INFECTION AND IMMUNITY, Nov. 1990, p. 3779-3787 Vol. 58, No. 110019-9567/90/113779-09$02.00/0Copyright C) 1990, American Society for Microbiology

Streptococcus pyogenes Clinical Isolates and Lipoteichoic AcidOFRA LEON AND CHARLES PANOS*

Department of Microbiology, Jefferson Medical College, Thomas Jefferson University, Philadelphia, Pennsylvania 19107

Received 10 May 1990/Accepted 20 August 1990

Minimally subcultured clinical isolates of virulent nephritogenic and nonnephritogenic Streptococcuspyogenes of the same serotype showed major differences in lipoteichoic acid (LTA) production, secretion, andstructure. These were related to changes in coccal adherence to and destruction of growing human skin cellmonolayers in vitro. A possible relationship between cellular LTA content and group A streptococcal surfacehydrophobicity was also investigated. Nephritogenic S. pyogenes M18 produced twice as much total (i.e.,cellular and secretory) LTA as did the virulent, serologically identical, but nonnephritogenic isolate. Also, theLTAs from these organisms differed markedly. The polyglycerol phosphate chain of LTA from thenephritogenic isolate was longer (1.6 times) than was that from the nonnephritogenic isolate. Likewise, bothLTAs indicated the presence of alanine and the absence of glucose. Amino sugars were found in LTA from onlynephritogenic S. pyogenes. Teichoic acid, as a cellular component or secretory product, was not detected. Theadherence of two different nephritogenic group A streptococcal serotypes (M18 and M2) exceeded that of theserologically identical but nonnephritogenic isolates (by about five times), indicating a correlation betweenvirulent strains causing acute glomerulonephritis and adherence to human skin cell monolayers. Likewise, LTAfrom nephritogenic S. pyogenes M18 was more cytotoxic (1.5 times) than was that from the nonnephritogenicisolate for human skin cells, as determined by protein release. This difference was not perceptible by the moresensitive dye exclusion method (i.e., requiring less LTA), which emphasizes changes in host cell morphologyand death. Also, the secretion of LTA by only virulent nephritogenic S. pyogenes M18 was exacerbated bypenicillin (a maximum of four times). Finally, while the adherence of nephritogenic S. pyogenes M18 decreasedmarkedly after continued subculturing in vitro, the surface hydrophobicity did not.

A dramatic decline in the prevalence of serious infectionscaused by group A streptococci (GAS) has occurred duringthis century. However, streptococcal infections of surprisingseverity have now begun to reappear. For example, recentoutbreaks of acute rheumatic fever among children andmilitary recruits have been recorded (3, 4, 38, 39). In aregional outbreak within the United States, patients withsevere GAS infections had soft tissue infections, shock,renal impairment, and acute respiratory distress syndrome.In addition, a mortality rate of 30% was noted. The strains ofStreptococcus pyogenes isolated were not of a single sero-type. In addition, most produced pyrogenic exotoxin A, atoxin not commonly observed in the recent past (35). SevereGAS outbreaks were also recently reported in Great Britain(9). These were related to changes in other GAS virulencefactors. In at least one of these outbreaks, host factors didnot appear to explain the increased severity of these strep-tococcal infections (35). A current consensus is that thisresurgence is associated with the emergence of more virulentorganisms.A prerequisite for successful bacterial infection in vivo is

adherence to a susceptible host cell. Considerable data havenow accumulated indicating that the adherence mechanismwithin the pathogenic GAS involves lipoteichoic acid (LTA)(2, 6, 26, 41). LTA is a cellular component as well as asecretory product of S. pyogenes. In addition to beingamphipathic and amphoteric, LTA is highly cytotoxic for avariety of growing eucaryotic cell monolayers and for glo-meruli in tissue cultures (6, 15, 36). Deacylation of LTAabolishes its cytotoxicity. Likewise, treatment of host cellswith LTA or intact cells of S. pyogenes with anti-LTA serumnegates coccal adherence.

* Corresponding author.

Pyrogenic exotoxin A has begun to reappear in certainvirulent GAS (35). This toxin has several biological proper-ties in common with streptococcal LTA. These includecytotoxicity, mitogenicity, and immunosuppression. LTAalso induces the Shwartzman reaction. Therefore, the mal-adies indicated above and associated with the resurgentvirulence of certain GAS may be due to an additive effect ofpyrogenic exotoxin A and other secretions, including LTA.Given the variety of toxins produced by S. pyogenes, thiseffect is more than just a probability.

Studies with fresh isolates of virulent and nonvirulentgroup B streptococci (GBS) from symptomatic and asymp-tomatic individuals have already revealed significant differ-ences in the structure and production of their LTAs (22, 23).Also, adherence, like virulence, is a transitory property ofonly virulent GBS (20). Such information is not available forfresh isolates of S. pyogenes. Changes which enhance thepotential role of LTA in disease are of prime importance forour understanding of the current resurgence of GAS viru-lence, and in this regard toxemia may be an important factor.This study compares the adherence and cytotoxicity ofclinical isolates of virulent nephritogenic and nonnephrito-genic GAS with differences in their LTA production, struc-ture, and secretion.

MATERIALS AND METHODSBacteria and LTA. With one exception, all organisms were

virulent isolates. The designations of each as used here,including serotype and pertinent isolation details, were asfollows: M18AGN from the throat of a patient with acuteglomerulonephritis (AGN), M18NAGN from the peritonealfluid of a patient with bacteremia and peritonitis (indicatingits invasive property) but without AGN (i.e., NAGN), andM2AGN from a patient with AGN. M2NAGN, the NAGNcontrol, was from a classmate of a patient with rheumatic

3779

on Decem

ber 13, 2020 by guesthttp://iai.asm

.org/D

ownloaded from

3780 LEON AND PANOS

fever. These organisms were supplied by E. Kaplan, WorldHealth Organization Collaborating Center for Reference andResearch on Streptococci, University of Minnesota. Addi-tional details and the source of each isolate may be requestedfrom that center. The corresponding center numbers forthese strains (in parentheses) were as follows: M18AGN(88-145), M18NAGN (88-126), M2AGN (87-407), andM2NAGN (88-120). From the time of patient isolation untiluse in these studies, none of these isolates was transferredmore than 10 times. Frozen stock inocula of each isolatewere prepared, stored, and treated before use exactly asdetailed previously (20). When necessary, each experimentwas initiated from an original stock inoculum.LTA was isolated from intact M18AGN and M18NAGN

cells by the cold phenol method and purified as detailedpreviously (10). Its high purity was similar to those ofprevious LTA preparations from S. pyogenes and S. agalac-tiae (e.g., 0.35% protein, no nucleic acids) (10).

