Developmental expression of a tandemly repeated, proline-and glutamine-rich amino acid motif on hyphal surfaces on Candida albicans.

cDNA sequences encoding a cell wall protein have been isolated from the opportunistic pathogen, Candida albicans, an organism that can cause serious disease in immunocompromised patients such as those with AIDS. The cDNA encodes a peptide that is largely composed of an acidic, repeated motif 10 amino acids in length that is rich in proline and glutamine residues. The cDNA gene product was found to be present on hyphal surfaces by immunofluorescence assays using monospecific antisera raised to the recombinant protein produced in Pichia pastoris. The hyphae-specific surface location was also seen on organisms colonizing the gastrointestinal mucosa of mice, indicating that the antigen is produced and developmentally regulated during growth in host tissues. The cDNA clone hybridized to an abundant messenger RNA 2.3 kilobases in size that was present in hyphal but not yeast forms. These studies demonstrate that the bud-hypha transition is accompanied by the de novo synthesis of proteins that are targeted to hyphal surfaces. The primary sequence of the unique amino acid motif shares features with surface proteins of other lower eukaryotic microorganisms and with host acidic salivary proline-rich proteins.

cDNA sequences encoding a cell wall protein have been isolated from the opportunistic pathogen, Candida albicans, an organism that can cause serious disease in immunocompromised patients such as those with AIDS. The cDNA encodes a peptide that is largely composed of an acidic, repeated motif 10 amino acids in length that is rich in proline and glutamine residues. The cDNA gene product was found to be present on hyphal surfaces by immunofluorescence assays using monospecific antisera raised to the recombinant protein produced in Pichia pastoris. The hyphae-specific surface location was also seen on organisms colonizing the gastrointestinal mucosa of mice, indicating that the antigen is produced and developmentally regulated during growth in host tissues. The cDNA clone hybridized to an abundant messenger RNA 2.3 kilobases in size that was present in hyphal but not yeast forms. These studies demonstrate that the bud-hypha transition is accompanied by the de novo synthesis of proteins that are targeted to hyphal surfaces. The primary sequence of the unique amino acid motif shares features with surface proteins of other lower eukaryotic microorganisms and with host acidic salivary proline-rich proteins.
The cell surface structures of Candida albicans mediate several important processes in the molecular interactions that occur between host and parasite. To establish a commensal host parasite relationship, C. albicans must be adherent enough to persist on mucosal surfaces where cell-mediated immune responses are induced that serve to counteract overgrowth. When risk factors such as immunosuppression due to AIDS or chemotherapy are present, fungal surface structures mediate attachment to and invasion of host tissues. Thus surface structures of C. albicans are of central importance in the complex balancing mechanisms that serve to maintain the host-parasite equilibrium and are also involved in pathogenesis.
The facile transition between yeast and hyphal growth and the presence of intermediate pseudohyphal forms are hallmarks of C. albicans that are accompanied by molecular sur-face changes that confer differing antigenic, chemical, and physical surface properties to the growth forms (1)(2)(3)(4)(5)(6)(7)(8)(9)(10)(11). Thus, the existence of yeast and hyphal forms increases the repertoire of surface structures of C. albicans that are available for survival within the host environment. The differences between growth forms in interactions with phagocytic cells (12)(13)(14)(15), adherence to epithelial cells (16), and binding to host proteins (17)(18)(19)(20)(21) probably result from the differing surface biochemical compositions of the growth forms.
Although the numerous binding activities that have been attributed to hyphae involve a variety of host cells and host proteins, a frequent finding is that surface proteins of hyphae are involved in binding (8,9,15,17). Therefore it is surprising that primary structures of candidal hyphal wall proteins have not been described. We have shown that unique antigens exist on hyphal surfaces (8) and propose that identification of the primary sequences of these antigens will provide insight into the structure and function of hyphal surfaces. In this report, we describe a novel proline-and glutamine-rich amino acid segment that is exposed on surfaces of hyphae grown in mammalian hosts and in laboratory cultures. Synthesis of this protein occurred exclusively during hyphal growth, showing that the bud-hypha transition controls the antigenic surface composition of hyphae by production of de novo proteins. The predicted protein sequence provided new insights into possible mechanisms of interaction between C. albicans and host cells.

