INFECTION AND IMMUNITY, Aug. 1986, p. 411-419 Vol. 53, No. 20019-9567/86/080411-09$02.00/0Copyright © 1986, American Society for Microbiology

Characterization of a Secretory Proteinase of Candida parapsilosisand Evidence for the Absence of the Enzyme during Infection

In VitroR. RUCHEL,* BIRGITT BONING, AND MARGARETE BORG

Department of Medical Microbiology, Institute ofHygiene, University of Gottingen, D-3400 Gottingen,Federal Republic of Germany

Received 13 February 1986/Accepted 8 May 1986

The opportunistic yeastlike fungi of the genus Candida comprise three species which are proteolytic in vitro.Among them, C. albicans and C. tropicalis are of foremost medical importance. However, a strict correlationbetween extracellular proteolytic activity and virulence is opposed by the low virulence of the third proteolyticspecies, C. parapsilosis. We purified the secretory acid proteinase of C. parapsilosis (clinical isolate 265). Theenzyme is a carboxyl proteinase (EC 3.4.23) like all other secretory Candida proteinases handled so far.Proteinase 265 is distinguished by a lower molecular weight (approximately 33,000); it has increasedhydrophobicity, which accounts for inhibition of the enzyme by hemin, and required the presence of nonionicdetergent in the initial steps of purification. The enzyme already undergoes alkaline denaturation at neutrality.Its activity is thus confined to the acid microenvironment of the fungal cel wall. Within this range, the enzymemay degrade immunoglobulins like immunoglobulin Al (IgAl), IgA2, and secretory IgA. No indication wasfound for glycosylation of proteinase 265 and the related enzyme of C. albicans CBS 2730. However, thecomparable proteinase of C. tropicalis 293 was identified as a manno protein. Antiserum against proteinase 265cross-reacted strongly with corresponding enzymes from other Candida species. Antisera against proteinases ofC. albicans and C. tropicalis reacted only weakly with proteinase 265. Thus, secretory Candida proteinases arelikely to possess common and species-specific antigenic sites. In contrast to C. albicans, infection of phagocytesby C. parapsilosis 265 was not accompanied by secretion of fungal proteinase. This lack of induction of theenzyme under conditions of infection may account for the low virulence of most isolates of C. parapsiosis.

Among the yeastlike fungi of the genus Candida, C.albicans is of foremost medical importance as an opportu-nistic organism, followed by C. tropicalis and possibly C.parapsilosis (1, 3). These three species are distinguished bythe secretion of acid proteinase in vitro. Candida proteinaseswere first discovered in C. albicans by Staib (37). Subse-quently, extracellular proteolytic activity was also detectedamong the majority of isolates of C. tropicalis and C.parapsilosis (17, 36).

Considerable evidence subsequently accrued indicatingthat secretory proteinases are factors in the virulence of C.albicans (15, 17, 19, 20a, 32, 38). Such correlation may alsoexist among isolates of C. tropicalis (4), but it is missing in C.parapsilosis. Isolates of this species are mostly proteolytic invitro (17, 36, 39), but they possess only low virulence in vitro(4) and in mice (2). In humans, deep mycoses due to C.parapsilosis had a better prognosis than those that werecaused by other Candida species (6).To elucidate the discrepancy between the high proteolytic

activity in vitro and the low virulence of C. parapsilosis, wepurified and characterized a secretory proteinase of thisspecies, thus allowing comparison with enzymes of the morevirulent species and also permitting the production of spe-cific antibodies, which were used to trace the enzyme underconditions of infection.

(These data were presented in part at the 9th Congress ofthe ISHAM, Atlanta, Ga., 19-24 May 1985.)

* Corresponding author.

MATERIALS AND METHODS

C. parapsilosis 265/80 and 21/8 and C. tropicalis 293/80are clinical isolates handled in this institute; C. parapsilosis78K was kindly provided by 0. Zimmermann, Gottingen,Federal Republic of Germany. C. albicans CBS 2730+ is avariant of the type strain; it produces little mycelia in vitro.The secretory proteinases of the type strain and the varianthad identical properties (unpublished data). All yeasts wereclassified by the API 20C Auxanogram (API-Bio Mdrieux,Nurtingen, Federal Republic of Germany).

All second-antibody peroxidase and fluorescein conju-gates were from Dakopatts, Hamburg, Federal Republic ofGermany. Yeast carbon base was from Difco Laboratories,Detroit, Mich.Bovine serum albumin (BSA), butanedione, dithiothreitol,

o-phenylenediamine, and all common laboratory chemicalswere from E. Merck AG, Darmstadt, Federal Republic ofGermany. Endoglycosidase F was purchased from NewEngland Nuclear Corp., Boston, Mass. Microtest plates andvials for tissue culture were from Nunc, Roskilde, Denmark.Isoelectric focusing Sephadex, Sephacryl S200, andaminohexyl Sepharose 4B were from Pharmacia, Inc.,Piscataway, N.J. Nitrocellulose membranes (BA85) wereobtained from Schleicher & Schuell, Inc., Keene, N.H.

Molecular weight standards, carrier ampholytes, reagentsfor sodium dodecyl sulfate-polyacrylamide gel electrophore-sis (SDS-PAGE), DEAE-cellulose, ethylene glycol-bis(p-aminoethyl ether)-N,N,N',N'-tetraacetic acid (EGTA), o-

411

on October 1, 2018 by guest

http://iai.asm.org/

Dow

nloaded from

412 RUCHEL ET AL.

phenanthroline, alcian blue, basic fuchsin, and fast greenFCF were from Serva, Heidelberg, Federal Republic ofGermany.Crude bovine hemoglobin, concanavalin A (ConA), ConA

agarose, diazoacetylnorleucine methyl ester (DAN), hemin,iodoacetamide, methyl green, p-phenylenediamine, secre-tory immunoglobulin A (sIgA) from human colostrum, wheatgerm agglutinin (WGA), and antisera against ConA andWGA were from Sigma Chemical Co., St. Louis, Mo.Human myeloma proteins of the IgAl and A2 isotypes

were kindly provided by H. Kratzin (Gottingen, FederalRepublic of Germany) and F. Skvaril (Berne, Switzerland).