Culture media. Viable counts were determined with Todd-Hewett broth (THB; BBL Microbiology Systems, Cockeys-ville, Md.) with agar (1.5% [wt/vol] purified Bacto-Agar;Difco, Detroit, Mich.) added. Adherence studies were donewith organisms grown in THB (37°C, 5% [vol/vol] inoculum)and harvested during the mid-logarithmic phase of growth(about 3 h). The organisms were harvested by centrifugation(4,340 x g, 10 min) and washed twice with phosphate-buffered saline (PBS) before use. For biochemical studies,10- to 15-liter batches of THB were inoculated with organ-isms (5% [vol/vol]) and inoculated overnight at 37°C, corre-sponding to the late stationary phase of growth. Cells wereharvested by centrifugation (6,130 x g, 20 min), washedtwice with 0.9% (wt/vol) saline and once with distilled water,and lyophilized. Coccal generation times were calculatedfrom viable counts obtained by inoculating a mid-logarith-mic-phase culture of each organism (5% [vol/vol], 2.5 x 108CFU) into 50 ml of THB in an Erlenmeyer flask (125 ml)equipped with a side arm for determination of the opticaldensity (OD) at 500 ,um and by plating replicates periodicallyfor 24 h.

Tissue culture cells. Human (21-year-old male) primaryskin (CRL-1474/CCD-25SK) and established mouse fibro-blast (CCL-1/L-929) cells from the American Type CultureCollection, Rockville, Md., were used. Skin cells werereceived at passage number 6 and were used in experimentsat passages 9 to 12. These cells were propagated in minimalessential medium (MEM) containing 20% (vol/vol) heat-inactivated fetal bovine serum (MEM-20), 2 mM L-glU-tamine, 0.1 mM nonessential amino acids, 100 U of penicillinper ml, and 100 ,ug of streptomycin per ml. Mouse fibroblastswere grown in the same medium but with 10% (vol/vol)serum and without the nonessential amino acids added. Thefinal pH of all media was 7.2. Media and reagents were fromGIBCO Laboratories, Grand Island, N.Y. Both cell lineswere grown in flat, 75-cm2, 250-ml plastic bottles (CorningWorks, Corning, N.Y.) at 37°C in at atmosphere ofCO2 (5%)and air (95%). Viable cell counts were determined witherythrosine B and a hemacytometer.

Streptococcal adherence. This assay was conducted asdetailed previously except that 104 human skin cells in 1 mlof MEM-20 were seeded into glass Leighton tubes (10, 20).Also, 1 ml of a mid-logarithmic-phase suspension of approx-imately 2.5 x 108 CFU of each S. pyogenes isolate per mlwas prepared and used as described before (20). The distinctadvantages of this assay have been detailed previously (10,20). Results are expressed as the number of attached bacte-

ria and the number of human cells binding bacteria per 100tissue culture cells.

Cytotoxicity studies with LTA. Cytotoxicity was evaluatedby measuring radioactivity and viability. For radiolabelingexperiments, a confluent monolayer of human skin cellsgrown in 75-cm2 bottles with supplemented MEM-20 (seeabove) was trypsinized and 104 cells in 1 ml of medium wereseeded into each of 24 wells (Linbro 76-033-05; well area, 2.0cm2; Flow Laboratories, Inc., Hamden, Conn.). Incubationfor 24 h at 37°C in 5% C02-95% air provided a subconfluentmonolayer. Spent medium was removed, and 1 ml of freshmedium containing 2.6 ,uCi L-[355]methionine per ml wasadded (specific activity, 1,156 Ci/mmol; 10 mCi/ml of H20;ICN Biomedicals Inc., Irvine, Calif.; stored as 1-ml aliquotsof 1 mCi/ml of H20 at -80°C and further diluted withMEM-20 to a concentration of 2.6 ,uCi/ml before use). Aftereach monolayer was labeled for 24 h, the medium wasremoved and each monolayer was washed twice with PBS.Next, 0.5 ml of fresh MEM-0 (i.e., without serum) and 0.5 mlof MEM-0 with various concentrations of LTA adjusted topH 7.2 with HCO3 were added. Medium without serumwas necessary to prevent overgrowth and to maintain sub-confluency. Normal skin cell morphology could not bemaintained without serum for more than 24 h. All solutionsofLTA were prepared prior to use; the concentrations testedranged from 1 to 200 ,ug per well. After 24 h of incubationwith LTA, monolayers in all wells were washed twice withPBS and digested with 0.1 ml of 1 N NaOH for 30 min at37°C. The digests and the well washings (twice with 0.5 ml ofPBS) were combined, transferred to scintillation vials (10 mlof Bray cocktail), and counted in an LKB 1209 Rackbetaliquid scintillation counter. Percent cytotoxicity was calcu-lated as described before (15).

Cytotoxicity was also determined by measuring the de-cline in viability in the multiple-well plates described above.Seeded human skin or mouse fibroblast cells (104 per well)were incubated for 48 and 24 h, respectively, and then refedwith MEM-0 containing various concentrations of LTA asdescribed above. After 24 h in the presence of LTA, viabilitywas assessed as follows. Medium was removed, each mono-layer was washed twice with PBS, and 0.5 ml of PBS wasadded to all wells. Following the addition of 0.05 ml oferythrosine B and incubation at room temperature for 20min, live versus dead cells were counted at a magnificationof x 100 with a Nikon inverted microscope equipped with a21-mm reticle eyepiece, with a control monolayer showingapproximately 500 cells. Dead cells were defined as cellsretaining the stain. Percent toxicity was calculated as fol-lows: percent toxicity = [(number of cells without LTA -number of cells with LTA)/number of cells without LTA] x100.

Secretion of LTA and TA. Direct and indirect passivehemagglutination assays (PHA) were performed essentiallyas described previously with purified LTA from S. pyogenesM18AGN serving as the standard (21, 26). In brief, thesensitizing activity of LTA in the medium was measured bydirect PHA with sheep erythrocytes, whereas indirect PHAassessed the antigenic activity of secreted LTA and teichoicacid (TA). Differences between antigenic and sensitizingactivities were taken as the amount of TA present in themedium. For these studies, the anti-LTA serum used previ-ously was used again. This antiserum had been prepared inrabbits with purified LTA from S. agalactiae (10). It cross-reacted with LTA from this pathogenic group B type IIIcoccus and a nephritogenic type 12 S. pyogenes to the sameextent (titer, 1:256) in PHA with rabbit erythrocytes.