EXPERIMENTAL PROCEDURES
C. albicans Strains-C. albicans strain SC5314 (22) was used in all experiments unless otherwise indicated. Organisms were stored at Ϫ70°C and cultured on yeast peptone dextrose agar plates at room temperature according to standard techniques (23).
DNA Manipulations-All restriction enzymes, T4 ligase, and DNA polymerase Klenow fragment were purchased from Promega Biotech (Madison, WI) or Life Technologies, Inc. and were used according to the manufacturers' instructions. Deoxynucleotide stocks were purchased from Boehringer Mannheim. Recombinant plasmids were maintained in Escherichia coli strains SURE or SURE2 (Stratagene, La Jolla, CA) to minimize cloned DNA rearrangements. DNA inserts were purified using Geneclean II (BIO 101, La Jolla, CA). Synthetic oligos were prepared by Genosys Biotechnologies, Inc. (The Woodlands, TX).
The 5Ј region of HWP1 mRNA was amplified using the 5Ј RACE System kit (Life Technologies, Inc.) according to the manufacturer's instructions. First strand cDNA synthesis was performed on total RNA isolated from yeast and hyphal forms using a gene-specific oligo (5Ј-GGGTAATCATCACATGG) complementary to nucleotides 199 -214. A second nested gene-specific oligo (5Ј-GATAGTAATCATAAGATCTC) complementary to nucleotides 174 -193 was used as a primer to amplify the 5Ј end using PCR. The resulting PCR 1 product was cloned directly into pBluescript SKϪ (Stratagene, La Jolla, CA) that had been digested with EcoRV and treated with TaqPlus DNA polymerase (Stratagene) in the presence of dTTP to generate a T-vector (24). The nucleotide sequence of four independent clones was determined (U. S. Biochemical Corp.).
Antisera to C. albicans-Berkeley Antibody Company (Richmond, CA) prepared rabbit anti-C. albicans antiserum to strain SC5314 bearing germ tubes. Upon adsorption with yeast forms as described previously (8), the antiserum was specific for hyphal surfaces when assayed by indirect immunofluorescence and was denoted hyphae-specific antiserum. Immunofluorescence assays were performed as previously (8) described using goat anti-rabbit IgG (HϩL) conjugated to fluorescein isothiocyanate (Zymed Laboratories Inc., South San Francisco, CA) at a dilution of 1:50. Additional antiserum adsorptions with an E. coli ZAP crude extract, a Zymolyase digest of yeast cells that had been coupled to a solid matrix, and a soluble, purified C. albicans enolase-GST fusion protein (25) were performed prior to screening a cDNA library to reduce the levels of antibodies to internal proteins common to yeast and hyphae. The adsorbed antiserum was affinity purified with protein A (25,26). The titer of the resulting antiserum for hyphal surfaces was 1:400.
Histology and Immunofluorescence Staining of Tissues of Colonized Mice-Paraffin-embedded gastrointestinal tissues from 11-month-old female bg/bg-nu/ϩ mice that had been colonized for 9 months with C. albicans B311 were generously provided by Edward Balish. Stomach, esophagus, and tongue tissues were sectioned (4 m) and stained with the periodic acid-Schiff reagent or with rabbit antiserum as described previously (27) except that blocking with bovine serum albumin was carried out at room temperature. Primary antibodies were rabbit antiserum to C. albicans (Berkeley Antibody Company) that had not been adsorbed and preimmune serum (1:100), as well as anti-rhwp1 (see below) and preimmune rhwp1 serum (1:50). Secondary antibodies were fluorescein isothiocyanate-conjugated goat anti-rabbit IgG (HϩL chains) (Zymed Laboratories Incorporated) at a 1:50 dilution.
