Clin Exp Immunol 1993; 93:189-194

Tumour necrosis factor production in vivo and in vitro in response toParacoccidioides brasiliensis and the cell wall fractions thereof

F. FIGUEIREDO, L. M. C. ALVES* & C. L. SILVA* Department of Pathology and *Department of Parasitology,Microbiology and Immunology, School of Medicine of Ribeirao Preto, University of Silo Paulo, Sao Paulo, Brazil

(Acceptedfor publication 8 April 1993)

SUMMARY

Tumour necrosis factor (TNF) was detected in serum from mice challenged with Paracoccidioidesbrasiliensis. The serum TNF levels of mice challenged with an avirulent strain were significantlyhigher than those of mice challenged with a virulent strain, and the same was observed for the TNFlevels of mice challenged with a cell wall fraction (Fl) from the two fungal strains. Fraction F1consisted of chitin and fl-glucan; but although the chitin contents were similar for the two strains, theavirulent strain allowed a greater content of fl-glucan. The fl-glucan, purified from both strains,increased serum TNF levels in an identical dose-dependent manner, whereas purified chitin did notinduce serum TNF levels. P. brasiliensis, the Fl fractions and f,-glucan induced macrophages tosecrete TNF in vitro. The differences in TNF levels, induced by the different fungal strains, were

correlated with the f-glucan concentrations in the cell walls of both the avirulent and virulent strainsof P. brasiliensis. These findings support a role for TNF in the pathogenicity of P. brasiliensis.

Keywords Paracoccidioidomycosis Paracoccidioides brasiliensis tumour necrosis factorP-glucan

INTRODUCTION

Paracoccidioidomycosis (Pbmycosis) is a chronic and granulo-matous disease caused by the imperfect dimorphic fungusParacoccidioides brasiliensis [1]. The disease involves mucosae,lymph nodes, the reticuloendothelial system, the pulmonaryparenchyma and other internal organs [2].

The pathogeny of Pbmycosis depends in part upon at leasttwo factors, i.e. the degree and kind of host resistance, and thevirulence and pathogenicity of the fungus. Many people areexposed to the organism, but only a small number developclinical symptoms, suggesting the presence of innate andacquired resistance mechanisms [3]. Thus, cell-mediated immu-nity is generally a more significant factor than humoralimmunity, and activated macrophages appear to provide themajor line of defence [4]. The virulence and pathogenicity of thefungus are correlated with several components of the fungal cellwall. Cell wall a-glucan seems to be related to virulence [5]; and,cell wall JJ-glucan can induce leucocyte migration and agranulomatous inflammatory reaction [6,7].

The inflammatory response induced by the Pbmycosis ischaracterized by macrophage activation and granuloma forma-tion. The granuloma is a mononuclear tissue (macrophage)inflammatory reaction evoked by cell wall components (fraction

Correspondence: Dr Celio L. Silva, Department of Parasitology,Microbiology and Immunology, School of Medicine of Ribeirao Preto,USP, 14049-900 Ribeirao Preto, Sdo Paulo, Brazil.

Fl), and represents a balance between macrophage differentia-tion and destruction of the causal agent [7]. The full develop-ment of granulomas in cell-mediated immunity requires a Tlymphocyte immune response, and involves the continuedattraction of macrophages and their subsequent differentiationinto epithelioid cells capable of preventing dissemination of thecausal agent and of efficient killing [8-11].

Immunological and inflammatory events can be induced bythe P. brasiliensis and their cell wall components. Macrophagesplay a pivotal role in these events. They are the main source oftumour necrosis factor (TNF), a cytokine with multiple biologi-cal activities [12,13], a number of which are involved in variousaspects of the inflammatory process [14,15]. TNF levels havebeen associated with various infectious diseases [16-20], andhigh concentrations of circulating TNF have been demon-strated by our group in Pbmycosis patients [21]. The presentstudy was performed to investigate whether TNF is produced inmice in response to injection of P. brasiliensis or its cell wallcomponents, and whether TNF might be a determining factorinvolved in the genesis of the inflammatory response observed inPbmycosis infection.