Cultivation. C. parapsilosis 265/80 was grown at 30°C in a9-liter fermentor at 200 rpm and with 5 liters of air per min.The growth medium contained 1% bovine hemoglobin, 1.2%yeast carbon base, and 1% glucose (pH 6.2). The mediumwas sterilized by filtration and inoculated with 50 ml of anovernight culture of the yeast in glucose-peptone broth.After 2 days, the pH had dropped below 5 and the acidproteolytic activity had reached a plateau at approximately22 U/ml. The content of the fermentor was adjusted to pH6.5, and the cells were removed by centrifugation (4,000 x g,60 min).

Purification of acid proteinase. The culture supernatantwas concentrated by polyethylene glycol dialysis and chro-matographed on DEAE-cellulose as described previously(29). Addition of 0.05% nonionic detergent (EmulphogeneBC-720) to all buffers of the chromatographic system wasnecessary to avoid nonspecific adsorption of the enzyme.Fractions with enzymic activity were located by hemo-globin assay (see below); they were pooled and stored at- 180C.The crude enzyme was purified further by (i) gel filtration

through Sephacryl S-200 at pH 6.2 in 0.2 M citrate buffer, (ii)preparative isoelectric focusing in granulated gel in thepresence of nonionic detergent, and (iii) affinity chromatog-raphy on pepstatin linked to aminohexyl Sepharose-4B(29).Enzyme characterization. Acid proteolytic activity was

quantitated by the Anson test (16) with 1% hemoglobin as asubstrate at pH 3. One unit of enzymic activity was definedarbitrarily as the amount of enzyme that released trichloro-acetic acid-soluble fragments of 0.1 absorption units (280nm) from hemoglobin in 1 h at 37°C.The pH-dependent activity profile of proteinase 265 was

determined in 0.1 M sodium citrate buffers of pH 2 to 6 withhemoglobin as the substrate. Separate blanks were run foreach pH step.The stability of proteinase 265 under alkaline conditions

was tested by exposure of the enzyme to 0.2 M phosphatebuffer (pH 6 to 8) for 30 min at 220C. Residual enzymicactivity was determined at pH 3.5 by hemoglobin assay.

Inhibition of enzyme activity. EGTA and o-phenanthrolinewere used at a final concentration of 1 mM. Pepstatin A waskept as a stock solution of 1 mM in methanol at - 18°C; itwas applied at a final concentration of 10-5 M. Dithiothreitoland iodoacetamide were used at 20 mM. DAN was used at afinal concentration of 1 mM in the presence of copper ions(14), and butanedione was used as described by Gripon andHofmann (10). Hemin was used from a freshly preparedstock solution (12 mM) in 1% aqueous ammonium hydrox-ide. The final concentration of hemin was 1.8 mM versus10-7 M proteinase. The pH was adjusted by addition of 0.1M citrate buffer (pHs 2.8 to 6.5). The pH was controlled witha microprobe as described above. Endoglycosidase F wasapplied to Candida proteinase as described by Elder and

Alexander (7); the effect of the treatment was monitored bycomparative SDS-PAGE.

Electrophoresis. PAGE was performed essentially as pre-viously described (35). The gels were cast in batches of 10using equipment from Hoefer, San Francisco, Calif. Gelswere either stained with Coomassie blue R or processedfurther for immunoblotting. Electrophoretically separatedCandida proteinases were stained for bound phosphate bythe methyl green method (5).

Isoelectric focusing. Preparative electrofocusing in IEFSephadex was performed essentially as described by Radola(26) with 1.6% carrier ampholytes of the pH 2 to 11 range.Emulphogene BC-720 (0.05%) was added to the gel slurry; a10-cm-long gel bed was used for preparative purposes, whilea 24-cm bed length was used for analytical runs. Methylgreen and fast green FCF were used as cationic and anionicmarkers, respectively. The runs were performed at constantpower and cooling to approximately 12°C. After terminationof the run, the pHs of the gel fractions were determined witha microelectrode (type MI-410; Microelectrodes Inc.,Londonderry, N.H.). The liquid in the gel fractions wasrecovered by brief centrifugation at 800 x g of the gel slurrythrough cheesecloth (10-,um mesh). After titration to pH 6.2,fractions were dialyzed and concentrated by nitrogen pres-sure dialysis if necessary.Western blotting and lectin assays. Electrophoretic transfer

of proteinase from polyacrylamide gel slabs to nitrocellulosemembranes was performed in 10 mM sodium tetraborate, pH9.2, as described by Renart and Sandoval (27). After thetransfer, the membrane was saturated with 2% BSA inphosphate-buffered saline (PBS) for 2 h under gentle agita-tion. Proteinase fractions that had been transferred ontonitrocellulose membrane were reacted with the lectins ConAand WGA essentially as described previously (9). The blotmembrane was exposed to the lectins (5 ,ug/ml) for 2 h in thepresence of 2% BSA in PBS-1 mM MgCl2. Next, themembrane was rinsed five times for 5 min each inPBS-0.05% Tween 20; subsequently it was exposed tolectin-specific antibodies (approximately S ,ug/ml in PBS-2%BSA). After another 2 h, the membrane was rinsed asdescribed above and exposed for 2 h to the second-antibodyperoxidase conjugate (diluted 2,000-fold in PBS-2%BSA-0.05% Tween 20). After final rinsing as describedabove, the membrane was exposed to the chromogenicdiaminobenzidine-H202 substrate (12).

Controls of the specificity of the lectin reactions wereperformed in parallel. Droplets of ovalbumin and BSAsolutions (1 mg/ml in PBS) were placed on small chips ofnitrocellulose membrane. After 15 min in a moist chamber,the chips were rinsed five times in PBS prior to immersion inPBS-2% BSA for 2 h. The subsequent steps of the reactionwere identical with those described above. In the course ofthe lectin reaction, only BSA treated with periodate andglycerol was used to inactivate carbohydrate contaminants(9).

