthiazoline-sulfonate) (ABTS;Sigma) in 0.1 M citrate buffer pH 4.2, withthe addition of 0.075% hydrogen peroxidejust before use. The optical density wasread at 414 nm on a microplate reader(BioKinetics reader; BioTek Instruments,Winooski, VT, USA). Values expressedin ng/mL were extrapolated using linearregression from the standard curve ob-tained with the known concentrations ofLukM or LukF’.

2.4. Titration of leucotoxic activity inmilk and neutralization by specificantibodies

The assay is based on the shape change(spreading) induced on PMN by staphylo-coccal leucotoxins, as described by [1,16].PMN were isolated from goat blood as de-scribed [1], and adjusted to 2 × 106/mLin RPMI 1640 without bicarbonate, with20 mM Hepes (Sigma) and 1 mg/mLbovine serum albumin (low endotoxin, Up-tima Interchim, Montluçon, France). Via-

bility of PMN was higher than 98% (ex-clusion of trypan blue) and more than 95%of cells were granulocytes. Flat-bottomed96-well microtiter plates were coated with1 mg/mL BSA in HBSS for 15 min at37 ◦C. After washing, 80 µL of serialtwofold dilutions of the sample under testwere distributed in the wells of the plate,20 µL of PMN suspension was added, andthe plate was incubated at 37 ◦C for 60 min.Then, the shape of PMN was examinedwith an inverted microscope. The inverseof the last dilution inducing the spreadingof more than 90% of the cells was takenas the leucotoxic titer. Positive control wascarried out with purified LukM/F’ (2 nM)and negative control with RPMI 1640.

Neutralization of the leucotoxic activityof milk samples was achieved by addingaffinity-purified rabbit Ab against LukMand LukF’ [16], and incubating for 4 h at4 ◦C before performing the PMN spreadingassay. Control of specificity was carriedout by incubating the milk samples withthe same concentration (see Results) ofaffinity-purified rabbit Ab to hen ovalbu-min.

2.5. Quantitation of Ab to LukMand LukF’ by ELISA

Antibodies to LukM and to LukF’ weretitrated in milk and sera with indirectELISA. Washings, diluting buffer and in-cubations were as for the leucotoxin com-ponents ELISA. The sequence of incuba-tion steps was as follows: (1) coating of theplate with 2 µg/mL of either purified LukMor LukF’ [16] in carbonate-bicarbonatebuffer 0.1 M pH 9.6; (2) blocking with0.5% gelatin for 30 min; (3) incubationwith sera diluted 1/500 or milk diluted1/4, along with serial dilutions of immunegoat serum to establish the standard curve;(4) incubation with rabbit anti-LukM oranti-LukF’ immune serum diluted 1/1000;(5) incubation with 1/10000 dilution ofperoxidase-conjugated rabbit antibody to

Production of leucotoxin during mastitis 689

goat IgG (Jackson Immunoresearch Labo-ratories); (6) ABTS substrate. The opticaldensity was read at 414 nm on a microplatereader (BioKinetics reader). Caprine seraand milk were titrated in units by extrap-olation, using linear regression, from thestandard curve obtained with the dilutionsof the immune serum which was arbitrar-ily assigned 10 000 Ab units. To obtain thisserum, a goat was immunized three timesat two-month intervals with 50 µg of LukMand LukF’ (in separate preparations) emul-sified in incomplete Freund’s adjuvant. Thetwo toxin components were injected sub-cutaneously at separate sites. Serum wasobtained two weeks after the last injection.

2.6. Assay of neutralizing activityof serum or milk

The capacity of serum or milk toneutralize the cytotoxic activity ofLukM/LukF’ was measured with the PMNspreading assay. The mixture (ratio 1/1)of 50 µL LukM and LukF’ at 4 nM con-centration was incubated for 1 h at 37 ◦Cwith serial twofold dilutions (50 µL) ofserum in wells of 96-well microtiter plates.Then, 20 µL of caprine PMN suspension(2×106/mL) were added, and the plate wasincubated for 1 h at 37 ◦C at rest. Visualassessment of the spreading of PMNwas done under an inverted microscope.The inverse of the last dilution of serumprecluding the toxin-induced spreadingof PMN was taken as the neutralizingtiter. A positive control was performedwith immune goat serum. The immuneserum neutralized 2 nM LukM/F’ up tothe 1/400 dilution. Owing to its opacity,normal milk had to be diluted at leastfour times to allow visual assessment ofPMN spreading, but serum-like mammarysecretion could be tested undiluted.