MATERIALS AND METHODS

Fungal strain and culture conditionsThe two strains of P. brasiliensis, Pbl8 and Pb265, have beenpreviously described [22]. Yeast-phase fungal cells were cultured

189

F. Figueiredo, L. M. C. Alves & C. L. Silva

at 370C for 28 days in Fava Netto's medium [23]. The cells werethen harvested, autoclaved at 120'C, 150 kPa and washedseveral times with distilled water.

Preparation of cell wallsThe washed yeast cells were disrupted by ultrasonic vibration at200 W for 3 min; this process was repeated six times. Extensivedisruption was confirmed by microscopy. The walls werecollected and washed three times with distilled water bycentrifugation at 5000 g for 5 min. Lipids were extracted bysoaking the walls in chloroform/methanol (2: 1, v/v) withstirring at room temperature for 2 h. The extracts wereseparated by centrifugation at 5000g for 5 min and the insolubleresidue was re-extracted three more times as described above.The resulting insoluble cell residue was named the cell wallfraction.

Fractionation of cell wallsPartial fractionation was done by alkaline extraction as de-scribed by Kanetsuna & Carbonell [24]. Briefly, cell walls weresuspended in 1 N NaOH (10 mg/ml) and gently stirred at roomtemperature for 1 h. After centrifugation at 5000 g for 10 min,the supernatants were collected, and the procedure was repeatedfour times combining all the supernatants. The alkali-insolublesediment was washed with water until it reached pH 7-0 and thenwashed with ethanol, followed by acetone and diethyl ether. Theresulting white powder was named the Fl fraction ofeither Pbl 8(fraction F1 Pbl8) or Pb265 (fraction Fl Pb265). The pooledsupernatants were neutralized with acetic acid and left to standovernight at 4°C, after which a precipitate formed. Thesuspension was then centrifuged as before. The precipitate andthe supernatant were collected, dialysed separately againstdistilled water and freeze-dried, yielding fractions F2 (ac-glucan)and F3 (galactomannan) respectively.

Preparation of fi-glucan and chitin,B-glucan and chitin were purified by enzymatic treatment of theFl fractions. For ,B-glucan recovery, 10 mg aliquots of the F1fractions were incubated at 37°C for 24 h with 1 ml of 0-6 mg/mlchitinase in 0 05 M sodium acetate buffer (pH 5 0) and 5 91toluene. After incubation, the supernatant was removed and theresidue was further treated with chitinase. The procedure wasrepeated five times and the resulting sediment was washed withwater and ethyl alcohol and dried in an oven at 37°C. Protease(pronase CB; Calbiochem) was incubated with cell wall samplesfor 5 h at 37°C in a 0 05 M Tris (hydroxymethyl) aminomethane-hydrochloride buffer, pH 7-5, containing 555 pg/ml CaCl2, 50pg/ml cycloheximide, and 25 yg/ml chloramphenicol. Thechitinase-digested Fl fractions were also treated with trypsin(100 ,g/ml) in 0-01 M potassium phosphate buffer (pH 7 2) for 4h at 37°C. After trypsin treatment, the suspensions werecentrifuged at 20000 g for 10 min, the supernatant solution wasdiscarded, and the cell wall pellet was washed repeatedly withdistilled water.

For chitin preparation, Fl fractions were treated withpronase and trypsin as described above, followed by incubationwith cellulase (EC 3.2.1.4; Sigma; one unit liberated 1 0 pmole ofglucose from cellulose in 1 h at pH 5-0 at 370C) and ,B-glucosidase (EC 3.2.1.21; Sigma; one unit liberated 1 0 ymole ofglucose from salicin per min at pH 5-0 at 37°C). One millilitre0-05 M acetate buffer, pH 5-0, was added to 1 mg of the Fl

fractions, and the mixture was incubated at 370C. After 24 hincubation, the reaction mixture was placed in a boiling-waterbath for 5 min to inactivate the enzymes. Glucose was estimatedby chemical or chromatographic methods. The concentrationsof 13-glucan and chitin recovered from FlPbl8 and FlPb265,respectively, were determined gravimetrically (four determina-tions) and the materials were characterized by infrared spectro-scopy.