Screening of the cDNA Library-The construction of the cDNA library in ZAP II from hyphal mRNA and production of plaques containing recombinant C. albicans proteins for immunoscreening have been previously described (25). This screening resulted in the identification of 139 positive plaques that were stained by the antiserum. Twelve plaques from the initial screening were purified by plating at low density and rescreened with antiserum until all plaques from each original isolate were positive. Bluescript SKϪ phagemids containing inserts from the positive plaques were rescued by the in vivo excision protocol using Exassist helper phage and Solr E. coli cells (Stratagene) according to the manufacturer's directions.
Purified clones were further tested for their ability to selectively adsorb antibodies that cross-reacted with hyphal surfaces from the anti-C. albicans antiserum. Each of the purified clones was used to produce nearly confluent isopropyl-1-thio-␤-D-galactopyranoside-induced plaques on 100-mm NZY agar plates that were then transferred to nitrocellulose. Filters were soaked overnight in TBST (20 mM Tris, pH 7.5, 50 mM NaCl, and 0.05% Tween 20) and washed three times in TBST prior to blocking with TBST containing 1% skim milk, 1 mM EDTA, and 0.02% sodium azide. Filters were washed and incubated in the anti-C. albicans antiserum described above in blocking solution for 2 h at 37°C. Unattached antibodies were removed with five washes in TBST followed by one wash in 0.15 M NaCl containing 0.05% Tween 20. Bound antibodies were eluted with a solution containing 0.2 M glycine, pH 2.8, 0.15 M NaCl, 1 mM EDTA, and 0.05% Tween 20. Antibodies were immediately neutralized with Tris base and concentrated using Centricon membranes (Amicon, Beverly, MA). The concentrated antibodies were assayed for their ability to bind hyphal surfaces in the immunofluorescence assay (8).
DNA Sequencing and Analysis-The DNA sequences of positive clones were determined (U. S. Biochemical Corp.) on plasmid DNA. One phagemid clone (pBSϩ13), that harbored a 609-base pair insert (HWP1), was used in subsequent studies. The complete sequence of HWP1 on both strands was derived from nested clones produced using the Erase-a-Base System (Promega Biotech). Sequence analysis was carried out using the DNA Inspector IIE program (Textco, Boston, MA) and Software from the Genetics Computer Group at the University of Wisconsin (28). Computations to identify similar proteins in nonredundant protein data bases according to the algorithm of Altschul et al. (29) were performed at NCBI using the BLAST network service. The sequence has been placed in Genbank. The accession number is U29369.
Northern Blot Analysis-Total RNA was isolated from C. albicans SC5314 grown under conditions that promote yeast or hyphal morphologies. Growth conditions in modified Lee's medium (30) and M199 (Life Technologies, Inc.) were the same as described previously (26,31), except that an isolated colony from a yeast peptone dextrose agar plate, rather than stationary phase organisms grown in broth, was used to inoculate M199. The cultures were examined microscopically after 3 h of incubation prior to RNA isolation. Cells incubated in Lee's medium, pH 4.5 and 6.5, at 25°C were growing as budding yeasts (Ͼ99%); cells incubated in Lee's medium, pH 4.5, at 37°C were growing as budding yeasts, with a few cells (10 -15%) growing as pseudohyphae; and cells incubated in Lee's medium, pH 6.5, at 37°C possessed germ tubes (Ͼ99%). 80% of the cells grown in M199 for 4 h had germinated.
Procedures for electrophoresis, transfer, and hybridization conditions and radiolabeling of probes were similar to those previously described (26,32) except that 2 ϫ 10 6 cpm of gel-purified HWP1 and cENO were used and the incubation temperature during hybridization was 37°C. Molecular sizes of mRNAs were determined using a standard curve based on gel migrations of RNA standards (0.24 -9.5 kilobases) (Life Technologies, Inc.).