Following the lectin peroxidase reaction, the blot mem-brane was once more rinsed in PBS. The membrane wasthen exposed for 2 h to antibodies against secretory Candidaproteinases (approximately S jg/ml) in PBS-2% BSA. Afteranother rinsing in PBS, the membrane was immersed in asolution of the suitable second-antibody peroxidase conju-gate (in PBS-2% BSA). Finally, the membrane was rinsed asusual and exposed again to the chromogenic peroxidasesubstrate. For the sake of immunological compatibility, thesequentially used antibodies were from the following spe-cies: anti-ConA and anti-WGA from rabbit, anti-rabbit per-

INFECT. IMMUN.

on October 1, 2018 by guest

http://iai.asm.org/

Dow

nloaded from

SECRETORY PROTEINASE OF C. PARAPSILOSIS 413

oxidase from hog, anti-secretory Candida proteinase fromguinea pig, and anti-guinea pig peroxidase from rabbit. Allincubation steps were performed at room temperature andunder constant, gentle agitation.Immunologic methods. Antisera against proteinase 265, the

proteinase having been purified by ion-exchange chromatog-raphy and gel filtration, were raised in guinea pigs byrepeated intramuscular injections of active enzyme inFreund adjuvant at pH 6.5 6 and 2 weeks prior to bleeding.Enzyme-linked immunosorbent assays for the detection ofCandida proteinase or specific antibodies were performed aspreviously described (33).For immunofluorescence, fixed cell cultures were submit-

ted to absolute methanol (15 min, -22°C), which renders theplasma membrane penetrable for immunoglobulins (21).Prior to reaction with the first antibody, the sample wasexposed to BSA (1% [wt/vol] in PBS, 30 min, 22°C). Subse-quently, the. cells were reacted with antiserum againstCandida proteinase (diluted 100-fold in PBS-1% BSA, 12 h,8°C). This serum had been elicited in guinea pig against thesecretory proteinase of C. parapsilosis 265; it showed strongcross-reactions in enzyme-linked immunosorbent assaysagainst the related proteinases of C. albicans CBS 2730 andC. tropicalis 293. The controls were treated with normalguinea pig serum under identical conditions. Afterwards, thecells were rinsed carefully with PBS (four times for 5 mineach). Prior to the second antibody, the cells were exposedto normal rabbit serum (diluted 30-fold in PBS, 30 min,22°C). The second antibody was a fluorescein-conjugatedanti-guinea pig immunoglobulin from rabbit (DAKO; diluted30-fold in PBS-1% BSA). After 90 min at 37°C, the cellswere rinsed again in PBS. For examination by fluorescencemicroscopy the preparations were mounted in PBS-glycerol(1:9) containing 0.1% (wt/vol) p-phenylenediamine; the latterwas added to reduce fluorescence fading (13).

Immunoglobulins as substrates. Purified human myelomaproteins of the IgAl, IgA2, and IgGl (heavy chain) types (2mg/ml) were exposed to proteinase (5 pug/ml) in 0.1 M citratebuffer, pH 3.5, at 37°C for up to 12 h. Samples of the digestswere taken at intervals, and pepstatin was added to a finalconcentration of 10-6 M. The samples were dialyzed against20 mM citrate buffer, pH 6.5, prior to analytic PAGE.For subsequent degradation of secretory IgA, C. parapsi-

losis 265 was grown in yeast carbon base-hemoglobin me-dium until the pH had dropped below 4. At this stage thesupernatant was clear and did not contain recognizableamounts of high-molecular-weight polypeptides. Two milli-liters of the supernatant, containing highly proteolytic yeastcells, was transferred into a 10-ml Erlenmeyer flask andsupplemented with 0.1 ml of 10% glucose in yeast carbonbase solution. Secretory IgA was added to a final concentra-tion of 1 mg/ml. The flask was placed on a rocker platformand kept at 37°C. Samples (0.1 ml) were removed everyother day. The volume was replenished by glucose-yeastcarbon base solution as described above. In the test intervalof 10 days, the pH stayed below 4. The controls were kept atpH 3.7 under comparable conditions. Samples that had beentaken were treated with pepstatin as described above anddialyzed prior to PAGE.

Infection of phagocytes in vitro. Peritoneal macrophageswere collected from male NWNI mice (approximately 35 g)3 days after stimulation with 3 ml of thioglycolate broth. Thecells were rinsed in Hanks balanced salt solution and trans-ferred to tissue culture medium 1940-199-10% fetal calfserum (40). One milliliter of the suspension with 3 x 105 cellswas placed on a round cover slip (12-mm diameter) on the

a b +

FIG. 1. Coomassie blue staining of SDS-PAGE of (a) pooledproteolytic fractions after DEAE chromatography and (b) the frac-tion of highest enzymic activity on DEAE chromatography afterrechromatography on Sephacryl S200. The arrow indicates migra-tion of proteins toward the anode.

bottom of a well of a tissue culture plate (Nunc; no. 134673).For adherence, the cells were incubated for 24 h (37°C, 5%C02). The adherent macrophages were infected with 106Candida blastospores from an overnight culture in glucosebroth. The yeast cells had been pretreated with normalmouse serum (10% in cell culture medium as describedabove; 30 min, 37°C) to facilitate phagocytosis, and theywere suspended in 1 ml of regular culture medium prior toinfection. After infection, the tissue culture plates werecentrifuged (200 x g, 2 min) to enhance contact betweenmacrophages and yeasts. Subsequently the plates wereincubated for 20 h as described above. Finally the cultureswere rinsed with PBS and fixed with freshly preparedparaformaldehyde (1% [wt/vol]) in PBS (1 h, 22°C). Thecover slips with fixed cells were directly subjected to immu-nohistochemical analysis.

RESULTS

Purification of proteinase 265. After 48 h of growth inproteinaceous medium, culture supernatants typically con-tained 22 U of enzymic activity per ml. Subsequent DEAEchromatography of the supernatant at pH 6.5 to 6.2, whichhad been applied successfully to related proteinases of C.albicans and C. tropicalis (36), caused 90% loss of enzymicactivity. Since the pH-dependent activity profile of alkalinedenaturation did not reveal a particular instability of theenzyme (see below), attention was focused on hydrophobicinteractions.