INFECT. IMMUN.

on Decem

ber 13, 2020 by guesthttp://iai.asm

.org/D

ownloaded from

S. PYOGENES CLINICAL ISOLATES AND LTA 3781

Secretion of glycerol-labeled LTA with and without pen-icillin was also determined. First, S. pyogenes M18 isolateswere grown overnight in THB (30 ml) containing 10 ,uCi of[3H]glycerol per ml (specific activity, 500 mCi/mmol; Amer-sham Corp., Arlington Heights, Ill.). Cultures were centri-fuged (4,340 x g, 10 min), washed three times with 10 ml ofPBS, suspended in fresh THB (30 ml), and divided into 3-mlportions. These portions, with or without penicillin (forconcentrations, see Fig. 6) were incubated at 37°C for 3 h.Aliquots of each were centrifuged, the supernatants were

inactivated at 56°C for 30 min, and 0.5 ml of each superna-

tant was counted as described above for assessment ofglycerol secretion. Next, aliquots (100 pul) of each superna-

tant were mixed with 100 ,ul of sheep erythrocytes (108/ml)and incubated at 37°C for 3 h with occasional mixing inBeckman Microfuge tubes. The binding of LTA to sheeperythrocytes was stopped by centrifugation in a BeckmanMicrofuge B for 2.5 min. Erythrocytes were washed twicewith 100 ,ul of PBS, the washes were added to the superna-

tants, and each was counted as described before. Finally,remaining sheep erythrocytes were lysed with distilled water(100 ,ul), and the lysed cells were transferred to scintillationvials. Also, all Microfuge tubes were washed with 1 NNaOH (100 pu1), and the washes were added to erythrocytepreparations. All erythrocyte-NaOH mixtures were allowedto digest by incubation for 15 min at room temperaturebefore being bleached with 300 p.l of 30% (vol/vol) H202.Radioactivity was measured as described above after theaddition of Bray cocktail. Remaining medium aliquots were

used to determine LTA and TA contents by direct andindirect PHA (see above).

Differences in the hydrophobicity of intact coccal cells. Themethod of Rosenberg et al. was used to measure hydropho-bicity (29). Ten-milliliter cultures (mid-logarithmic and sta-tionary phases) of S. pyogenes M18AGN and M18NAGNwere grown in THB. Cells were obtained by centrifugation(4,340 x g, 10 min) and washed twice with 10 ml of PUMbuffer (0.1 M potassium phosphate containing 1.8 g of urea

and 0.2 g of MgSO4 - 7H20 per 1,000 ml; final pH, 7.1). Afterbeing suspended in this buffer, cells were vortexed todisperse any clumping before being adjusted to an OD at 450p.m of 0.500 with a Spectronic 20 spectrophotometer(Bausch & Lomb, Rochester, N.Y.). Three-milliliter suspen-

sions of each isolate were placed in a series of 15-ml glasstubes, and increasing volumes ofp-xylene (25 to 200 ,ul) wereadded (see Fig. 7). Each mixture was vortexed for 60 s andallowed to separate at room temperature for 15 min. Com-parably aged cells of a laboratory strain of Escherichia coligrown in THB served as a control. The OD of aqueous layerswas read at 450 ,um, and changes in hydrophobicity were

calculated as follows: percent hydrophobicity = 100 - [(100x OD after treatment)/OD before treatment].LTA structure. Chain length determinations (i.e., total

phosphorus/organic phosphorus) of each purified LTA weredone by alkaline phosphatase digestion as described before(33). The composition of each LTA was established afteracid hydrolysis (1 mg of LTA and 0.5 ml of 2 N HCl in a

sealed tube at 100°C for 3 h) by descending paper chroma-tography (n-butanol-pyridine-water, 90:60:45 [vol/vol/vol]).Amino acids and amino sugars were detected by dipping inninhydrin. Glycerol and sugars were also visualized withammoniacal AgNO3. One microgram of each standard (ala-nine, glycerol, glucose, glucosamine, and N-acetyl-D-glu-cosamine) was easily discernible by this procedure (10, 33,34). Estimation of the phosphorus/amino sugar ratio of LTAwas achieved by visual comparison of chromatograms with

TABLE 1. Cell-associated and secreted LTA yields of virulentS. pyogenes M18 isolatesa

Cell-associated LTASecreted TotalS.

8Dyygenes Crude extract LTA LTAM- 8 Dry cell rafter phenol Pure (mg) (mg)b (mg)

wt (g) extraction (mg)'

AGN 15.21 310.05 (17.86) 77.60 (8.75) 1,440.00 1,517.60NAGN 17.10 145.96 (13.90) 38.65 (5.68) 720.00 758.65

aPooled batch culture (total, 45 liters); overnight incubation in THB.b Determined by PHA.' Numbers in parentheses represent weight in milligrams of phosphorus.

incremental increases in the standard (glucosamine) and thehydrolysate (1 to 100 and 5 to 100 ,ug per spot, respectively)and by phosphorus equivalency calculations (see Table 1).

Statistics. Statistical analysis was performed with thepaired Student t test.

RESULTS

LTA yields. The generation time of S. pyogenes M18AGNin THB was 20 min; that of S. pyogenes M18NAGN was 40min. The yields of LTA from these clinical isolates areshown in Table 1. The cellular content ofLTA was greater inthe former (0.51%) than in the latter (0.23%). Also, the actualamounts of LTA secreted by both isolates were larger thanthe amounts retained by each as a cellular component, withM18AGN secreting still more (2.13 times) LTA on anequivalent-cell-weight basis. Finally, the total amounts ofLTA produced (i.e., cellular plus secreted) by these organ-isms differed markedly. Overall, M18AGN produced 2.3times more total LTA than did M18NAGN. The calculatedphosphorus equivalency of purified cellular LTA from eachof these two clinical isolates was not the same, indicating adifference in molecular weight between their LTAs. Onemilligram of LTA phosphorus was found to be equivalent to8.9 and 6.8 mg ofLTA from the virulent AGN and NAGN S.pyogenes isolates, respectively (Table 1). Exhaustive studiesfailed to indicate the secretion ofTA (or deacylated LTA) byeither of these M18 clinical isolates (see below).LTA structure. Analyses of purified LTA from these M18

clinical isolates substantiated the molecular difference indi-cated by the phosphorus equivalency calculations. Chainlength determinations of LTA from the AGN coccus re-vealed a chain length of 25 + 1.0 glycerol-phosphate units;the chain length for the NAGN coccus was 16 + 0.7 units (n= 2). In addition, compositional differences were docu-mented. Alanine was barely perceptible and glucose was notdetected in acid hydrolysates of LTA from either isolate bypaper chromatography. Only LTA from AGN S. pyogenescontained traces of glucosamine and an unidentified butmore pronounced amino sugar. The estimated phosphorus/total amino sugar ratio of this LTA was 1:0.44, with theunidentified but slower-moving (Rf, 0.13) amino sugar beingin excess of glucosamine (Rf, 0.25). The method used for thehydrolysis of LTA deacylated N-actylglucosamine (Rf, 0.58)to glucosamine, precluding the establishment of this acety-lated derivative as the hexosamine of LTA in this AGN S.pyogenes isolate.