Expression of Recombinant Hyphal Wall Protein Antigen-The partial C. albicans cDNA clone was incorporated into the P. pastoris expression vector pPIC9 (Invitrogen, San Diego, CA) in frame with the ␣-factor secretion signal of Saccharomyces cerevisiae as follows: the 691-base pair XbaI-XhoI fragment from pBSϩ13 harboring HWP1 cDNA (in addition to ␤-galactosidase DNA sequences 5Ј and 3Ј to the clone) was treated with Klenow fragment to produce blunt ends (33) and then ligated to the pPIC9 SnaBI site. The reading frame of HWP1 relative to the ␣-factor secretion signal was verified via DNA sequencing (U. S. Biochemical Corp.). The recombinant plasmid, pPIC13, was then transformed into P. pastoris strain GS115. Hisϩ, mut colonies were tested for their ability to secrete the C. albicans gene product into cell culture supernatants as directed by the manufacturer. Secreted proteins were examined by Western blotting (25,34) using either anti-C. albicans or hyphae-specific antiserum (IgG fraction) at dilutions of 1:400, followed by incubation with goat anti-rabbit IgG (L ϩ H chains) conjugated to horseradish peroxidase (Zymed Laboratories Inc.) and developed with ECL reagents (Amersham Corp.) according to the manufacturer. One P. pastoris transformant (GS115-13) secreting relatively large amounts of rhwp1 was used in further studies.
Purification of Recombinant hwp1 and Production of Monospecific Antiserum-P. pastoris GS115-13 culture medium containing rhwp1 was electrophoresed in 16 ϫ 18-cm slab gels by standard methods (35), and the protein band corresponding to hwp1 was excised. Recombinant hwp1 was electroeluted from the gel slices (Centrilutor, Amicon), and its purity was assessed by two-dimensional gel electrophoresis (36) using a SE250 unit (Hoefer Scientific Instruments, San Francisco, CA) followed by Western blotting using the anti-C. albicans antiserum described above at a 1:200 dilution. Primary rabbit antibodies were detected using the streptavidin-alkaline phosphatase kit as directed by the manufacturer (Zymed Laboratories Inc.). The Western blot of the electroeluted hwp1 showed a predominant protein near the anode as predicted for hwp1 and a protein present in lesser amounts near neutral pH, indicating the presence of a comigrating P. pastoris protein.
Recombinant hwp1 was purified by standard column chromatography. P. pastoris GS115-13 culture supernatant (30 ml) was fractionated by (NH 4 ) 2 SO 4 precipitation; the fraction enriched for rhwp1 (40 -60% saturation) was dialyzed against column buffer (50 mM Tris-Cl, pH 7.5, 1 mM EDTA, 10% glycerol, 2 mM 2-mercaptoethanol, and 1 mM phenylmethylsulfonyl fluoride) and applied to a 2.5 ϫ 18-cm DEAE-Sephacel (Pharmacia Biotech Inc.) column at 4°C. The bound proteins were eluted with a linear gradient of KCl from 0.0 to 0.5 M. Aliquots of column fractions were tested for rhwp1 by SDS-polyacrylamide gel electrophoresis (PAGE) and Western blotting, followed by immunostaining. Fractions containing rhwp1 were pooled and concentrated. A single protein adjacent to the anode was seen following two-dimensional SDS-PAGE and Western blotting as described above showing that rhwp1 was pure. Attempts to quantitate rhwp1 with Coomassie-based protein assays and derivative methods (Lowry and biuret enhanced with bicinchoninic acid; Pierce) were unsuccessful. Therefore the absorbance at 205 nm was used to approximate the protein concentration (37). The molecular size of rhwp1 was determined using a standard curve based on gel migrations of purchased protein standards ranging from 14.3 to 200 kDa (Rainbow Markers, Amersham Corp.). The amino acid composition of rhwp1 was determined at the Ohio State Biochemical Instrument Center.