Proteinase 265 that had been adsorbed to the column in theabsence of detergent could not be mobilized again even byaddition of 1% (vollvol) nonionic detergent (EmulphogeneBC-720). However, when DEAE chromatography was per-formed in the presence of 0.05% Emulphogene, more than70% of the enzyme was recovered. Partially purified enzymefractions were also obtained from the culture supernatants oftwo additional strains of C. parapsilosis, 21/78 and 78K.PAGE of partially purified proteinase 265 revealed the

persistence of contaminating protein species. By gel filtra-tion through Sephacryl S200, relative separation of theproteinase from the contaminants was accomplished (Fig. 1).The pooled proteolytic fractions derived from gel filtrationwere used for further experimentation unless stated other-wise. Whereas ion-exchange chromatography raised thespecific activity approximately 100-fold, gel filtration raisedthe specific activity another 5-fold. Yield in this purificationstep was 80%. Owing to incomplete separation of contami-

VOL. 53, 1986

on October 1, 2018 by guest

http://iai.asm.org/

Dow

nloaded from

414 RUCHEL ET AL.

F 6 7 8 9

a 40o-c

10o-pH 5.0 5.2 5.6 6.2 +

b

5-4-

FIG. 2. SDS-PAGE with Coomassie blue staining of fractions (F)obtained from preparative electrofocusing of partially purified pro-teinase 265 after DEAE chromatography: (a) after 3 h at 50 V/cm; (b)after 4 h at 50 V/cm; (c) fractions 7 and 9 from 4 h of focusingcombined. The pHs apply to the corresponding fractions of bothruns. The arrow indicates migration of proteins toward the anode.

nants, however, only half of the enzyme was available in thepure state shown in Fig. 1.

Preparative isoelectric focusing. Partially purified protein-ase after DEAE chromatography was electrofocused inSephadex gel with carrier ampholytes of the pH 2 to 11range. Focusing was performed at constant power andstopped after 3 h, when the voltage had barely reached itsplateau (50 V/cm).At this stage, the enzymic activity in the gel was focused

in a symmetric peak (approximately pH 5.2), which corre-sponded to a single protein of high purity as shown bySDS-PAGE (Fig. 2). When electrofocusing was continuedfor another hour, a gradual shift of the enzyme back toneutrality was observed (Fig. 2b). This shift was accompa-nied by loss of enzymic activity. Both the active enzyme andinactive protein (Fig. 2b, fractions 8 and 9) were reactiveupon immunoblotting with antibodies against proteinase 265(data not shown).No difference in mobility between the active enzyme and

the inactive variant was observed in SDS-PAGE (Fig. 2c).Thus, inactivation of proteinase 265 in the course ofelectrofocusing is not due to proteolytic autolysis but ap-pears to involve a more subtle structural conversion, themechanism of which has not yet been elucidated. The lowyield of active enzyme prohibited the further use of prepar-ative electrofocusing.

Estimation of molecular mass. Proteinase 265 was submit-ted to gel filtration through Sephacryl S200 in the presence of0.05% Emulphogene. Its position relative to different stan-dard proteins indicated a molecular mass of approximately33,000 daltons. This estimate was confirmed by SDS-PAGE(Fig. 3). SDS-PAGE in the presence or absence of reducingagent yielded a single peak in a virtually identical position;proteinase 265 thus appears to be a single-stranded polypep-tide without intramolecular disulfide bonds.

Isoelectric point. For estimation of the isoelectric point ofproteinase 265, isoelectric focusing in granulated gel wasperformed as described above. The use of a long gel bed (24cm) enhanced resolution. Soon after the voltage plateau at 50V/cm had been reached, enzyme was recovered from the gelfractions. The profile of proteolytic activity revealed awell-defined peak at pH 5.3, which was accompanied by asecondary peak at pH 6.5 (Fig. 4). When the application site

2-

D

a /

0

/

* T

b

B0

0+

MFIG. 3. Estimation of the molecular mass of proteinase 265 by (a)

gel filtration in Sephacryl S200 in 20 mM citrate buffer (pH6.5)-0.05% Emulphogene BC-720 and (b) SDS-PAGE in a gradientgel of 2.5 to 20% total acrylamide and 2.5% cross-linkage. M, Linearplot of migration velocity (gel filtration) and distance (PAGE). D,Log of molecular mass (104 daltons). Standard proteins: horseferritin (F; 440,000 daltons), BSA (B; 68,000 daltons), ovalbumin (0;45,000 daltons), and soy bean trypsin inhibitor (T; 21,500 daltons).The proteinase (0) was detected by hemoglobin assay (gel filtration)or Coomassie blue staining (PAGE).

of the sample was shifted from a position corresponding to afinal pH of 7 into the position corresponding to a final pH of5, essentially the same pattern of activity was obtained. Withprolonged focusing time, the enzymic activity shifted to theposition at pH 6.5. The shift was accompanied by the grossloss of activity that had been observed previously on pre-parative electrofocusing (see above). The results suggest anisoelectric point of active proteinase 265 at approximatelypH 5.3.

Carbohydrate moiety. Some secretory acid Candida pro-teinases are manno proteins (18). However, staining ofvarious Candida proteinases in polyacrylamide gel by us

pH

10.0

90

8.0

7.0

60

5.0

4.0

3.0

+FIG. 4. Estimation of the isoelectric point of proteinase 265 by

electrofocusing in a bed of granulated gel with carrier ampholytes ofpHs 2 to 11 and a nonionic detergent (Emulphogene BC-720, 0.05%).After termination of the run, pH (@) was determined in each fraction(Fr) of the gel. Subsequently, proteolytic activity at pH 3.5 (x) wasdetermined by hemoglobin assay (A280). The sample was applied inthe position of fraction 22; the position of maximum activitycorresponds to pH 5.3.

INFECT. IMMUN.

on October 1, 2018 by guest

http://iai.asm.org/

Dow

nloaded from

SECRETORY PROTEINASE OF C. PARAPSILOSIS 415

with periodic Schiff reagent or alcian blue resulted in incon-sistent data (unpublished data). Thus, we performed im-munoblots with anti-mannan antibody that had been elicitedin rabbits against mannan of C. albicans CBS 2730 (preparedas described in reference 22). Such antibodies in enzyme-linked immunosorbent assays cross-reacted strongly withmannans of C. tropicalis and C. parapsilosis. The im-munoblot revealed that only the proteinase of C. tropicalis293 was a manno protein (Fig. 5a and b); the weak reactionsof the enzymes from C. albicans and C. parapsilosis wereconsidered nonspecific.