AnimalsSpecific pathogen-free, inbred C57B1/6 mice (6 weeks old) wereused in all experiments. Mice were injected intraperitoneallywith the preparations to be tested. At appropriate times afterchallenge, mice were bled and the serum was collected. Serumsamples were stored at - 70'C until the time for TNF measure-ment.

Preparation ofmacrophage supernatantThioglycollate-elicited macrophages were suspended in Hanks'balanced salt solution (HBSS) to remove non-adherent cells,and cultured with fresh RPMI 1640 medium. The cell mono-layers were routinely found to contain > 95% macrophages asdetermined by morphology using a Giemsa stain or a histoche-mical assay for non-specific esterase. After 24 h, macrophageswere treated with the various stimuli for the time indicated; thesupernatants were collected at 24 h, filtered (0-2 ,m pore sizemembrane), aliquoted, and stored at - 70°C. All reagents andmedia/serum used for macrophage cultures contained less than0125 ng of endotoxin (lipopolysaccharide (LPS)) per ml asquantified by the Limilus amoebocyte lysate assay.

Measurement of TNF levelsTNF was assayed using L929 cells [21,25]. Briefly, 3 x 105 cells/ml were suspended in supplemented RPMI 1640 mediumcontaining 10% fetal calf serum (FCS) and 2 pg/ml actinomycinD, and 0 1 ml of the cell suspension was added to microtitrewells. RPMI 1640 medium (0-1 ml) containing differentamounts of recombinant murine TNF (standard) or dilutedserum samples was added to each well. The culture wasincubated at 37°C for 20 h in a tissue culture incubator. Themedium was removed by aspiration, and 0 I ml of crystal violet(0 1% in ethanol and water 1: 4) was added to each well. After 20min at room temperature, the crystal violet solution wasremoved by aspiration and the wells were washed three timeswith cold water. The wells were dried, and 0-1 ml of 95%methanol was added. The absorbance of each well was meas-ured at 490 nm with a microELISA autoreader (model BT-100;Biotech Research Labs, Rockville, MD). The amount ofbioactive TNF in the serum sample was determined usingrecombinant murine TNF as standard.

RESULTS

In initial experiments, we assessed the capacity of P. brasiliensiscells to induce TNF secretion in infected mice. Mice were

challenged with 106 CFU of strain Pbl8 or Pb265 of P.brasiliensis, and serum samples were collected at 2, 4, 8, 16 and32 days after challenge. Serum samples were then assayed forTNF (Fig. I a). Samples from five mice were tested at each point.TNF secretion was detectable after a 4-day latency, reached amaximum peak at 16 days for both strains of P. brasiliensis, and

190

TNFproduction in response to P. brasiliensis

2000 r-

E 1600- 1200 \

In1200 -900

~800 -600-

zH400 -300

0 0024 8 16 32 024 8 16 32

Time after inoculation (days)

Fig. 1. Tumour necrosis factor (TNF) units found in the serum ofC57B 1/6 mice after challenge with 106 cells of the Pb265 (A) or Pbl 8 (-)strain of Paracoccidioides brasiliensis (a) or after challenge with 100 Ygin 0 1 ml PBS of fractions FIPb265 (A) or FIPbl8 (-) (b). Values aremean + s.e.m. (n = 5).

1600 -

E

N.

ZL

z

1200 I-

800 -

0 2 4 8 16

Time after inoculation (days)

35

28

C

. 21

0

0

0°o 4

il

0 1

FIPbI8 FIPb265

Fig. 2. Comparison of f-glucan and chitin concentrations present infraction Fl isolated from the Pbl8 and Pb265 strains of Paracocci-dioides brasiliensis. Values are mean+ s.e.m. (n = 3). 0, fl-glucan;*, chitin.

still persisted at 32 days. Biological activity could be neutralizedby the addition of polyclonal rabbit anti-TNF antisera at alltime points.