2.7. Statistical analysis

The results were analyzed with theStatXact 5 software (Cambridge, MA,

USA), using the permutation test for pairedsamples and Spearman correlation (exactnonparametric inference).

3. RESULTS

3.1. Clinical signs

Mild mastitis symptoms were evokedby intramammary inoculation of about16 000 cfu of S. aureus ch122 in 5 of the6 goats. There were local signs of moderateintensity (sensitive to palpation) during theday following the challenge, then no localor systemic clinical signs. One goat (1214)developed a severe form of mastitis, withlocal and systemic signs as early as 12 hpi. At 24 h pi, the inoculated gland wasswelled, red, tender, and there were numer-ous clots in the milk. At 48 h pi, a blue skincircular area (about 7 cm in diameter) wasvisible at the rear of the gland, signalling acase of gangrenous (necrotic) mastitis. Thedecision was made to slaughter the animalfor ethical reasons. The mammary secre-tion at this time was serum-like.

3.2. Inflammatory responseand shedding of bacteria

It was not possible to determine pre-cisely the cell concentration in the milkof the goat (1214) which developed gan-grenous mastitis, because of massive clot-ting, nevertheless cell concentrations of theliquid phase markedly increased as earlyas day 1 pi (Fig. 1). Cell concentrationsincreased sharply in the five inoculatedglands which developed mild mastitis onthe day after inoculation. Thereafter, thecourse of infection differed widely amonganimals (Fig. 1). Cell concentrations re-mained elevated in the milk of two goatsbut declined in the milk of the other three.One goat got rid of the infection veryearly (goat 1398): bacteria were not iso-lated from milk after day 2 pi. Goat 1389never shed cultivable S. aureus, despite a

690 P. Rainard

Figure 1. Log cells in milk of inoculated mammary glands of the five goats that developed mildclinical mastitis. Control values are the means of the non-inoculated gland of the same goats. (Pleaseconsult www.vetres.org for a colour version of this figure.)

strong initial cellular response in milk, sug-gesting that bacteria multiplied initially inthe gland but were promptly eliminated.Pure cultures of S. aureus were consis-tently recovered from milk of the otherthree goats until the end of the experiment.The highest shedding of S. aureus occurredin the milk of goat 1214, with bacterialconcentration >100 000 cfu/mL. The sec-ond highest shedding was by goat 1208.At day 1 pi, the shedding was more than10 000 cfu/mL, about 5000 cfu/mL at day2 pi, and only 1500 cfu/mL at day 3 pi.In milk samples of the other goats, theconcentration of bacteria remained below1000 cfu/mL.

3.3. Leucotoxin in milk of S. aureus-infected glands

The milk was searched for leucotoxinby means of sandwich ELISA. With theLukM-ELISA, capable of detecting 0.4 ngof toxin per mL, all the milk sampleswere negative, except four samples frominfected glands of goats 1208 and 1214.Concentrations of 4.8 and 6.2 ng/mL werefound in the milk of goat 1208, and 140and 180 ng/mL in the milk of goat 1214,at days 1 and 2 pi, respectively. The com-ponent LukF’ was also detected in thesesamples, at concentrations comparable to

that of LukM (203 ng/mL in milk of goat1214 at day 2 pi).

At day 2 pi, the milk of goat 1214,which suffered from gangrenous masti-tis, was examined microscopically. A cy-tospin preparation and May-Grünwald-Giemsa staining showed cells with swelledrounded nucleus, and no intact neutrophils(Fig. 2). This cell appearance contrastedwith the intact morphology of neutrophilsin the milk of the infected glands of theother goats, but was quite comparable tothat of blood neutrophils treated in vitrowith LukM/LukF’ leucotoxin (Fig. 2).