All three enzymes were tested for terminal glucose ormannose by reaction with ConA. The reaction was per-formed on nitrocellulose blots after SDS-PAGE. BSA sam-ples that had been treated with periodate (9) or ovalbuminwere used as negative and positive controls, respectively.There were no reactions of the three proteinases with ConA,but certain glycosylated contaminants of the preparationsreacted (Fig. 5c). Binding ofWGA was also attempted under

a

I4

b

2 3

c

1 2 3

d

1 2 3 1 2 3

FIG. 5. Detection of carbohydrate moieties in proteinases fromC. albicans CBS 2730 (lane 1), C. tropicalis 293 (lane 2), and C.parapsilosis 265 (lane 3). Panel a is the reference SDS-PAGE stainedwith Coomassie blue; panel b is a nitrocellulose blot of a corre-sponding gel section after immunoperoxidase reaction for demon-stration of mannan; panel c is another blot of a correspondingsection of the slab gel after immunoperoxidase reaction of boundConA; panel d is the same as panel c, but the membrane was treatedwith an anti-proteinase antibody as well. P, Position of theproteinases. The arrow indicates migration of proteins toward theanode.

A -100% 0

0

-50

-10

6 7 8 9 pH

FIG. 6. Alkaline denaturation of proteinase 265 (0) when ex-posed to various pHs for 30 min at room temperature. Residualenzymic activity was monitored by hemoglobin assay at pH 3.5 (A,%). The denaturation profile of the comparatively stable proteinaseof C. albicans CBS 2730 (0) has been entered for comparison.

comparable conditions. This lectin reacts with N-acetyl-glucosamine, which is the constituent of chitin (23). Noreaction of WGA with the three proteinase preparations wasobserved (data not shown).

In a subsequent immunoreaction the same blots wereexposed to antibodies against Candida proteinase. Theseantibodies had been elicited in guinea pigs against theproteinase of C. parapsilosis; they cross-reacted stronglywith corresponding enzymes of the other two proteolyticCandida species. The spots of the three proteinases becameapparent on the blot (Fig. Sd) only after application of theanti-proteinase antibody, indicating that the three enzymeshave no terminal glucose or mannose (ConA) nor N-acetylglucosamine (WGA).

Further characterization of proteinase 265. When testedwith hemoglobin as a substrate, proteinase 265 had a primaryproteolytic activity maximum at pH 4.3 and a secondarymaximum at pH 3. However, the profile covered a broadrange of pHs, since 25% of activity was detected at pHs 2.2and 5.8, respectively (data not shown).The resistance of proteinase 265 to weakly alkaline con-

ditions was tested at room temperature by exposure for 30min to phosphate buffer at pHs 6 to 8. An assay of residualacid proteolytic activity revealed gradual denaturation of theenzyme above pH 7 (Fig. 6). The denaturation profileresembled profiles that had been obtained from various labileproteinases of C. albicans and C. tropicalis. More stableenzymes have only been detected among isolates of C.albicans (29, 31).

Proteinase 265 was fully inhibited by pepstatin, allowingfor its classification as an acid proteinase (EC 3.4.23). Noinhibition was observed with DAN at pH 5.

Butanedione caused 25% inhibition at pH 6. Under iden-tical conditions porcine pepsin was fully inhibited bybutanedione, whereas DAN caused 55% inhibition. Thepattern of inhibition of proteinase 265 is in agreement withthe pattern obtained with various corresponding proteinasesof C. albicans and C. tropicalis (34, 36).

Purified proteinase 265 lost one-third of its activity afterincubation (22°C, 15 min) with hemin at pH 6.5; at pH 3.5approximately 60% of the activity was lost. Inhibition by

VOL. 53, 1986

on October 1, 2018 by guest

http://iai.asm.org/

Dow

nloaded from

416 RUCHEL ET AL.

H .@

F]ity (protc, clsecrC. Iparcing clightano(

henenctheproenztroj

redi(20Thusulfchelaffeheap

det(minredi0.2!tivefiveysisEm265T

bufiAft(10,ext(of tbutdist

Shancullapprefeaftemo]ratiproalb,

hemoglobin four times faster than did C. parapsilosis"> _ {~~~~~~36).

Human immunoglobulins (myeloma proteins of the Al andA2 subtypes) were hydrolyzed by purified proteinase 265 at

,, pH 3.5. The heavy chains were the prime targets of degra-~ + dation (Fig. 7a to d). Comparable degradation was seen with

a b c d e f g h polyclonal human IgG and the isolated heavy chain of ahuman IgGl myeloma protein. Live yeast cells in liquid

IG. 7. SDS-PAGE with Coomassie blue staining showing activ- culture hydrolyzed human secretory IgA from colostrum asof purified proteinase 265 at pH 3.5 against human myeloma well (Fig. 7e to h), provided that the cultures were agitatedeins of subtypes A2 (lanes: a, control; b, digest) and Al (lanes: constantly and supplemented repeatedly with glucose.ontrol; d, digest). Panels e to h show the hydrolysis of human Immunological aspects. Antibodies that were elicited*etory IgA when used as the sole nitrogen source in a culture of against purified proteinase 265 in guinea pigs cross-reactedparapsilosis (lanes: e, 2-day control incubation without C. with partially purified secretory proteinases of two unrelatedipsilosis; f, 2-day culture with the fungus; g and h, correspond ilte of C. parloisecretor an as demonrated:ontrol and culture after 4 days of incubation. H, heavy chain; L, isolates of C parapsilosis (21/78 and 78K), as demonstratedtchain. The arrow indicates migration of proteins toward the by enzyme-linked immunosorbent assay. However, cross-de. reactivity of the antibodies was not confined to proteinases

from isolates of the same species, in that antibodies againstnin depends on hydrophobic interactions, since the pres- proteinase 265 reacted with secretory proteinase of C. albi-e of 0.05% nonionic detergent (Emulphogene) prevented cans and C. tropicalis (Fig. 5). On the other hand, antibodieseffect. Sensitivity to hemin is a specific feature of that discriminated between proteinases of C. albicans and C.