Streptococcal adherence to human skin cells. Maximalcoccal adherence was achieved with subconfluent monolay-ers and GAS grown to the mid-logarithmic phase of growth.Two sets of AGN and NAGN S. pyogenes isolates ofdifferent serotypes (M18 and M2 GAS) were examined. Two

VOL. 58, 1990

on Decem

ber 13, 2020 by guesthttp://iai.asm

.org/D

ownloaded from

3782 LEON AND PANOS

12]

10

CD 0C.)-i

z 0

ul

z2

cxzm

AGN NAGN

STREPTOCOCCI

IL

AGN NAGN

8

4

V(1,2)

1-3 4-6 7-9 10-12 13-16 17-22

NUMBER OF CFU/INDIVIDUAL HOST CELL

FIG. 2. Streptococcal distribution per 100 human skin cells bind-ing cocci in tissue cultures. Shown are data for clinical isolates of S.pyogenes M18AGN (M) and M18NAGN (n). Values are averagesfor one experiment done in duplicate.

V(1,2)

STREPTOCOCCI

FIG. 1. Streptococcal adherence to human skin cell monolayersin tissue cultures. AGN, Acute glomerulonephritogenic S. pyogenesM18 and M2 isolates; NAGN, nonacute glomerulonephritogenic S.pyogenes M18 and M2 isolates; V(1,2), recent human (neonatal)isolates of virulent S. agalactiae type III (controls; see also refer-ence 20). Values are averages + standard deviations for six exper-iments done separately.

virulent neonatal clinical isolates of S. agalactiae type IIIserved as controls for adherence specificity, while avirulentM2 was the NAGN control. Since there were no significantdifferences in adherence within the different coccal groupsexamined (P c 0.05), data for each group were combinedand are presented as averages (Fig. 1). A mean of 48% of thehuman skin cell population after 24 h of growth was capableof binding S. pyogenes M18AGN and M2AGN (Fig. 1A).Their virulent and avirulent NAGN counterparts and thevirulent GBS adhered to not more than 20% of this same cellpopulation. Figure 1B shows the actual CFU (i.e., chains) ofeach of these coccal groups adhering to 100 human skincells. Again, the AGN isolates far exceeded the NAGNcontrols in binding to these host cells. Of note was thealmost complete lack of adherence by the virulent GBScontrols. Finally, stationary-phase cells of the AGN GASadhered to the same extent as did mid-logarithmic-phasecells of the NAGN GAS (Fig. 1).

Figure 2 quantitates a typical distribution difference inadherence between the M18AGN and M18NAGN isolatesby only that segment of the human skin cell populationbinding these GAS (approximately 55 and 15%, respectively;see also Fig. 1A). For example, the CFU bound per host cellranged from 1 to 22. While approximately 10% of theadherent skin cell population was able to bind the maximumnumber of AGN coccal chains (17 to 22 CFU per cell), nonebound the NAGN isolate to this extent. Finally, the AGNisolate was also more uniformly distributed among theadherent host cell population than was the NAGN isolate.Table 2 shows the results of treatments that interfere with

coccal adherence. Treatment of S. pyogenes M18AGN withanti-LTA serum or human skin monolayers with LTA inhib-ited the CFU binding to host cells almost completely.Likewise, the percentage of host cells receptive to thispathogen after each of these treatments was also lowered byapproximately 80%.

Effect of daily transfers in vitro on coccal adherence to

TABLE 2. Cell pretreatments and changes in adherence ofS. pyogenes M18AGN'

Mean + SD CFU Mean + SD %Pretreatment bound/100 host cells receptive host cells

None 379.8 ± 40.8 57.2 ± 6.3Normal rabbit serum 386.7 ± 35.9 (2) 55.2 + 7.6 (2)

(1:128 final dilution)Anti-LTA serum (1:128 5.0 ± 2.8 (2) 11.0 ± 4.2 (2)

final dilution)LTA treatment of skin 7.5 ± 2.1 (2) 10.5 ± 5.0 (2)

cell monolayersa The duration of pretreatments was 30 min in PBS at 37°C. Untreated cells

were kept in PBS for the corresponding time period. The numbers of coverslips counted in separate experiments are shown in parentheses. Growing skinmonolayers were 24 h old; the coccus was at its mid-logarithmic phase ofgrowth.

A 60 -

50 -

CD

e 40-

0x,, 30-

Pa.C 20-

a:10-

o

B500 -

> 400-

Oy(C 300z z

200-

o 100-

0O

INFECT. IMMUN.

on Decem

ber 13, 2020 by guesthttp://iai.asm

.org/D

ownloaded from

S. PYOGENES CLINICAL ISOLATES AND LTA 3783

500 -

(A

w0

azIO00C)00

_Oz0 0v-x<: 'o,0

IL.C.-

400 -

300 -

200 -

100 -

00

100

0-J-i

LUw

0

I.-0L0iw)II0.wg

5 7 9 11 13 15 17 19 21 23 25

DAILY TRANSFERS

FIG. 3. Effect of consecutive daily transfers on S. pyogenes M18AGN and M18NAGN adherence to growing (24-h-old) human skin cellmonolayers. Symbols: O, CFU; *, percent receptive host cells; , AGN cocci; - - - -, NAGN cocci. Values are averages ± standarddeviations for two separate experiments.

human skin cells. Figure 3 shows the loss of adherence (CFUper 100 host cells and percentage of host cells with cocci)with successive transfers of the S. pyogenes M18AGNclinical isolate at its mid-logarithmic phase of growth. Thenumber of transfers indicated does not include those fewrequired after patient isolation and preparation for storage(see Materials and Methods). Adherence remained nearlyconstant for approximately 15 consecutive transfers in THBbefore declining abruptly. After 20 daily transfers, adher-ence had decreased to almost that of the M18NAGN clinicalisolate. The adherence of the NAGN isolate remained lowand unchanged over this same time period (see also Fig. 1).

Cytotoxicity of LTA for human skin cell and mouse fibro-blast monolayers. Two methods were used to assess theeffects of exposure to LTA from S. pyogenes M18AGN andM18NAGN for 24 h on primary human skin cell monolayers.Figure 4A compares the cytotoxicities of LTA from thesetwo coccal sources by measurement of the release of labeledprotein from prelabeled host cells. Since the mechanism ofcell death during exposure to LTA is not known, percenttoxicity probably includes cell death (detachment) and cellleakage of labeled protein at only the lower concentrations ofLTA tested. As indicated above, the chain lengths of theLTAs from both M18 isolates differed considerably. On anequal-weight basis, LTA from the AGN isolate was signifi-cantly more cytotoxic than was that from the NAGN isolate,even though approximately 0.5 as many molecules of theformer were used (based on the probable molecular weightdifference). This difference is indicated by the equal amountsof each LTA required for maximal cytotoxicity (50 ,ug perwell) but the significant difference in the ultimate levels oftoxicity realized (i.e., 90 and 60% for AGN and NAGNLTAs, respectively, at the higher concentrations of LTAtested). The results of the paired Student t test for the LTAcytotoxicities of these two coccal isolates revealed a signif-icant difference (P s 0.001).