The cytotoxic activity was titrated inmilk of goat 1214 at day 2 pi by thePMN spreading assay. Activity was seenup to dilution 1/20, which corresponds to180/20 = 9 ng/mL of leucotoxin. Affinity-purified antibodies to LukM and LukF’were used to check that the spreadingactivity was due to a member of theleucotoxin family. A molar ratio of 10 to 1was chosen to attempt blocking the toxinactivity: antibodies to LukM were addedto the milk sample at a final concentra-tion of 10 µg/mL, along with 10 µg/mL ofantibodies to LukF’, and the mixture wasincubated for 4 h at 4 ◦C before additionof caprine neutrophils. Antibodies totallyinhibited the spreading of neutrophils, even

Production of leucotoxin during mastitis 691

Figure 2. Appearance of PMN subjected to LukM/F’ and of cells in milk of gangrenous mastitis.(a,b,c) Effect of LukM/F’ on granulocytes (mainly neutrophils) on nucleus appearance. Granulo-cytes were incubated with LukM/F’ (2 nM) during 0 min (a), 5 min (b) and 30 min (c) beforecytocentrifugation and May-Grünwald-Giemsa staining. Arrow in (b) points to an eosinophil. (d)the appearance of cells in the milk of goat 1208 at day 2 pi (mild mastitis); most cells are neutrophilswith normal appearance. (e) appearance of cells in the milk of goat 1214 (gangrenous mastitis) onday 2 pi. A few intact neutrophils are seen among numerous cells with altered nucleus. Small ar-rows point to bacterial aggregates. The biggest arrow points to an altered nucleus. (Please consultwww.vetres.org for a colour version of this figure.)

in undiluted milk. Anti-ovalbumin anti-bodies (20 µg/mL, final concentration) didnot inhibit the activity of milk, which re-mained able to induce neutrophil spread-ing at the 1/20 dilution. Cytocentrifuga-tion and May-Grünwald-Giemsa stainingwas performed to check the morphologyof the neutrophil nucleus. The results cor-responded to the characteristic alterationprovoked by leucotoxins, and the cyto-toxic effect was inhibited by antibodies toLukM/F’ but not by antibodies to ovalbu-min (not shown) in milk samples of goat1214. Altered cells were also seen in themilk sample of goat 1208 at day 1 pi (notshown), but not at day 2 pi (Fig. 2).

3.4. Antibody response to infection:leucotoxin ELISA titers

Before infection, Ab to the two com-ponents of the toxin were detected in the

serum of the six goats, with wide variationsamong animals (Fig. 3). Ab titers in serumof goat 1214 before inoculation were com-parable to the titers of other goats, 356and 48 for Ab anti-LukM and anti-LukF’,respectively. Only three goats developedAb titers higher than pre-infection titers.Titers began to increase at days 7 pi, anda peak was reached at day 21 pi. Titers re-mained low in the serum of the two goatsthat eliminated precociously their infection(No. 1389 and 1398). The highest titerswere in the serum of goat 1208, whichhad the highest numbers of bacteria inmilk. Differences between anti-LukM oranti-LukF’ ELISA titers at day 0 versus ei-ther day 14, 21 or 28 pi were statisticallysignificant (p = 0.03; permutation test fortwo related samples).

In milk, initial Ab titers were very low,from 1/2000 to 1/9250 of serum titers. Inthe milk of inoculated glands, there was

692 P. Rainard

Figure 3. Antibody ELISA titers in serum of the five goats which developed mild clinical mastitis.(A) Ab to LukM as the antigen. (B) Ab to LukF’ as the antigen. Titers were calcutated with referenceto a standard immune goat serum which was given arbitrarily 10 000 ELISA units. Titers at day 0differed significantly from titers at days 14, 21 or 28 (p = 0.031; permutation test for paired samples,one-sided). (Please consult www.vetres.org for a colour version of this figure.)