When mice were challenged with cell wall fractions Fl (100pg in 0 1 ml of PBS) from both fungal strains (Fig. I b), serum

TNF levels increased to a mean of 840 U/ml (F I Pb265) and 440U/mI (Fl Pbl 8) by 4 days after challenge, which was significantcompared with control serum levels (150 U/ml). The serum TNFlevels measured on the 8th day were significantly increasedcompared with levels at 4 days (1360 U/ml and 760 U/ml forFlPb265 and FlPbl8, respectively). Serum TNF levels thendecreased to a mean of 900 U/ml (FlPb265) and 760 U/ml(Fl Pbl 8) by 16 days after challenge, and to 560 U/ml (FlPb265)and 440 U/ml (FIPbl 8) by 32 days after challenge, levels whichwere still significantly higher than in controls (150 U/ml).

Fig. 3. Tumour necrosis factor (TNF) units in the serum of C57B1/6mice at various times after challenge with 100 yg in 01 ml PBSof fl-glucan (A) or chitin (-) purified from fraction Fl-derivedParacoccidioides brasiliensis Pbl8. Values are mean+ s.e.m. (n= 5).

a)DI0-C

0

E)

u0

E

0

"I

ILLz

T

T

- LPS Pb265 PbI8

Fig. 4. Tumour necrosis factor (TNF) units in the supernatant of 106macrophages challenged for 3 h with the Pb265 or Pbl8 strain ofParacoccidioides brasiliensis at a ratio of 10 organisms/macrophage.The supernatants were collected 24 h after the challenge. Macrophageswere incubated with medium alone (negative control) or lipopolysac-charide (LPS) 10 ug/ml (positive control). Values are mean+s.e.m.(n =6).

191

400 F

n I32

I

F. Figueiredo, L. M. C. Alves & C. L. Silva

1200 r-

loooF

8001-

600 I-

400 [-

200

0 '

Fig. 5. Tumour necrosis factor (TNF) units in the supernatant of 106macrophages challenged for 3 h with different cell wall componentspurified from Pbl8. The supernatants were collected 24 h after thechallenge. Values are mean+s.e.m. (n=6). 0, Medium; *, fl-glucan;U, a-glucan; U, chitin; *, galactomannan.

1200

1000

0

.O 800

0.

CO0

-0

E 6000

AD

2-

200

0I

10 20 50 100

,3-glucan concentration (,ag/ml)

200

Fig. 6. Tumour necrosis factor (TNF) units in the supernatant of 106macrophages challenged for 3 h with f-glucan concentrations rangingfrom 10 to 200 Mg/ml. The supernatants were collected 24 h after thechallenge. Values are mean + s.e.m. (n = 6).

Biological activity could be neutralized by the addition ofpolyclonal rabbit anti-TNF antisera at all time periods. Serumsamples from F2- and F3-challenged mice (100 pg in 0 1 ml ofPBS) had no detectableTNF levels at any time point tested (datanot shown).

Chemical analysis of fraction Fl from the two strainsrevealed that it contained chitin and f3-glucan (Fig. 2). Differ-

ences between the Fl fractions of the two strains were quantita-tive, particularly in terms of f-glucan content. The F1 of theavirulent strain contained almost four times as much f-glucanas that from the virulent strain. In contrast, similar amounts ofchitin were found in the cell walls of both strains. To determinewhether increases in serum TNF were due to either of these twocell wall components, mice were challenged with 100 pg in 0 I mlPBS of fi-glucan or chitin purified from FlPbl 8 (Fig. 3). Whenmice were challenged with f-glucan, serum TNF levels reached amaximal peak at 16 days (1600 U/ml) and then decreased to amean of 450 U/ml by 32 days after challenge. These levels weresignificantly greater than the 150 U/ml formed in controls.Biological activity could be neutralized by the addition ofpolyclonal rabbit anti-TNF antisera at all time periods. Incontrast to the results with f-glucan, serum samples from chitin-challenged mice had no detectable TNF levels at any time pointtested. Similar results were obtained when mice were challengedwith P-glucan and chitin purified from FlPb265 (data notshown).