Monospecific antiserum to rhwp1 was prepared by immunizing New Zealand White rabbits with rhwp1 that had been excised from onedimensional SDS-PAGE as described above, using conventional methods (38). Antibodies to P. pastoris were removed by adsorptions with P. pastoris GS115 untransformed cells. The immune serum (1:200 dilution in phosphate-buffered saline (170 mM NaCl in 10 mM potassium phosphate buffer, pH 7.4)) was used for binding to surfaces of C. albicans using the immunofluorescence assay (8). Immunofluorescence blocking experiments were performed by incubating chromatographically purified rhwp1 (see above) at concentrations of 4.25 or 8.5 g/ml at 37°C for 15 min with diluted primary antibody before applying to C. albicans cells. In control experiments, irrelevant proteins enolase and lactate dehydrogenase (Boehringer Mannheim) were used at concentrations of 1.25 mg/ml. Detection of Native hwp1 by Western Blotting-C. albicans cells were grown to stationary phase in yeast nitrogen base (Difco, Detroit, MI) with 50 mM glucose and diluted into two flasks of M199 at cell densities of 5 ϫ 10 6 /ml. The cultures were grown for 2.5 h at room temperature (yeasts) and at 37°C (yeasts bearing germ tubes) with gentle shaking. Cells were digested enzymatically as in previous studies (39,40) except that Zymolyase 20T 6 mg/ml (ICN, Biomedicals, Inc., Costa Mesa, CA), was used to digest the cells and protease inhibitors EDTA (2.6 mM), pefabloc (0.26 mM) (Boehringer Mannheim), leupeptin, and pepstatin (2.6 and 2.8 M respectively) (Sigma) were included. Digestions were carried out for 1 h at 37°C with rocking. Following centrifugation to remove particulates, digests were desalted using 10 DG columns (Bio-Rad) prior to separation on an analytical SDS-PAGE gel (12% acrylamide) using 6.5 g of protein per lane. Immunoblotting and molecular size determinations were performed as described for recombinant hwp1 except that the primary antibody was monospecific rabbit anti-rhwp1 antiserum (1:100 dilution). For blocking experiments, monospecific rabbit anti-rhwp1 serum was incubated with 7.08 and 14.2 g/ml of chromatographically purified rhwp1 at 37°C for 15 min prior to incubating the antiserum with immunoblots. Enolase (31.3 g/ml) was used as a control protein in blocking experiments.

Immunoselection and Preliminary
Characterization of Recombinant cDNA Encoding Surface Proteins of C. albicans-Previous attempts to obtain clones for surface proteins resulted in isolation of cDNA clones for cytoplasmic proteins. To increase the probability of isolating a clone encoding a hyphal cell wall antigen, the antiserum was adsorbed with yeast proteins as described under "Experimental Procedures." In addition, DNA sequencing of insert ends from plasmid DNA of purified clones and sequence comparison with proteins in national data bases was performed to help identify undesired clones encoding highly conserved housekeeping protein cDNAs. Two of the 12 clones initially selected were not homologous to other sequences in the data bases and were found to have nearly identical DNA sequences at their 5Ј and 3Ј ends.
Preliminary results showed that two clones were able to select antibodies from the screening antiserum that cross-reacted with hyphal surfaces of C. albicans in an immunofluorescence assay that employs germ tube forms as antigens (not shown). Yeast surfaces were negative or weakly fluorescent. In control experiments, antibodies selected by a clone containing cDNA encoding the cytoplasmic enzyme enolase did not stain C. albicans surfaces, indicating that the reactivity with hyphal surfaces of antibodies selected by the unknown clones was not a result of nonspecific binding of immunoglobulins. One of the positive clones, 13, was selected for further study. The cDNA from this clone was termed HWP1 for hyphal wall protein 1, and the protein product, hwp1.