Figure 4B shows the death of human skin monolayerscaused by LTA as measured by dye exclusion. There was nodifference in cytotoxicity between the LTAs from the twoclinical isolates, with both exhibiting maximal toxicity at 5,ug/ml. However, considerably less LTA was required formaximum cell death in this method than as judged by the

release of labeled protein (Fig. 4A). Greater than 90%destruction of the cells occurred, with those still attachedbeing without cytoplasm, staining intensely, and being inca-pable of recovery after LTA removal by washing and refeed-ing.

Past studies in this laboratory were done with mousefibroblast cells for assessing death caused by coccal LTA.To compare current with past findings, we tested LTAs fromthe M18AGN and M18NAGN isolates and that from anearlier nephritogenic laboratory strain of S. pyogenes (type12) with established mouse fibroblast monolayers (Fig. 5) bythe dye exclusion method. Contrary to the results obtainedwith primary human skin cells (Fig. 4B), differences incytotoxicity (i.e., cell death) in the dye exclusion methodwere observed for the LTAs from the M18AGN andM18NAGN isolates. For example, at the 50% lethal dose,LTA from the AGN isolate was 33 to 50% more cytotoxicthan was that from the NAGN isolate of the same serotype.Finally, the same degree of cytotoxicity for these mousecells was only observed for LTA from the laboratory strainof S. pyogenes (type 12) and LTA from the M18NAGNclinical isolate (Fig. 5).Enhancement of secretion of LTA by penicillin in S. pyo-

genes M18AGN and M18NAGN. Measurement of the en-hancement by penicillin of LTA secretion was done withintact cells labeled with glycerol. The total labeled glycerolcomponents secreted by each of the organisms was esti-mated first, followed by that portion of this milieu alsocapable of binding to sheep erythrocytes (e.g., labeled LTA,lipids, etc.). Final assessment of only the LTA present wasachieved by direct PHA. There was no significant differencein the secretion of the total labeled glycerol components bythe two clinical isolates until the highest concentration ofpenicillin (10 ,ug/ml) was reached (Fig. 6A). However, amajor change was evident in the content of secreted material(e.g., labeled LTA, lipids, etc.) able to bind to sheeperythrocytes (Fig. 6B). This was clearly evident from theonset and continued over the range of concentrations ofpenicillin tested, reaching a peak at 10 ,ug of penicillin per mlof medium. At this antibiotic concentration, the erythrocyte-binding capacity of material secreted by the AGN coccuswas 200% greater than that of the NAGN coccus. By

VOL. 58, 1990

on Decem

ber 13, 2020 by guesthttp://iai.asm

.org/D

ownloaded from

3784 LEON AND PANOS

CDnw0l

LL IILz

x-I

z49

0 25 50 75 100 125 150 1 75 200

LTA CONCENTRATION (Ag/ml)

100 -

cr-

00ILZ

J0

0-i(0C.

z4

80 -

60-

40 -

20 -

0

I0

B

5 1 0

LTA CONCENTRATION (jig

FIG. 4. Cytotoxicity of purified LTA from S. pyhuman skin cell monolayers. (A) Percent cytotoxicilwith [35S]methionine (see the text). (B) Percent suskin cells as determined by the dye exclusion methiAGN cocci; *, NAGN cocci. Age of human skin c24 h. Values are averages for two separate expelduplicate (standard deviations are perceptible in papanel B).

comparison, the actual amounts of LTA rehisolates differed even more, with that releaseAGN isolate continuing to increase as the coantibiotic was increased; e.g., at 10 ,ug of per130 and 40 ,g of LTA per ml of medium were sAGN and NAGN cocci, respectively (Fig. 6Cparticular interest was the amount of LTA siAGN isolate. It was more than four timespresence than in the absence of the highest copenicillin tested (130 versus 32 ,ug of LTA percompare Table 1 and Fig. 6C). Similar diffmagnitude or amount of LTA secreted by thein the presence and absence of these conpenicillin did not occur.

Differences in hydrophobicity between clinic;

4 I

60-j-j

Cf 404

0m\

. 20

0~0 25 50 75 100 125 150 1 75 200

LTA CONCENTRATION (jig)

FIG. 5. Cytotoxicity of LTA for mouse fibroblast monolayers intissue cultures as determined by the dye exclusion method. PurifiedLTAs were from clinical isolates of S. pyogenes M18AGN (O) andM18NAGN (-) and from an established laboratory strain ofAGN S.pyogenes type 12 (A). Values are averages for five separate exper-iments done in duplicate (standard deviations are not always per-ceptible).

pyogenes M18. Figure 7 shows the hydrophobicity in xyleneof intact, mid-logarithmic-phase cells of the virulent AGNand NAGN isolates. After six subtransfers, the hydropho-bicity of the AGN coccus was greater than was that of theNAGN coccus (approximately double). However, after 26subtransfers, this difference in hydrophobicity increasedconsiderably (approximately fourfold). Also, whereas thehydrophobicity of the AGN isolate did not change withsubtransferring, that of the NAGN isolate declined morethan twofold. Finally, identical results were observed when

1'5 * 2'0 stationary-phase cells of each isolate were compared. Simi-larly aged cultures of E. coli, a negative control, were not

i/mi) hydrophobic. It should be mentioned that these GAS isolateswere encapsulated (India ink capsule procedure) and re-

togenes M18 for mained so even after being washed and suspended in PUMty as determined buffer (see Materials and Methods).Irvival of humanod. Symbols: O,cell monolayers, DISCUSSIONriments done in This is the first comparison showing the significantly

greater production of LTA by a virulent AGN GAS than byits virulent and serologically identical but NAGN counter-part. LTA is known to be involved in the adherence of GAS

eased by both to human host cells. It is also cytotoxic for host cells. ThisId by only the dual involvement of LTA plus its increased production and)ncentration of secretion by select pathogenic GAS may be contributingnicillin per ml, factors in necrotizing fascitis, the most common soft tissuesecreted by the infection in the recent resurgence of certain GAS infections; P c 0.05). Of (35).ecreted by the LTA is capable of an array of biological activities. Itlarger in the stimulates bone resorption (13) and DNA synthesis in lym-

ncentration of phocytes (31), suppresses immune responses (19), inhibitsml of medium; complement activity (30), induces nephritis (40) and arthritiserences in the (25), and causes the release of lysosomal enzymes fromNAGN isolate macrophages (12). It induces the Shwartzman reaction.centrations of Biochemically, nephritogenic GAS LTA causes the forma-

tion and accumulation of defective (i.e., partially hydroxy-al isolates of S. proline-free) collagen in mouse fibroblast monolayers in