Since macrophages are a major source of TNF [14,26] andare known to play a key role in host defence against P.brasiliensis [27], further experiments were performed to deter-mine whether the insoluble polysaccharide fl-glucan present onthe cell wall of the fungi directly stimulate macrophages toproduce TNF in vitro.

First, peritoneal macrophages (106) from uninfected micewere challenged with Pbl8 or Pb265 for 3 h at a ratio of 10organisms/macrophage. Supernatants were collected at 24 hafter challenge and assayed for TNF secretion (Fig. 4). Macro-phages incubated with medium alone for negative control didnot secrete a significant amount of TNF (90 U/ml). However,macrophages incubated with medium plus 10 pg/ml of LPS(phenol extract of Escherichia coli 026: 136; Difco Labs, Detroit,MI) for positive control secreted large amounts ofTNF (890 U/ml). The supernatants from Pb265- and Pb18-challenged mac-rophages secreted detectable TNF levels, six-fold (540 U/ml)and 4-5-fold (410 U/ml) above the levels found in the negativecontrol.

Second, macrophages from uninfected mice challenged for3 h with 100 pg in 0-1 ml PBS of different cell wall componentspurified from Pbl8 (a-glucan, P-glucan, chitin and galactoman-nan), and the supernatants collected 24 h after challenge wereassayed for TNF secretion (Fig. 5). In multiple experiments,high levels of TNF (880 U/ml) were detected in supernatantswhen macrophages were challenged with f-glucan. Biologicalactivity could be neutralized by the addition ofpolyclonal rabbitanti-TNF antisera. Macrophages incubated with a-glucan,chitin or galactomannan did not secrete any significant amountsof TNF compared with macrophage incubated with mediumalone (negative control). A similar result was obtained whenmacrophages were challenged with these cell wall componentspurified from Pb265 (data not shown).

Third, we then attempted to determine whether TNFsecretion by macrophages challenged with fl-glucan was concen-tration-dependent. Macrophages were incubated for 3 h with ,B-glucan concentrations ranging from 10 to 200 Mg/ml and thesupernatant was assayed for TNF secretion. TNF secretion wasdose-dependent (Fig. 6) and required 100 pg/ml of fl-glucan formaximal induction (900 U/ml). Biological activity at all dosepoints could be neutralized by the addition of polyclonal rabbitanti-TNF antisera. Similar results were obtained when macro-

192

to0'

Im1'000E0

z

TNFproduction in response to P. brasiliensis

phages were challenge with fl-glucan purified from Pb265 (datanot shown).

DISCUSSION

In the present study, TNF was detected in serum from micechallenged with strains Pbl 8 and Pb265 (virulent and avirulent,respectively) ofP. brasiliensis. The results show that P. brasilien-sis infection is associated with substantially increased serum

TNF levels, and high TNF levels have been implicated in thepathogenesis of a number of other serious infectious diseases[16-20]. We recently demonstrated high levels of circulatingTNF in humans infected by P. brasiliensis [21]; however, thestimuli for TNF production and the role of TNF in theinfectious process are unknown. In the present study we madesome striking observations concerning the relationship betweenP. brasiliensis infection in vivo and in vitro and TNF secretion.First, TNF levels increased when the mice were challenged withthe cell wall fraction (Fl) from both fungal strains. Second, theserum TNF levels were significantly higher in mice challengedwith FlPb265 than in mice challenged with FlPbl8, and thepatterns were similar to the TNF levels of mice challenged withthe yeast-like forms of both fungal strains (Fig. la). Third,fraction Fl of the two strains consists of chitin and fl-glucan,with the Fl of the avirulent strain containing much more fi-

glucan than the Fl from the virulent strain. Finally, fl-glucanpurified from either strain increased serum TNF levels in an

identical dose-dependent manner.