Figure 4. Mean Ab ELISA titers (LukM +

LukF’) in milk of inoculated glands of the 5individual goats. Titers of Ab to LukM andto LukF’ were added for each sample. Titersat day 0 differed significantly from titers atdays 14, 21 or 28 (p = 0.031; permutation testfor paired samples, one-sided). (Please consultwww.vetres.org for a colour version of this fig-ure.)

a slight initial increase in Ab titers onday 1 pi, probably as a result of the exuda-tion of plasma associated with the inflam-matory response. Nevertheless, marked in-creases in titers were not seen before day14 pi and later on (Fig. 4), probably at leastpartly as a result of combined exudationof plasma and of increases in blood Abtiters. As in serum, the Ab titers did notaugment in the milk of the goats 1289 and1398 (Fig. 4). Differences between ELISA

titers (anti-LukM + antiLukF’) in the milkof inoculated glands at day 0 versus eitherday 14, 21, or 28 pi were statistically sig-nificant (p = 0.03; permutation test for tworelated samples). In control healthy glands,Ab titers followed an upward trend fromday 14 pi, although not significant (Fig. 5).The ratio of Ab titers after/before infectionranged from 2 to 5 in healthy glands and 10to 30 in infected glands.

3.5. Neutralizing antibodies in serumand milk

The serum of all the goats exerted someneutralizing activity on LukM/F’, but itwas rather low because undiluted or at least10% serum were necessary to block the ac-tivity of two times the lowest toxin concen-tration capable of inactivating neutrophils,i.e. 1-2 nM (18-36 ng/mL) (Fig. 5). Incontrast, the serum of the immunized goatinhibited the toxic activity of the toxin upto the 1/400 dilution. At the peak of theELISA serum Ab titers, the neutralizingtiters were increased (Fig. 6): the medianvalue, which before inoculation was 10,reached 40. The correlation between theneutralizing and the ELISA titers were of0.86 and 0.95 at days 0 and 21 pi (p =0.083), respectively, suggesting that most

Production of leucotoxin during mastitis 693

Figure 5. Mean Ab ELISA titers (5 goats) in milk of either inoculated or control healthy glands.Titers are given for each of the two antigens, LukM and LukF’. (Please consult www.vetres.org fora colour version of this figure.)

Figure 6. Neutralizing titers of goat serum before and 21 days after inoculation of the mammarygland with S. aureus ch122. Titer is the inverse of the serum dilution neutralizing twice the minimumactive concentration of LukM/F’ (i.e. 36 ng/mL) in the PMN spreading assay. Titers at day 0 differedsignificantly from titers at day 21 (p = 0.031; permutation test for paired samples, one-sided).(Please consult www.vetres.org for a colour version of this figure.)

of the induced Ab were neutralizing. Neu-tralizing activity was not detected in milk,except in the samples taken 14, 21 and28 days pi from the inoculated mammarygland of goat 1208, which had the high-est milk Ab titers. Nevertheless, only thesmallest dilution that can be tested due tothe opacity of milk (1/4) was inhibitory.

4. DISCUSSION

The main objective of this study wasto investigate whether leucotoxins are pro-duced in vivo in the course of mammary

infection. Several lines of evidence indi-cate that this was the case. First, leucotoxinwas detected in certain milk samples byELISA. Second, leucotoxin-specific cyto-toxic activity was present in milk samplesof a case of gangrenous mastitis. Third, Abtiters increased in the serum and milk ofall the animals that developed lasting in-fection.

The direct demonstration of intramam-mary production of leucotoxins by theirdetection in milk is hindered by the dilu-tion by milk of the minute amounts whichare likely to be generated in the foci of

694 P. Rainard

infection, unless high bacterial concentra-tion are reached in milk where leucotoxinproduction could then take place. The oc-currence of naturally acquired antibodiesand potential elicited antibodies could alsointerfere with immunochemical detectionof the toxins. In the present study, bothantigenic and cytotoxic leucotoxin activi-ties were found in several samples of milkduring infection. Twenty times the mini-mal concentration active on caprine PMNwas found in mammary secretion duringthe case of gangrenous mastitis, along withmassive shedding (more than 105 cfu/mL)of staphylococci.