DNA Sequence of HWP1-The DNA sequence of HWP1 was found to contain two open reading frames (ORFs) that trans- versed the entire clone. Initially, we suspected ORF1 to be the correct reading frame because of the similarity between the codon usage profiles of ORF1 but not ORF2 with that of other highly expressed C. albicans genes, EF1-␣ and enolase (25,31). In addition, ORF1 was found to be in frame with ␤-galactosidase, providing further evidence that ORF1 was the correct reading frame because only those clones fused in frame with ␤-galactosidase would be translated, a requirement for synthesis of native antigens that would be recognized by screening antibodies. To further confirm that ORF1 was in frame with an ATG start codon, a PCR product representing the 5Ј end of HWP1 message was isolated. Oligonucleotide primers internal to the partial cDNA were used in combination with C. albicans hyphal total RNA to generate a PCR product that represented the 5Ј end of HWP1 message according to the rapid amplification of cDNA ends protocol (41). No PCR products were produced when identical procedures were performed using yeast total RNA. The DNA sequence of the amplified product revealed a single open reading frame contiguous with ORF1 beginning with an AUG consensus start codon 32 codons upstream from the initial glutamate codon of clone 13 (amino acid 33 in Fig. 1A), followed by a typical signal sequence (42), and a signal cleavage site for the yeast KEX2-encoded endoprotease (43) that is located within the partial clone 13 cDNA after amino acid 39 (arginine) (Fig. 1A). A stop codon existed 8 codons upstream of the putative AUG initiation codon. Stop codons in the other reading frames ruled out the possibility that ORF2 of clone 13 was translated. A composite sequence of the rapid amplification of cDNA ends product and clone 13 cDNA is shown in Fig. 1A. In addition, the amino acid composition of chromatographically purified recombinant hwp1 (from clone 13) was consistent with that predicted by ORF1 (Table I).
The original mRNA probably bound to the oligo(dT) column by a run of five adenosines near the 3Ј end that are within the open reading frames.
The deduced amino acid sequence of the partial cDNA revealed a series of tandem repeats 10 amino acids in length that comprised the majority of the open reading frame. An alignment of the repeats starting with amino acid 14 is shown in Fig.  1B. Common features included proline residues at positions 2, 6, and 9 in all but two of the repeats. Cysteine residues pre-dominated at position 3 and aspartate was found at position 4. The first six repeats had a tyrosine in the fifth position and glutamate in the tenth position, whereas the more carboxy proximal repeats had asparagine and aspartate, at the fifth and tenth positions, respectively. The amino acid composition of the entire ORF was notable in having 27% proline, 16% glutamine, and 12% aspartate residues (mole percents). The C terminus of hwp1 was threonine-and serine-rich providing abundant potential sites for O-glycosylation. The deduced amino acid composition was hydrophilic throughout the entire sequence with a net negative charge of 32 at neutral pH and a calculated M r of 22,750. No consensus sequence N-glycosylation sites were found. Chou-Fasman predictions of secondary structure (44) showed the sequence to be primarily composed of turns with two helical sections in locations containing five and six consecutive glutamine residues. The surface probabilities (45) were also highest for the sites with consecutive glutamine residues and for glutamines bounded by prolines. Searches of computer data bases showed that hwp1 was similar to a wide variety of proteins containing amino acid repeats having periodic proline residues and/or glutamine residues; however, no identical sequences were found.
Northern Blot Analysis-To determine if the morphologyspecific surface expression of hwp1 was related to HWP1 mRNA levels, Northern blotting of RNA from C. albicans grown either in yeast or hyphal forms in Lee's medium or M199 was performed. The presence of HWP1 mRNA depended on production of hyphae and was not an isolated effect of temperature or pH of the growth medium. HWP1 mRNA was present in cells grown at 37°C in both in Lee's medium at neutral pH and in M199, conditions that induced hyphae formation (Fig. 2). HWP1 mRNA was not produced in any of the Lee's medium conditions supporting yeast growth without hyphae production. Enolase mRNA, which is known to be present in both growth forms (25), served as an internal control for the presence of mRNA in all samples. The size of HWP1 mRNA was 2.3 kilobases, much larger than the size of the cDNA clone. Thus the cDNA was not complete but encoded an amino acid segment of the HWP1 gene.

Production of Recombinant hwp1 and Confirmation That hwp1 Is Localized to Hyphal Surfaces of C. albicans-HWP1
cDNA was cloned into the P. pastoris vector, pPIC9 in frame with the S. cerevisiae ␣-agglutinin signal peptide (pPIC13). The resulting plasmid PIC13 was transformed into P. pastoris strain GS115 (his4). Culture supernatants from four clones produced a protein that was not present in untransformed cells. Whereas the anti-C. albicans serum recognized several proteins in the P. pastoris supernatants, the hyphae-specific serum recognized only the protein that was specific to the transformed cells. One clone, GS115-13, that produced rela-  tively large amounts of rhwp1, was chosen for further study.