teinase 265; it was not found with the corresponding tropicalis (36) reacted weakly with proteinase of C. parapsi-;ymes of C. albicans HP and CBS 2730 or that of C. losis. This pattern of immunological reactivity suggests the)icalis 293/80. existence of common and species-specific domains in thelo inhibition of proteinase 265 was caused by exposure to secretory acid Candida proteinases.ucing agent (20 mM dithiothreitol) or carboxymethylation Proteinase production following phagocytosis. Cultures ofmM iodoacetamide), both at pH 6 and 22°C for 15 min. murine peritoneal macrophages were chosen as targets foris, neither intact disulfide bonds nor functional infection by C. parapsilosis in vitro. Adherent macrophageshydryls are essential for its enzymic activity. Also, that had been kept for 24 h in cell culture were infected withlating agents such as o-phenanthroline and EGTA did not blastospores at a cell ratio of 1:3. The experiments werect the enzyme, excluding a functional role of Ca2" and performed in parallel with blastospores of the proteolyticLvy metal ions. reference strain C. albicans CBS 2730+. After 20 h at 37°C,roteinase 265 was resistant to treatment with nonionic the cocultivation was terminated by fixation with paraform-ergent (Emulphogene BC-720 up to 1% at 37°C for 90 aldehyde, and the samples were submitted to indirect immu-). It did not tolerate SDS in that 0.03% (22°C, 90 min) nofluorescence with polyvalent anti-proteinase antibody asuced the activity of proteinase 265 by 40%. At 0.12 and the first antibody (for details, see Materials and Methods).5%, inhibition by detergent was 85 and 100%, respec- Blastospores of both C. albicans and C. parapsilosis hadly. A trial for the displacement of bound SDS by a been ingested in large numbers by the phagocytes (Fig. 8).fold excess of Emulphogene followed by prolonged dial- However, Candida proteinase was detectable only on cellsi against 10 mM citrate buffer (pH 6.2) and 0.05% of C. albicans (Fig. 8c). Samples with C. parapsilosisulphogene did not restore the activity of proteinase resembled the controls (Fig. 8b). This unexpected difference

may provide an explanation for the largely differing'he enzyme could be precipitated from 20 mM citrate virulences of the Candida species in vivo.fer (pH 6.2) by addition of acetone (4 volumes, 0WC).er 12 h on ice, a pellet was collected by centrifugation DISCUSSION,000 x g, 10 min), which was dissolved and dialyzedensively against the starting buffer. Approximately 50% C. albicans, C. tropicalis, and C. parapsilosis are distin-;he enzymic activity was recovered from the precipitate, guished from other yeastlike fungi by the secretion ofno activity was detected if the pellet was dissolved in proteinase in vitro (17, 36). For C. albicans, secretory

tilled water. proteolytic activity is considered an important factor in;ubstrate specificity. Like all other Candida proteinases virulence (for a review see reference 20a), and the same mayIdled so far, proteinase 265 cleaved BSA. In liquid be true for C. tropicalis (4). For C. parapsilosis, a correla-ture, hydrolysis of BSA by C. parapsilosis 265 was tion between proteolytic activity and virulence has not been)arent after 24 h, whereas the proteolytic activity of the established. In spite of its proteolytic potential, the virulencewrence strain, C. albicans CBS 2730, was apparent only of C. parapsilosis is low in vitro (4) as well as in humans and-r 48 h. Beyond 48 h, the proteinase of C. albicans was animals (2, 6).re effective than that of C. parapsilosis. The changing To elucidate the questionable contribution of secretorylo of hydrolysis reflects the broader pH optimum of proteinase to fungal virulence, we undertook the purificationiteinase 265 and the superior proteolytic potential of C. and characterization of the acid proteinase of C. parapsilosisicans at low pH. Thus, at pH 3.5, C. albicans hydrolyzed 265. This strain is an isolate derived from a case of urinary

FIG. 8. Mouse peritoneal macrophages infected with C. parapsilosis 265 (a and b) or C. albicans CBS 2730+ (c). (a) Phase-contrastmicroscopy showing ingested yeast cells 20 h after infection of the cell culture. (b and c) Samples treated with anti-proteinase antibody anda fluorescein-conjugated second antibody. Note the distortion of the phagocytes and the proteinase-dependent fluorescence of ingested yeastcells in c compared with the round shape of the cells and weak image of yeast cells (<-) in b. The bar equals 9 ,um.

INFECT. IMMUN.

on October 1, 2018 by guest

http://iai.asm.org/

Dow

nloaded from

VOL. 53, 1986 SECRETORY PROTEINASE OF C. PARAPSILOSIS 417

op

<awe1 S%W9,

a' o i 7e *1-t sX- 'g:~~ ¼t>*rr.b x i v # , s Z tr + l%\wIF

W ts, st' 1 1 s .., ~~~~~A

on October 1, 2018 by guest

http://iai.asm.org/

Dow

nloaded from

418 RUCHEL ET AL.

tract infection; it was strongly proteolytic on serum albuminagar (36). However, the proteolytic activity of strain 265 wasmore dependent on the supply of glucose (.1%) than wasthe proteolytic reference strain, C. albicans CBS 2730 (un-published data). In liquid medium with 1% glucose andhemoglobin as the sole nitrogen source, C. parapsilosis 265secreted acid proteolytic activity up to an apparent concen-tration of 5 Fg/ml, which is comparable to that released by C.albicans CBS 2730 (29). The enzyme of C. parapsilosis 265,proteinase 265, was classified as an aspartic proteinase (EC3.4.23). Purification of the enzyme was attempted along linesthat had proved useful with proteinases of C. albicans and C.tropicalis (29, 36). A particular problem was the almost totalloss of enzyme in the DEAE column. Hydrophobic forceswere responsible for the loss, since a nonionic detergent wasable to prevent it. Emulphogene BC-720 was chosen, since itlacks the phenolic ring of the related detergents Triton X-100and Nonidet P-40; thus, it does not interfere with themonitoring of proteins at 280 nm (8, 11).The particular hydrophobicity of proteinase 265 was also

revealed by its sensitivity to hemin inhibition. Such inhibi-tion had been observed previously with a membrane protein-ase of erythrocytes (25). Proteinase 265 underwent alkalinedenaturation at pH 7 and room temperature. More stableenzymes were found only with C. albicans (31). The profileof alkaline denaturation may govern the pathogenic potentialof Candida proteinases and thus influence the virulence ofthe fungus.

Alkaline denaturation certainly was the reason for the lossof enzymic activity on pepstatin affinity chromatography(unpublished data), which had been applied successfully tothe stable enzyme of C. albicans CBS 2730 (29). Instead, thepurification of proteinase 265 had to be continued by ordi-nary gel filtration, which yielded a virtually clean enzyme atan acceptable loss. A high degree of purification wasachieved as well by preparative electrofocusing. However,the yield was very low. Inactivation of the proteinase onelectrofocusing was accompanied by a shift in apparent pI.Inactivation was not due to autolysis; it may involve inter-action with carrier ampholytes. The estimation of the mo-lecular weight yielded values around 33,000. Thus, protein-ase 265 is considerably smaller than the secretoryproteinases of C. albicans and C. tropicalis, which are in therange of 45,000 (32).