Cell wall components have been associated with the dimor-phism and virulence of P. brasiliensis [5,28]. oz-glucan has beenstrongly implicated in the virulence of the fungus [29]. Thispolysaccharide exists only in the parasitic yeast form of thefungus, but not in the mycelial form. The reduced cell wall a-glucan in mutants or laboratory-passaged strains is directlycorrelated with a decrease in virulence. On the other hand,mutants with a greater amount of a-glucan have increasedvirulence. The avirulent strain Pb265, used here, contains 2-6-fold less a-glucan than the virulent Pbl8 strain (unpublisheddata). We recently demonstrated that the Fl cell wall fractionplays an important role in the inflammatory response evoked bythe fungi, inducing leucocyte migration when injected into theperitoneal cavity [22]. When injected into the subcutaneouslayer of mice, the F fraction also induces macrophage accumu-

lation, differentiation of epithelioid cells, and the persistence ofwell formed granulomas [7]. Since the F1 fraction consists of amixture of chitin and fl-glucan, and chitin does not induceleucocyte migration or granuloma formation, it is likely that f,-glucan is the main active component of the Fl fraction.

The activation of macrophages is correlated with metaboli-zation of the Fl fraction, whose active components are

continuously and slowly degraded. In a study of the cell wallarchitecture, the position of a-glucan and of the chitin and fl-glucan complex, the strategic arrangement of a-glucan, which islocated more externally on the cell wall, protects fl-glucan frommetabolism by the host [30]. Because it is retained in the cellwall, 15-glucan could regulate TNF production by macrophages.

JJ-glucan, purified from both fungal strains, was equallycapable of inducing TNF production in vivo. In contrast, the F Ifraction from the avirulent strain induced higher TNF produc-tion than the Fl fraction from the virulent strain. A similarcorrelation was observed for yeast-like forms of both fungal

strains. These results are explained by our finding that TNFproduction was directly correlated with fl-glucan concentration.This role of fl-glucan is further supported by studies on TNFproduction in vitro.

TNF may somehow serve as a host defence mechanismagainst a wide range of infectious pathogens, and TNF levelsappeared to be correlated with the severity of disease caused bythese organisms. However, comprehensive understanding of therole of TNF in the pathogenesis of these infectious diseasesrequires definition of the biochemical components of microbialpathogens which induce TNF release. Ofinterest in this regard isthe interaction between mycobacteria and host cells. Lypoarabi-nomannan (LAM), a major component of Mycobacteriumtuberculosis cell-wall, is considered a virulence/pathogenicityfactor [311, and LAM isolated from a virulent strain and from anavirulent strain, which have been shown to differ markedly interms of the structural basis, elicited TNF release by macro-phages in a dose-dependent manner, although LAM from theavirulent strain was 100-fold more potent at inducing TNFsecretion than LAM from the virulent strain [32].

Regarding the possible role of TNF, an interesting observa-tion in P. brasiliensis infection is the inflammatory response.Virulent strains, such as Pbl8, induce a large inflammatoryinflux, and the lesion is characterized by a progressive andgranulomatous reaction associated with numerous organisms.In contrast, the avirulent strain Pb265 induces a sparse inflam-matory reaction and the lesion is characterized by a non-progressive macrophage accumulation associated with very feworganisms, as suggested by the in vivo experiments. Thegranulomatous inflammatory reaction represents the mostspecialized and efficient tissue response of an organism against aparasite [9,14,15]. The full development of granulomas requiresa T lymphocyte immune response, and involves the continuedattraction ofmonocytes and their differentiation into epithelioidcells [10,1 1]. TNF is required for macrophage accumulation anddifferentiation into epithelioid cells, and for the persistence ofwell formed granulomas [14,15]. It has recently been demon-strated that F1 induces leucocyte migration, attraction, ac-cumulation and differentiation ofmonocytes to epithelioid cells.Those effects were abrogated with enzymatic cleavage of sugarscontaining the fl-1,3 linkage yield chitin as a final product(unpublished data). These results suggest that fl-glucan is themain active component for these effects. Since fl-glucan isassociated with the development of granulomas, and antibodyanti-TNF treatment prevented the formation and persistence ofmature granulomas (data not shown), we assume that TNFreleased by macrophages is involved in a process of granulomadevelopment.