INFECT. IMMUN.

on Decem

ber 13, 2020 by guesthttp://iai.asm

.org/D

ownloaded from

S. PYOGENES CLINICAL ISOLATES AND LTA 3785

E

X 20000E

a. 100000

A

i-m0IL00r

sRPENICILLIN CONCENTRATION (gig/mI)

100001 B

8000 -

E

X 6000-E

E 4000-

O 2000

PENi C-I4 3 l -2 10 10g°PENICILLIN CONCENTRATION (jig/ml)

E'a 140

120 -

I

> 100-

z 80-0

60-r

z 40-w

0

z 20-00

0

I.-

IC

0 1 4 1 0r3 1 -2 i0 l 1° 1 1

PENICILLIN CONCENTRATION (gtg/mi)

FIG. 6. Assays establishing enhancement of the secretion ofradiolabeled LTA from S. pyogenes M18 by penicillin. Intact cellswere labeled with [3H]glycerol (see the text). (A) [3H]glycerolsecretion. (B) Erythrocyte binding of secreted material. (C) LTAsecretion (as determined by direct PHA). Symbols: O, AGN cocci;U, NAGN cocci.

vitro (15, 16). Basement membrane thickening, fusion ofepithelial foot processes, etc. also occur in the mouse kidneyafter repeated injections of native LTA from nephritogenicS. pyogenes (17). Likewise, GAS LTA is believed to regu-late diverse functions of mammalian cells by phosphorylat-ing the tyrosine residues of certain proteins (8). Therefore,profuse production and secretion of LTA can only potentiatethe pathogenesis of coccal disease.Glucose was found to be a structural component of the

hydrophilic and hydrophobic portions of LTA of S. pyo-genes type 12 (33, 34). Also, an earlier study reported anamino sugar as the only hexose in glycerol TA from type 3GAS (18). Glucose was not detected in the LTAs of thevirulent GAS M18AGN and M18NAGN isolates in thisinvestigation. Instead, only the former contained smallamounts of two amino sugars. The significance of thisdifference in LTA cytotoxicity is obscure, since LTA fromthe GAS M18NAGN isolate (without detectable carbohy-

0 25 50 75 100 125 150 175 200

XYLENE CONCENTRATION (i)i)FIG. 7. Changes in the hydrophobicity of S. pyogenes M18 in

xylene after 6 (-) and 26 (----) consecutive daily transfers.Symbols: -, AGN cocci; *, NAGN cocci; A, E. coli (control).Values are averages for two separate experiments done in duplicate(standard deviations are not always perceptible).

drate) was also cytotoxic, albeit to a lesser degree (seebelow). Also, in an earlier study, S. pyogenes type 12 LTAwith a minimal carbohydrate content was highly cytotoxicfor a variety of human and animal cells in vitro (27).Nevertheless, these findings imply that the carbohydratecomposition within the hydrophilic portion of the LTA of theGAS is not uniform. It has also been shown that thehydrophilic chain length of LTA can profoundly affect itsbiological activity (7). Thus, the significantly longer chainlength of LTA from the AGN isolate may be one of severalfactors instrumental in its increased cytotoxicity. Similardifferences in length between LTAs from virulent (30 to 35glycerol phosphate units) and avirulent (10 to 12 glycerolphosphate units) human isolates of S. agalactiae type IIIhave been documented (24). A phosphoglucolipid was estab-lished as the hydrophobic component of the LTA of S.pyogenes type 12 (33). However, the compositional differ-ences between the virulent GAS M18 isolates indicate thatthe complex lipid of the LTA from S. pyogenes need not bea phosphoglucolipid.The differences in the adherence of different serotypes of

virulent AGN and NAGN GAS to growing human skin cellswere striking. This limited study indicated a correlationbetween virulent strains causing AGN and adherence tohuman skin cells. The lack of a similar affinity of the NAGNstrains may be due to differences in LTA production and/orexposure (or availability) on the surface of the organism. Inturn, it may be related to the structural and compositionaldifferences already mentioned between the GAS M18 iso-lates. Clearly, although major differences have now beendocumented in the LTAs from these virulent GAS isolates, amore detailed chemical and structural characterization, es-pecially of the hydrophobic portion, is still needed. Finally,the results of treatment of host cells with coccal LTA orcocci exposed to anti-LTA serum proved that LTA is stillinvolved in the adherence of fresh clinical isolates of AGNGAS to human cells in vitro.Only a certain population within the randomly growing

human skin cell monolayer was capable of binding AGNGAS, indicating a limited cyclic expression of receptors on

VOL. 58, 1990

i

-.-a

on Decem

ber 13, 2020 by guesthttp://iai.asm

.org/D

ownloaded from

3786 LEON AND PANOS

the surfaces of these primary human cells in vitro. While thishas indeed been shown to be the case with certain humancells infected with virulent GBS, it is not true of all humancells. Human fetal lung cells, for example, do not exhibit thisexpression (10, 20).

Continuing to transfer an AGN GAS isolate in vitro led toa sudden and almost complete loss of adherence. These dataresembled those obtained with the virulent GBS (20). Likevirulence, adherence is a transitory property of pathogenicGAS which needs to be preserved if comparative studies ofvirulent isolates are to be meaningful. Fortunately, adher-ence of the AGN coccus in vitro remained fairly constant fora period of time before declining rapidly. Thus, a trueassessment of this characteristic is still possible after pri-mary isolation and multiple but not indefinite transfer invitro.The cytotoxicity of LTA in vitro was established in 1978

(6). Some earlier evidence, however, had suggested thatLTA might also be mobile in vivo. An affinity of LTA forkidney tissue was shown by inducing nephrocalcinosis inrabbits after injection of GAS LTA. LTA migration (andaffinity) was demonstrated by labeled antibody (40). A recentstudy with mice injected by different routes with purifiedcoccal LTA tended to confirm this ability to migrate (17).Even earlier, in 1969, streptococcal membrane constituents(and presumably LTA) in the glomeruli of human patientswith acute poststreptococcal glomerulonephritis were de-tected with fluorescent antibody (37). LTA is now known tofunction as a carrier for streptolysin S. Also, serum albuminbinds LTA reversibly and in a specific way and is believed tobe a carrier of this amphiphile (32). Therefore, secreted LTAis probably also involved in coccal pathogenesis. Directcontact of membranous and secreted LTA with the host cellsurface occurs during and after coccal adherence. Thus,direct transfer of LTA to a host cell during adherence seemslikely. Under these conditions, the time required for cyto-toxic effects to begin would be less than that followingsecretion and transportation of LTA to a distant susceptibleorgan (e.g., the kidneys). Thus, secretion and transporta-tion, plus the direct transfer of LTA to a host cell duringadherence, can enhance the destructive effects of this am-phiphile in vivo.