Proteinase 265 and the related enzymes of C. albicansCBS 2730 and C. tropicalis 293 did not react with the lectinsConA and WGA. This indicates that the enzymes lackterminal mannose, glucose, and N-acetylglucosamine resi-dues (23). Lack of reaction with ConA should exclude thepresence of mannan (23). Consequently, proteinase 265could not be purified on ConA-Sepharose, and it did notreact with antibodies against mannan. However, the protein-ase of C. tropicalis 293 clearly reacted with anti-mannanantibody. Thus, certain Candida proteinases may indeed bemanno proteins as has been suggested previously (18). Theproteinase of C. albicans CBS 2730 may not be glycosylatedat all, since it resisted degradation by endoglycosidase F.Proteinase 265 and the related enzymes of C. albicans CBS2730 and C. tropicalis 293 also failed to stain for boundphosphate (unpublished data).

Proteinase 265 is a potent antigen in guinea pigs. Theantibodies against proteinase 265 cross-reacted strongly withproteinases of two unrelated isolates of the same species. Aspecies-specific relationship of Candida proteinases hadalready been found among enzymes of C. albicans and C.tropicalis (36). Such discriminating antibodies did not cross-

react with proteinases of C. parapsilosis or unrelated acidproteinases, including porcine pepsin (31) and human renin(a gift from B. Leckie, Glasgow, Scotland). On the contrary,antibodies against proteinase 265 cross-reacted withproteinases from other Candida species and porcine pepsin.These antibodies may be directed against sites that arecommon to aspartic proteinases, whereas the discriminatingantibodies recognize species-specific sites.

Proteinase 265 cleaved a number of immunoglobulins,including secretory IgA and both isotypes IgAl and A2. Thelatter is resistant to the IgA proteinases of various patho-genic bacteria (24). Cleavage of immunoglobulins in vitro isa feature of all of the Candida proteinases investigated sofar, and there are indications that such cleavage takes placein the course of infection by C. albicans (32). Living cells ofC. parapsilosis were able to utilize sIgA as a nitrogensource. Whether this process fosters the colonization ofmucosa by yeast cells needs to be investigated.

C. parapsilosis is conspicuously involved in the pathogen-esis of endocarditis (28). The fibrinaceous vegetations thatsurround cells of this yeast on infected heart valves (20)suggest that the yeast generates a procoagulant substance.Preliminary results with proteinase 265 indicate that thisenzyme is able to convert prothrombin and factor X (unpub-lished data), as has been found previously with C. albicansand C. tropicalis (30).

Isolates of C. parapsilosis were less cytotoxic versusmonocytelike cells (U 937) than were any of the testedisolates of C. albicans and C. tropicalis (4). C. parapsilosisalso had a low toxicity for murine pentoneal macrophages.The virulence of strain 265 in NWNI mice was minimal,which is in agreement with the results of Bistoni et al. (2).The lack of virulence may be related to the fact that C.parapsilosis 265 did not produce acid proteinase underconditions of infection in vitro. On infection of cell cultureswith C. albicans (and C. tropicalis), secretory fungal pro-teinase was clearly demonstrable. In mice, anti-proteinaseantibody was readily detectable after infection with C.albicans and C. tropicalis. No specific titer, however, couldbe detected after infection with C. parapsilosis 265, althoughpurified proteinase 265 was an effective antigen in mice ofthe same breed (unpublished data). These results suggestthat C. parapsilosis is unable to produce secretory acidproteinase on infection in vivo. This deficiency may be thereason for the comparatively low virulence of this proteo-lytic yeast.

ACKNOWLEDGMENTS

We are indebted to A. Siegmann for preparation of the manu-script, and to H. Wilmanns for assistance in a number of experi-ments.The investigation was supported by a grant from the Deutsche

Forschungsgemeinschaft.

LITERATURE CITED1. Ahearn, D. G. 1978. Medically important yeasts. Annu. Rev.

Microbiol. 32:59-68.2. Bistoni, F., A. Vecchiarelii, E. Cenci, G. Sbaraglia, S. Perito, and

A. Cassone. 1984. A comparison of experimental pathogenicityof Candida species in cyclophosphamide-immunodepressedmice. Sabouraudia 22:409-418.

3. Bodey, G. P., and V. Fainstain. 1985. Candidiasis. Raven Press,New York.

4. Borg, M., D. Kirk, H. Baumgarten, and R. Ruchel. 1984. Acolorimetric assay for the assessment of cytotoxicity of yeasts.Sabouraudia 22:357-367.

5. Cutting, J. A., and T. F. Roth. 1973. Staining of phospho-

INFECT. IMMUN.

on October 1, 2018 by guest

http://iai.asm.org/

Dow

nloaded from

SECRETORY PROTEINASE OF C. PARAPSILOSIS 419

proteins on acrylamide gel electropherograms. Anal. Biochem.54:386-394.

6. Dyess, D. L., R. N. Garrison, and D. E. Fry. 1985. Implicationsof polymicrobial blood-borne infection. Arch. Surg. 120:345-348.

7. Elder, J. H., and S. Alexander. 1982. Endo-p-N-acetylglu-cosaminidase F: endoglycosidase from Flavobacteriummeningosepticum that cleaves both high-mannose and complexglycoproteins. Proc. Natl. Acad. Sci. USA 79:4540-4544.

8. Furth, A. J., H. Bolton, J. Potter, and J. D. Priddle. 1984.Separating detergent from proteins. Methods Enzymol. 104:318-328.

9. Glass, W. F., R. C. Briggs, and L. S. Hnilica. 1981. Use oflectins for detection of electrophoretically separated glycopro-teins transferred onto nitrocellulose sheets. Anal. Biochem.115:219-224.