Others have reported the detection ofleucotoxin in milk samples from cowswith chronic staphylococcal mastitis, butnot leucotoxic activity by milk from in-fected glands [11]. In the present study,cytotoxic activity was found in milk ofgangrenous mastitis. The inhibition of thiscytotoxic activity by specific but not by un-related Ab demonstrated that it was dueto a toxin belonging to the family of bi-nary staphylococcal leucotoxins and not toanother staphylococcal cytotoxin such asa hemolysin or delta-toxin. The cytotoxicactivity found in milk of gangrenous mas-titis was high, since this milk withstood atwenty-fold dilution before loosing its ef-fect on caprine neutrophils. This findingagreed with the appearance in the milksamples of cells which showed the alter-ations characteristic of leucotoxin activity(Fig. 2). The identity of the leucotoxin de-tected in milk cannot be ascertained onlyon the basis of ELISA, because of thecross-reactivity of Ab due to sequenceidentity between class S components (55to 72%) and class F components (71 to79%), respectively [14]. Nevertheless, it isprobable that the major leucotoxin in pos-itive samples was LukM/LukF’, for tworeasons. First, S. aureus Ch122 essen-tially produces this leucotoxin [16]. Sec-ond, the robust cytotoxic activity of milkof gangrenous mastitis strongly suggests

that the leucotoxin produced was indeedLukM/F’, because at the least active con-centration (9 ng/mL in the 1/20 dilution),none of the other leucotoxins are activeon caprine neutrophils, since they are atbest 10 times less active than LukM/F’ [1].In fact, the toxin in milk seemed to beeven slightly more efficient than purifiedLukM/F’, which is active down to about18 ng/mL on caprine neutrophils ([16] andthis study). Possibly other toxins, such ashemolysins, secreted in milk along withleucotoxins, contributed to this strong ac-tivity.

Indirect evidence that leucotoxins weregenerated in vivo was the increases in an-tibody titers during or following infection,assuming that rising titers result from theproduction of the corresponding antigenby the mastitis-causing bacteria. The sharprises in Ab titers which occurred in the14 days following infection (Fig. 3) in-dicate that leucotoxins were produced invivo. Persistence of infection was neces-sary to elicit Ab, since the two goats thateliminated staphylococci from the inocu-lated gland within 2 days after inoculationdid not develop Ab. On the contrary, mildmastitis cases, associated with moderateshedding of bacteria, were sufficient to in-duce Ab, although bacterial multiplicationwas insufficient for Ag to be detected inmost milk samples. Higher leucotoxin AbELISA values in milk and serum of cowswith naturally acquired S. aureus masti-tis than in mastitis-free cows has beenreported [10]. This suggests that chronicinfections, mainly subclinical, elicit Ab re-sponses to leucotoxins also in cattle. Aparallel can be made with another potentialvirulence factor of mastitis-causing S. au-reus, enterotoxin D. This enterotoxin wasrecently shown to be produced in vivoand was found in the milk of a few in-fected cows, at concentrations active onbovine lymphocytes [19]. Specific Ab wereelicited, which were able to neutralize thetoxin. From these results and those of

Production of leucotoxin during mastitis 695

the present study, it can be put forwardthat during chronic mammary infections,although subclinical during most of theirduration, toxins are produced which are ca-pable of eliciting systemic Ab responses.

Outcomes of infections ranged fromprompt elimination without clinical signsto gangrenous mastitis. The group of in-oculated goats was not very homogeneous,and the conditions of their rearing did notpreclude contact with S. aureus prior toinoculation. This may account for the het-erogeneity of the responses to infection.No clear association between Ab titers be-fore inoculation and severity of mastitiswas obvious in this study.