TNF has also been shown to be necessary for T cell-independent macrophage activation and interferon-gamma(IFN-y) production [33]. IFN can activate macrophages tosecrete TNF and to inhibit the replication of P. brasiliensis [27].These observations suggest that TNF released by macrophagesenhances and perpetuates its own synthesis and secretion. Sucheffects might favour further macrophage accumulation, epithe-lioid differentiation and the persistence of well formed granulo-mas.

The findings presented here provide evidence that TNF isproduced in response to P. brasiliensis, and that production bymacrophages is directly induced and regulated by several cellwall constituents of the fungus. Though TNF may play a role in

193

194 F. Figueiredo, L. M. C. Alves & C. L. Silva

the pathogenicity of the inflammatory response induced by thefungus, the full sequence ofevents ultimately responsible for thegranuloma formation remains to be elucidated.

ACKNOWLEDGMENTS

This work was supported by grants from Conselho Nacional deDesenvolvimento Cientifico e Tecnol6gico (CNPq) and Fundacao deAmparo a Pesquisa do Estado de Sdo Paulo (FAPESP), Brazil. Wethank S. L. Brandao Filho and I. Tincani Brandao for excellenttechnical support, and Regina M. C. F. Bueno for her expert secretarialassistance.

REFERENCES

1 Restrepo A, Greer DL, Vasconcelos M. Paracoccidioidomycosis, areview. Review Med Vet 1973; 8:97-101.

2 Severo LC, Greyer GR, Londero AT et al. The primary pulmonarylymph node complex in paracoccidioidomycosis. Mycopathologia1979; 67:115-18.

3 Musatti CC, Rezkallah RT, Mendes E et al. In vivo and in vitroevaluation of cell-mediated immunity in patients with paracoccidi-oidomycosis. Cell Immunol 1976; 24:365-78.

4 Brummer E, Hanson LH, Restrepo A, Stevens DA. In vivo and invitro activation of pulmonary macrophages by IFN-y for enhancedkilling of Paracoccidioides brasiliensis or Blastomyces dermatitidis. J.Immunol 1988; 140:2786-9.

5 San-Blas G, San-Blas F. Paracoccidioides brasiliensis. Cell wallstructure and virulence. A review. Mycopathologia 1977; 62'77-86.

6 Silva CL, Fazioli RA. A Paracoccidioides brasiliensis polysaccharidehaving granuloma-inducing toxic and macrophage-stimulating acti-vity. J Gen Microbiol 1985; 131:1497-501.

7 Figueiredo F, Silva CL, Alves LMC et al. Participation of Paracocci-dioides brasiliensis lipids and polysaccharides in the evolution ofgranulomas. Brazilian J Med Biol Res 1986; 19:651.

8 Ando M, DannenbergAM Jr, Shima K. Macrophage accumulation,division, maturation and digestive and microbicidal capacities intuberculosis lesions. II. Rate at which mononuclear cells enter anddivide in primary BCG lesions and those of infections. J Immunol1972; 109:8-19.

9 Adams DO. The biology of the granuloma. In: Ioachim HL, ed.Pathology of granulomas, 1st edn. New York: Raven Press, 1983:1-20.

10 Adams DO. The structure of mononuclear phagocytes differentiat-ing in vivo. I. Sequential time and histologic studies of the effect ofBacillus Calmette-Guerin (BCG). Am J Pathol 1974; 76:17-48.

11 Adams DO. The granulomatous inflammatory response. Am JPathol 1976; 84:164-91.

12 Beutler B, Cerami A. Cachectin: more than a tumor necrosis factor.New Engl J Med 1987; 316:379-85.

13 Lane FC, Unanue ER. Requirement of thymus (T) lymphocytes forresistance to Listeriosis. J Exp Med 1972; 135:1104-12.

14 Kindler V, Sappino AP, Grau GE, Piquet PF, Vassalli P. Theinducing role of tumor necrosis factor in the development of

bactericidal granulomas during BCG infection. Cell 1989; 56:731-40.