It has been proposed that coccal LTA in a micellar statemay be necessary for disruption of the erythrocyte mem-brane and for destruction of eucaryotic cells in vitro. Criticalmicelle concentrations of LTAs from several bacteria werecalculated to be 25 to 60 ,ug/ml in PBS (5). The amount ofLTA (32 ,ug/ml) normally secreted by the M18AGN GASisolate (and increased with penicillin; Fig. 6) is within thisrange. However, arguing against a critical micellar concen-tration is the fact that appreciably smaller quantities of LTA(8 to 15 ,ug/ml) also caused extensive human cell damage invitro (see Results).

It was established that the GAS secrete LTA and deacy-lated LTA and that penicillin stimulates this process (1, 11,14). Also, coccal adherence to host cells is decreased withthe loss of LTA. This release of LTA (and lipids) bypenicillin is believed to be an "active process rather than acorrelate of viability loss, since streptococci tolerant topenicillin also exhibit penicillin-induced release of cell sur-face components" (11). The present data not only confirmthe enhanced secretion of LTA by penicillin but illustratethat the increase is most pronounced in only the M18AGNGAS isolate. Finally, while others observed the secretion ofdeacylated LTA by GAS, we did not detect this product byserological means in the medium of the virulent M18 clinical

isolates in the presence or absence of penicillin. Apparentlynot all GAS secrete deacylated LTA.The dramatic differences obtained with the two methods

for quantitating LTA cytotoxicity suggest that host celldeath (by dye exclusion) occurs before the release of cellularprotein with smaller quantities of LTA, which may resultfrom the inhibition of a select enzyme activity (16). Higherconcentrations of LTA lead to cell protein leakage after celldeath. Thus, a dual cytotoxic effect of LTA which entails aphysical perturbation of the host cell membrane is postu-lated. The differences also suggest that different host cellshave different sensitivities to LTA. These findings agree withprevious results showing greater changes in cell morphologythan in protein release with lower concentrations of LTA.Thus, while bacteria release little or no protein after penicil-lin treatment, eucaryotic cells lose more protein as theconcentration of LTA increases.Hydrophobicity studies with other streptococci showed

that slowly growing cells were more hydrophobic than wererapidly growing cells (28) and that bacterial hydrophobicitycan be markedly affected by the medium. Since it is gener-ally agreed that a number of mechanisms particularly relatedto surface characteristics (surface charge, hydrogen bond-ing, ligands, and hydrophobic bonding) play a role in bacte-rial adherence, the cell surface hydrophobicities of thesecoccal clinical isolates were compared. Contrary to previousfindings, the more rapidly growing AGN isolate was morehydrophobic than was the slowly growing S. pyogenesM18NAGN in the same medium. Since the AGN coccuscontained more membranous LTA and adhered better tohuman skin cells, suggesting a greater exposure of its surfaceLTA, these hydrophobicity results were expected. How-ever, unexpected was the lack of change in the hydropho-bicity with a loss in the adherence of the AGN isolate aftercontinued daily subtransfers. Apparently, the cell surfacehydrophobicity of the AGN isolate is not a determiningfactor in its adherence to growing human skin cell monolay-ers in vitro. Nevertheless, the difference in the hydropho-bicity of virulent M18AGN and M18NAGN GAS is not dueto the presence or absence of a capsule.These collective studies illustrate important biological and

biochemical differences between AGN and NAGN clinicalisolates of GAS. Also, for the AGN coccus they reveal achange in a parameter directly related to pathogenesisshortly after primary isolation. Such an alteration must beconsidered for comparative studies in vitro to be meaning-fully related to differences in GAS pathogenesis in vivo.

ACKNOWLEDGMENTS

This investigation was funded by the Department of Microbiologyand Immunology, Jefferson Medical College, Thomas JeffersonUniversity.We thank E. Kaplan for supplying the clinical isolates used in this

investigation.

LITERATURE CITED1. Alkan, M. L., and E. H. Beachey. 1978. Excretion of lipoteichoic

acid by group A streptococci. J. Clin. Invest. 61:671-677.2. Beachey, E. H. 1981. Bacterial adherence: adhesin receptor

interactions mediating the attachment of bacteria to mucosalsurfaces. J. Infect. Dis. 143:325-345.

3. Centers for Disease Control. 1987. Acute rheumatic fever-Utah. Morbid. Mortal. Weekly Rep. 36:108-115.

4. Centers for Disease Control. 1988. Acute rheumatic fever at anavy training center-San Diego, California. Morbid. Mortal.Weekly Rep. 37:101-104.

5. Courtney, H. S., W. A. Simpson, and E. H. Beachey. 1986.

INFECT. IMMUN.

on Decem

ber 13, 2020 by guesthttp://iai.asm

.org/D

ownloaded from

S. PYOGENES CLINICAL ISOLATES AND LTA 3787

Relationship of critical micelle concentrations of bacterial lipo-teichoic acids to biological activities. Infect. Immun. 51:414-418.

6. DeVuono, J., and C. Panos. 1978. Effect of L-form Streptococ-cus pyogenes and of lipoteichoic acid on human cells in tissueculture. Infect. Immun. 22:255-265.

7. Fiedler, F. 1981. On the participation of lipoteichoic acid in thebiosynthesis of wall teichoic acid, p. 195-208. In G. D. Shock-man and A. J. Wicken (ed.), Chemistry and biological activitiesof bacterial surface amphiphiles. Academic Press, Inc., NewYork.

8. Ganguly, C. L., J. B. Dale, H. S. Courtney, and E. H. Beachey.1985. Tyrosine phosphorylation of a 94-kDA protein of humanfibroblasts stimulated by streptococcal lipoteichoic acid. J. Biol.Chem. 260:13342-13346.

9. Gaworzewska, E., and G. Colman. 1988. Changes in the pattern

of infection caused by Streptococcus pyogenes. Epidemiol.Infect. 100:257-269.

10. Goldschmidt, J. C., Jr., and C. Panos. 1984. Teichoic acids ofStreptococcus agalactiae: chemistry, cytotoxicity, and effecton bacterial adherence to human cells in tissue culture. Infect.Immun. 43:670-677.

11. Gutmann, L., and A. Tomasz. 1982. Penicillin-resistant andpenicillin-tolerant mutants of group A streptococci. Antimicrob.Agents Chemother. 22:128-136.

12. Harrop, P. J., R. L. O'Grady, K. W. Knox, and A. J. Wicken.1980. Stimulation of lysosomal enzyme release from macro-

phages by lipoteichoic acid. J. Peridontal Res. 15:492-501.13. Hauseman, E., 0. Luderitz, K. Knox, A. Wicken, and N.

Weinfeld. 1975. Structural requirements for bone resorption byendotoxin and lipoteichoic acid. J. Dent. Res. 54:B94-B99.