10. Gripon, J. C., and T. Hofmann. 1981. Inactivation of aspartylproteinases by butane-2,3-dione. Biochem. J. 193:55-65.

11. Helenius, A., and K. Sinons. 1975. Solubilization of membranesby detergents. Biochim. Biophys. Acta 415:29-79.

12. Herzoq, V., and H. D. Fahimi. 1973. A new sensitive colorimet-ric assay for peroxidase using 3,3'-diaminobenzidine as hydro-gen donor. Anal. Biochem. 55:554-562.

13. Johnson, G. D., R. S. Davidson, K. C. McNamee, G. Russell, D.Goodwin, and E. J. Holborow. 1982. Fading of immunofluores-cence during microscopy: a study of the phenomenon and itsremedy. J. Immunol. Methods 55:231-242.

14. Kaehn, K., M. Morr, and M. R. Kula. 1979. Inhibition of theacid proteinase from Neurospora crassa by diazoacetyl-DL-norleucine methyl ester, 1,2-epoxy-3-(4-nitrophenoxy)propaneand pepstatin. Hoppe-Seyler's Z. Physiol. Chem. 360:791-794.

15. Kwon-Chung, K. J., D. Lehman, C. Good, and P. T. Magee.1985. Genetic evidence for role of extracellular proteinase invirulence of Candida albicans. Infect. Immun. 49:571-575.

16. Lanoe, J., and J. Dunnigan. 1978. Improvements of the Ansonassay for measuring proteolytic activities in acidic pH range.Anal. Biochem. 89:461-471.

17. MacDonald, F. 1984. Secretion of inducible proteinase by patho-genic Candida species. Sabouraudia 22:79-82.

18. MacDonald, F., and F. C. Odds. 1980. Inducible proteinase ofCandida albicans in diagnostic serology and in the pathogenesisof systemic candidosis. J. Med. Microbiol. 13:423-435.

19. MacDonald, F., and F. C. Odds. 1983. Virulence for mice of aproteinase-secreting strain of Candida albicans and a protein-ase-deficient mutant. J. Gen. Microbiol. 129:431-438.

20. Marrie, T. J., J. H. Cooper, and J. W. Costerton. 1984. Ultra-structure of Candida parapsilosis endocarditis. Infect. Immun.45:390-398.

20a.Odds, F. C. 1985. Candida albicans proteinase as a virulencefactor in the pathogenesis of Candida infections. Zentralbl.Bakteriol. Mikrobiol. Hyg. A 260:539-542.

21. Osborn, M., and K. Weber. 1977. The detergent-resistant cyto-skeleton of tissue culture cells includes the nucleus and themicrofilament bundles. Exp. Cell Res. 106:339-349.

22. Peat, S., W. J. Whelan, and T. E. Edwards. 1961. Polysac-charides of baker's yeast. Part IV. Mannan. J. Chem. Soc.1:29-34.

23. Pistole, T. G. 1981. Interaction of bacteria and fungi with lectinsand lectin-like substances. Annu. Rev. Microbiol. 35:85-112.

24. Plaut, A. G. 1983. The IgAl proteases of pathogenic bacteria.Annu. Rev. Microbiol. 37:603-622.

25. Pontremoli, S., F. Salamino, B. Sparatore, E. Melloni, A.Morelli, U. Benatti, and A. De Flora. 1979. Isolation and partialcharacterization of three acidic proteinases in erythrocyte mem-branes. Biochem. J. 181:559-568.

26. Radola, B. J. 1984. High-resolution preparative isoelectric fo-cusing. Methods Enzymol. 104:256-275.

27. Renart, J., and 1. V. Sandoval. 1984. Western blots. MethodsEnzymol. 104:455-460.

28. Rubinstein, E., E. R. Noriega, M. S. Simberkoff, R. Holzman,and J. J. Rahal. 1975. Fungal endocarditis: analysis of 24 casesand review of the literature. Medicine 54:331-344.

29. Ruchel, R. 1981. Properties of a purified proteinase from theyeast Candida albicans. Biochim. Biophys. Acta 659:99-113.

30. Ruchel, R. 1983. On the renin-like activity of Candidaproteinases and activation of blood coagulation in vitro.Zentralbl. Bakteriol. Mikrobiol. Hyg. I Abt. Orig. A 255:368-379.

31. Ruchel, R. 1983. On the role of proteinases from Candidaalbicans in the pathogenesis of acronecrosis. Zentralbl.Bakteriol. Mikrobiol. Hyg. I. Abt. Orig. A 255:524-536.

32. Ruchel, R. 1984. A variety of Candida proteinases and theirpossible targets of proteolytic attack in the host. Zentralbl.Bakteriol. Mikrobiol. Hyg. I Abt. Orig. A 257:266-274.

33. Ruchel, R., and B. Boning. 1983. Detection of Candida protein-ase by enzyme immunoassay and interaction of the enzyme withalpha-2-macroglobulin. J. Immunol. Methods 61:107-116.

34. Ruchel, R., B. Boning, and E. Jahn. 1985. Identification andpartial characterization of two proteinases from the cell enve-lope of Candida albicans blastospores. Zentralbl. Bakteriol.Mikrobiol. Hyg. I Abt. Orig. A 260:523-538.

35. Ruchel, R., and J. Gross. 1979. Preparations of continuousgradient gel slabs: a simple technique. Anal. Biochem. 92:91-98.

36. Ruchel, R., K. Uhlemann, and B. Boning. 1983. Secretion of acidproteinases by different species of the genus Candida. Zentralbl.Bakteriol. Mikrobiol. Hyg. I Abt. Orig. A 255:537-548.

37. Staib, F. 1965. Serum-Protein-Agar-pH 4,2 und 5,0 furSprof3pilze. Zentralbl. Bakteriol. Mikrobiol. Hyg. I Abt. Orig. A195:265-267.

38. Staib, F. 1969. Proteolysis and pathogenicity of Candida albi-cans strains. Mycopathologia 37:345-348.

39. Staib, F., S. K. Mishra, B. Tompak, G. Grosse, T. Abel, A.Blisse, U. Folkens, and B. Frohlich. 1980. Pathogenic yeast-likefungi in meat products. Zentralbl. Bakteriol. Hyg. I. Abt. Orig.A 248:422-429.

40. Sundsmo, J. S., and 0. Gotze. 1980. Human monocyte spreadinginduced by factor B of the alternative pathway of complementactivation. Cell. Immunol. 52:1-17.

VOL. 53, 1986

on October 1, 2018 by guest

http://iai.asm.org/

Dow

nloaded from