15 Shikama Y, Koabyashi K, Kasahara K et al. Granuloma formationby artificial macroparticles in vitro: macrophages and monokinesplay a critical role in granuloma formation. Am J Pathol 1989;134:1189-99.

16 Waage A, Halstensen A, Espevik T. Association between tumornecrosis factor in serum and fatal outcome in patients withmeningococcal disease. Lancet 1987; 1:355-7.

17 Silva CL, Faccioli LH, Rocha GM. The role ofcachectin/TNF in thepathogenesis of tuberculosis. Brazilian J Med Biol Res 1988; 21:489-92.

18 Scuderi P. Lam KS, Ryan KJ et al. Raised serum level of tumornecrosis factor in parasitic infections. Lancet 1986; 2:1364-5.

19 Silva CL, Foss NT. Tumor necrosis factor in leprosy patients. JInfect Dis 1989; 159:787-90.

20 Smith JG, Magre DM, Williams DM et al. Tumor necrosis factorplays a role in host defense against Histoplasma capsulatum. J InfecDis 1990; 162:1349-53.

21 Silva CL, Figueiredo F. Tumor necrosis factor in paracoccidioido-mycosis patients. J Infect Dis 1991; 164:1033-4.

22 Carareto Alves LM, Figueiredo F, Brandao Filho SL, Tincani I,Silva CL. The role of fractions from Paracoccidioides brasiliensis inthe genesis ofinflammatory response. Mycopathologia 1987; 97:3-7.

23 Fava Netto C. Contribui4ao para o estudo imunol6gico de Blasto-micose de Lutz (Blastomicose Sul-Americana). Rev Inst AdolfoLutz 1961; 21:99-122.

24 Kanetsuna F, Carbonell LM. Cell wall glucans of the yeast andmycelial forms of Paracoccidioides brasiliensis. J Bacteriol 1970;101:675-80.

25 Ruff MR, Gifford GE. Purification and physiochemical characteri-zation of rabbit tumor necrosis factor. J Immunol 1980; 125:1671-7.

26 Remick DG, Scales WE, May MA et al. In situ hybridizationanalysis of macrophage-derived tumor necrosis factor and inter-leukin- I mRNA. Lab Invest 1988; 59:809-16.

27 Brummer E, Hanson LH, Restrepo A, Stevens DA. In vivo and invitro activation of pulmonary macrophages by IFN-y for enhancedkilling of Paracoccidioides brasiliensis or Blastomyces dermatitidis. JImmunol 1988; 140:2786-9.

28 Hallak J, San-Blas F, San-Blas G. Isolation and wall analysis ofdimorphic mutants of Paracoccidioides brasiliensis. Sabouraudia1982; 20:51-62.

29 San-Blas G, San-Blas F, Serrano LE. Host-parasite relationships inthe yeast like form of Paracoccidioides brasiliensis strain IVIC Pb9.Infect Immun 1977; 15:343-6.

30 San-Blas G, San-Blas F. Molecular aspects of fungal dimorphism.Crit Rev Microbiol 1985; 11: 101-27.

31 Chatterjee D, Hunter SW, McNeil M, Brennan PJ. Lipoarabino-mannan; multiglycosylated form of the mycobacterial mannosyl-phosphatidylinositols. J Biol Chem 1992; 267:6228-33.

32 Chatterjee D, Roberts AD, Lowell K, Brennan PJ, Orme IM.Structural basis of capacity of lipoarabinomannan to inducesecretion of tumor necrosis factor. Infect Immun 1992; 60: 1249-53.

33 Bancroft GJ, Sheehan KCF, Schreiber RD, Umanue ER. Tumornecrosis factor is involved in the T cell independent pathway ofmacrophage activation in scid mice. J Immunol 1989; 143:127-30.