14. Horne, D., R. Hakenbeck, and A. Tomasz. 1977. Secretion oflipids induced by inhibition of peptidoglycan synthesis in strep-

tococci. J. Bacteriol. 132:704-717.15. Leon, O., and C. Panos. 1983. Cytotoxicity and inhibition of

normal collagen synthesis in mouse fibroblasts by lipoteichoicacid from Streptococcus pyogenes type 12. Infect. Immun.40:785-794.

16. Leon, O., and C. Panos. 1985. Effect of streptococcal lipotei-choic acid on prolyl hydroxylase activity as related to collagenformation in mouse fibroblast monolayers. Infect. Immun. 50:745-752.

17. Leon, O., and C. Panos. 1987. An electron microscope study ofkidney basement membrane changes in the mouse by lipotei-choic acid from Streptococcus pyogenes. Can. J. Microbiol.33:709-717.

18. Matsuno, T., and H. D. Slade. 1970. Composition and propertiesof a group A streptococcal teichoic acid. J. Bacteriol. 102:747-752.

19. Miller, G., A. J. Urban, and R. W. Jackson. 1976. Effects of a

streptococcal lipoteichoic acid on host responses in mice.Infect. Immun. 13:1408-1417.

20. Miyazaki, S., 0. Leon, and C. Panos. 1988. Adherence ofStreptococcus agalactiae to synchronously growing human cellmonolayers without lipoteichoic acid involvement. Infect. Im-mun. 56:505-512.

21. Nealon, T. J., E. H. Beachey, H. S. Courtney, and W. A.Simpson. 1986. Release of fibronectin-lipoteichoic acid com-

plexes from group A streptococci with penicillin. Infect. Im-mun. 51:529-535.

22. Nealon, T. J., and S. J. Mattingly. 1983. Association of elevatedlevels of cellular lipoteichoic acids of group B streptococci withhuman neonatal disease. Infect. Immun. 39:1243-1251.

23. Nealon, T. J., and S. J. Mattingly. 1984. Role of cellularlipoteichoic acids in mediating adherence of serotype III strainsof group B streptococci to human embryonic, fetal, and adult

epithelial cells. Infect. Immun. 43:523-530.24. Nealon, T. J., and S. J. Mattingly. 1985. Kinetic and chemical

analyses of the biologic significance of lipoteichoic acids inmediating adherence of serotype III group B streptococci.Infect. Immun. 50:107-115.

25. Ne'eman, J., and I. Ginsberg. 1972. Red cell sensitizing antigenof group A streptococci. Immunological and immunopathologi-cal properties. Isr. J. Med. Sci. 8:1807-1816.

26. Ofek, I., E. H. Beachey, W. Jefferson, and G. L. Campbell. 1975.Cell membrane-binding properties of group A streptococcallipoteichoic acid. J. Exp. Med. 141:990-1003.

27. Panos, C. 1986. Macromolecular synthesis, survival, and cyto-toxicity of an L-form of Streptococcus pyogenes, p. 59-97. In S.Madoff (ed.), The bacterial L-forms. Marcel Dekker, Inc., NewYork.

28. Rogers, A. H., K. Pilowsky, and P. S. Zilm. 1984. The effect ofgrowth rate on the adhesion of the oral bacteria Streptococcusmutans and Streptococcus milleri. Arch. Oral Biol. 29:147-150.

29. Rosenberg, M., D. Gutnick, and E. Rosenberg. 1980. Adherenceof bacteria to hydrocarbons: a simple method for measuringcell-surface hydrophobicity. FEMS Microbiol. Lett. 9:29-33.

30. Silvestri, L. J., K. W. Knox, A. J. Wicken, and E. M. Hoffman.1979. Inhibition of complement-mediated lysis of sheep eryth-rocytes by cell-free preparations from Streptococcus mutansBHT. J. Immunol. 122:54-60.

31. Simpson, W. A., J. B. Dale, and E. H. Beachey. 1982. Cytotox-icity of the glycolipid region of streptococcal lipoteichoic acidfor cultures of human heart cells. J. Lab. Clin. Med. 99:118-126.

32. Simpson, W. A., I. Ofek, and E. H. Beachey. 1980. Binding ofstreptococcal lipoteichoic acid to the fatty acid binding sites ofserum albumin. J. Biol. Chem. 255:6092-6097.

33. Slabyj, B. M., and C. Panos. 1973. Teichoic acid of a stabilizedL-form of Streptococcus pyogenes. J. Bacteriol. 114:934-942.

34. Slabyj, B. M., and C. Panos. 1976. Membrane lipoteichoic acidof Streptococcus pyogenes and its stabilized L-form and theeffect of two antibiotics upon its cellular content. J. Bacteriol.127:855-862.

35. Stevens, D. L., M. H. Tanner, J. Winship, R. Swarts, K. M. Ries,P. M. Schlievert, and E. Kaplan. 1989. Severe group A strepto-coccal infections associated with a toxic shock-like syndromeand scarlet fever toxin A. N. Engl. J. Med. 321:1-7.

36. Tomlinson, K., 0. Leon, and C. Panos. 1983. Morphologicalchanges and pathology of mouse glomeruli infected with astreptococcal L-form or exposed to lipoteichoic acid. Infect.Immun. 42:1144-1151.

37. Treser, G., M. Semar, M. McVicar, M. Franklin, A. Ty, I. Sagel,and K. Lange. 1969. Antigenic streptococcal components inacute glomerulonephritis. Science 163:676-677.

38. Veasy, L. G., S. E. Wiedmeier, G. S. Orsmond, H. D. Rutten-berg, M. M. Boucek, S. J. Roth, V. F. Tait, J. A. Thompson,J. A. Daly, E. L. Kaplan, and H. R. Hill. 1987. Resurgence ofacute rheumatic fever in the intermountain area of the UnitedStates. N. Engl. J. Med. 316:421-427.

39. Wald, E. R., B. Dashefsky, C. Feidt, D. Chiponis, and C. Byers.1987. Acute rheumatic fever in western Pennsylvania and thetri-state area. Pediatrics 80:371-374.

40. Waltersdorf, R. L., B. A. Fiedel, and R. W. Jackson. 1977.Induction of nephrocalcinosis in rabbit kidneys after long-termexposure to a streptococcal teichoic acid. Infect. Immun. 17:665-667.

41. Wicken, A. J. 1980. Structure and cell membrane-binding prop-erties of bacterial lipoteichoic acid and their possible role inadhesion of streptococci to eukaryotic cells, p. 137-158. InE. H. Beachey (ed.), Bacterial adherence, receptors, and rec-ognition, series B, vol. 6. Chapman & Hall, Ltd., London.

VOL. 58, 1990

on Decem

ber 13, 2020 by guesthttp://iai.asm

.org/D

ownloaded from