The possession of lukPV genes by iso-lates of human origin appears to be associ-ated with increased severity, ranging fromsevere cutaneous infection to necrotizingpneumonia [5, 8]. It may be that LukM/F’,which is the most active leucotoxin onruminant PMN, as the Panton-Valentineleucocidin is on human PMN, contributesto the severity of mastitis. The leucotoxinLukM/F’ seems to be an attribute of strainsisolated from mastitis of ruminants, since itis not present in the genome of strains iso-lated from other hosts [20,21]. Recently, ithas been shown that staphylococcal exose-cretion containing LukM/F’ can provokean inflammatory reaction when infused inthe mammary gland [22], suggesting thatthis leucotoxin can play a part in the sever-ity of the inflammatory response. The invivo production of LukM/F’ is a prereq-uisite for a contribution to pathogenesis,and the present study brings argumentsthat support this production during mam-mary infection. Moreover, the observedcase of gangrenous (necrotic) mastitis sug-gests that LukM/F’ could play a role inthis syndrome. It can be hypothesized thatwhen the population of a high-leucotoxin-producing strain reaches a sufficient sizein the mammary gland, leucotoxin is pro-duced in amounts that overwhelm thenatural neutralizing Ab. Then phagocytic

defenses are altered, and bacteria multi-plication escapes control. Accordingly, theconditions would be met for gangrenousmastitis to develop. This hypothesis wouldbe worth testing with a mutant strain defec-tive in LukM production.

In conclusion, this study shows that thelukM gene was functional, and activelytranscribed and translated in the course ofmammary infection. Leucotoxin and cyto-toxic activity were found in a few milksamples, and this activity was inhibited byAb to LukM/F’. Antibody titers increasedwhen mammary infection lasted more thana few days. In a case of gangrenous mas-titis, milk cells displayed the character-istic appearance of neutrophils incubatedwith purified LukM/F’. It can be concludedthat leucotoxin, most likely LukM/F’, wasproduced during mammary infection, andthe high leucotoxic activity found in gan-grenous mastitis secretion suggests thatthis toxin could play a role in the patho-genesis of severe forms of mastitis.

ACKNOWLEDGEMENTS

The author thanks Philippe Bernardet andthe staff of animal experimental facilities formanaging the goats and collecting blood andmilk samples, and Florence B. Gilbert (IASP)for fruitful discussions and critical reading ofthe manuscript.

REFERENCES

[1] Barrio M.B., Rainard P., Prevost G.,LukM/LukF’-PV is the most activeStaphylococcus aureus leukotoxin onbovine neutrophils, Microbes Infect. (2006)8:2068–2074.

[2] Bergonier D., de Cremoux R., Rupp R.,Lagriffoul G., Berthelot X., Mastitis ofdairy small ruminants, Vet. Res. (2003)34:689–716.

[3] Bronner S., Jehl F., Peter J.D., PloyM.C., Renault C., Arvis P., Monteil H.,Prevost G., Moxifloxacin efficacy and vit-reous penetration in a rabbit model of

696 P. Rainard

Staphylococcus aureus endophthalmitis andeffect on gene expression of leucotoxinsand virulence regulator factors, Antimicrob.Agents Chemother. (2003) 47:1621–1629.

[4] Contreras A., Luengo C., Sanchez A.,Corrales J.C., The role of intramammarypathogens in dairy goats, Livest. Prod. Sci.(2003) 79:273–283.

[5] Etienne J., Panton-Valentine leukocidin: amarker of severity for Staphylococcus au-reus infection? Clin. Infect. Dis. (2005)41:591–593.

[6] Fueyo J.M., Mendoza M.C., Rodicio M.R.,Muniz J., Alvarez M.A., Martin M.C.,Cytotoxin and pyrogenic toxin superantigengene profiles of Staphylococcus aureus as-sociated with subclinical mastitis in dairycows and relationships with macrorestrictiongenomic profiles, J. Clin. Microbiol. (2005)43:1278–1284.

[7] Genestier A.L., Michallet M.C., PrevostG., Bellot G., Chalabreysse L., Peyrol S.,Thivolet F., Etienne J., Lina G., Vallette F.M.,Vandenesch F., Genestier L., Staphylococcusaureus Panton-Valentine leukocidin directlytargets mitochondria and induces Bax-independent apoptosis of human neutrophils,J. Clin. Invest. (2005) 115:3117–3127.

[8] Gillet Y., Issartel B., Vanhems P., FournetJ.C., Lina G., Bes M., Vandenesch F.,Piémont Y., Brousse N., Floret D., Etienne J.,Association between Staphylococcus aureusstrains carrying gene for Panton-Valentineleukocidin and highly lethal necrotisingpneumonia in young immunocompetent pa-tients, Lancet (2002) 359:753–759.

[9] König B., Prévost G., Piémont Y., KönigW., Effects of Staphylococcus aureus leuko-cidins on inflammatory mediator releasefrom human granulocytes, J. Infect. Dis.(1995) 171:607–613.

[10] Loeffler D.A., Norcross N.L., Enzyme-linked immunosorbent assay for detection ofmilk immunoglobulins to leukocidin toxinof Staphylococcus aureus, Am. J. Vet. Res.(1985) 46:1728–1732.

[11] Loeffler D.A., Creasy M.T., Norcross N.L.,Paape M.J., Enzyme-linked immunosorbentassay for detection of leukocidin toxin fromStaphylococcus aureus in bovine milk sam-ples, J. Clin. Microbiol. (1988) 26:1331–1334.

[12] Paape M., Mehrzad J., Zhao X., DetilleuxJ., Burvenich C., Defense of the bovinemammary gland by polymorphonuclear neu-

trophil leukocytes, J. Mammary Gland Biol.Neoplasia (2002) 7:109–121.

[13] Poutrel B., Rainard P., Predicting theprobability of quarter infection (by majorpathogens) from somatic cell concentration,Am. J. Vet. Res. (1982) 43:1296–1299.

[14] Prévost G., The bi-component staphylo-coccal leucocidins and γ-haemolysins (tox-ins), in: Alouf J.E., Freer J.H. (Eds.), Thecomprehensive sourcebook of bacterial pro-tein toxins, Academic Press, London, 1999,pp. 402–418.

[15] Prévost G., Mourey L., Colin D.A.,Menestrina G., Staphylococcal pore-formingtoxins, Curr. Top. Microbiol. Immunol.(2001) 257:53–83.

[16] Rainard P., Corrales J.C., Barrio M.B.,Cochard T., Poutrel B., Leucotoxic activi-ties of Staphylococcus aureus strains isolatedfrom cows, ewes, and goats with masti-tis: importance of LukM/LukF’-PV leuko-toxin, Clin. Diagn. Lab. Immunol. (2003)10:272–277.

[17] Rainard P., Riollet C., Innate immunity ofthe bovine mammary gland, Vet. Res. (2006)37:369–400.

[18] Sutra L., Poutrel B., Virulence factors in-volved in the pathogenesis of bovine intra-mammary infections due to Staphylococcusaureus, J. Med. Microbiol. (1994) 40:79–89.

[19] Tollersrud T., Kampen A.H., Kenny K.,Staphylococcus aureus enterotoxin D is se-creted in milk and stimulates specific anti-body responses in cows in the course of ex-perimental intramammary infection, Infect.Immun. (2006) 74:3507–3512.

[20] Von Eiff C., Friedrich A.W., Peters G.,Becker K., Prevalence of genes encodingfor members of the staphylococcal leuko-toxin family among clinical isolates ofStaphylococcus aureus, Diagn. Microbiol.Infect. Dis. (2004) 49:157–162.

[21] Yamada T., Tochimaru N., Nakasuji S.,Hata E., Kobayashi H., Eguchi M., KanekoJ., Kamio Y., Kaidoh T., Takeuchi S.,Leukotoxin family genes in Staphylococcusaureus isolated from domestic animals andprevalence of lukM-lukF-PV genes by bacte-riophages in bovine isolates, Vet. Microbiol.(2005) 110:97–103.

[22] Younis A., Krifucks O., Fleminger G., HellerE.D., Gollop N., Saran A., Leitner G.,Staphylococcus aureus leucocidin, a viru-lence factor in bovine mastitis, J. Dairy Res.(2005) 72:188–194.