The finding that antiserum that had been raised to whole C. albicans organisms bearing germ tubes and adsorbed to yeast forms also recognized Pichia-produced rhwp1 of transformant GS115-13 (Fig. 3) provided strong confirmatory evidence that a hyphal surface antigen cDNA had been cloned. The M r of the rhwp1 was determined to be 55,000, much larger than the calculated 27,215 (see "Discussion"). The position of rhwp1 at the anode following isoelectrofocusing was consistent with the calculated pI of 3.39 (data not shown). Coomassie Blue did not stain the recombinant protein in SDS-PAGE. Preimmune serum was negative.
A further prediction of the conclusion that HWP1 cDNA encodes a hyphal surface protein is that antiserum raised to a recombinant protein encoded by HWP1 cDNA should crossreact with hyphal surfaces. Serum from rabbits immunized with rhwp1 was tested for the ability to recognize antigens on C. albicans hyphal surfaces in immunofluorescence assays (Fig. 4). Immune serum, at dilutions of 1:200, which was determined to be optimal by titration experiments, stained hyphal surfaces but not the parent blastoconidia of C. albicans. Thus the antiserum was specific for rhwp1 and hyphal surfaces of C. albicans. In addition, the hyphae-specific staining could be blocked by incubation of chromatographically purified rhwp1 with anti-rhwp1 antiserum used in immunofluorescence assays, and blocking was concentration-dependent (Fig. 4, C  and D). Control incubations with higher concentrations of irrelevant proteins (enolase and lactate dehydrogenase) did not affect antibody detection of native hwp1 (Fig. 4E). The blocking experiments showed that native hwp1 on hyphal surfaces and rhwp1 shared the same antigenic epitopes. Blastoconidia in logarithmic growth at temperatures of 25°C or 37°C were also negative for hwp1 (not shown). These results confirmed that the partial HWP1 cDNA encoded an antigen that is exposed on hyphal surfaces and not on yeast surfaces of C. albicans and that the rhwp1 antiserum was specific for a hyphal antigen encoded in part by the HWP1 cDNA clone.
To determine if hwp1 was the sole antigen recognized by the screening hyphae-specific antiserum, the blocking experiments were performed with this antiserum as well. Recombinant hwp1 did not block the fluorescence exhibited by the hyphaespecific antiserum, indicating that antigens other than hwp1 are present on hyphal surfaces (data not shown).
To ensure that hwp1 expression was not limited to strain SC5314, five additional strains including B311, ATCC 28367, ATCC 38696, and 15 recent clinical isolates were tested for the presence of hwp1 on hyphae induced by growth in M199 at 37°C (not shown). All C. albicans isolates tested expressed hwp1 on hyphal surfaces. In contrast, four strains of Candida tropicalis did not produce hwp1. These results demonstrated FIG. 3. Western blot and immunodetection of rhwp1 produced in Pichia pastoris. Samples of P. pastoris GS115-13 culture supernatant were subjected to SDS-PAGE, and the separated proteins transferred to an Immobilon-P membrane (Millipore Corporation, Bedford, MA). The membrane was cut and each portion was treated with anti-C. albicans hyphae-specific polyclonal antiserum (I) or preimmune serum (P) followed by goat anti-rabbit IgG conjugated to horseradish peroxidase. The membranes were developed with ECL reagents (Amersham Corp.) and exposed to x-ray film (Eastman Kodak's X-OMAT). Protein mass standards (Rainbow Markers, Amersham Corp.) are shown at left.

FIG. 4. Detection of hwp1 on hyphal surfaces of C. albicans grown in laboratory cultures or in host tissues.
Formalin-fixed C. albicans yeasts bearing germ tubes (A-E) were treated with various primary antisera followed by fluoresceinated goat anti rabbit IgG in indirect immunofluorescence assays. A, preimmune serum. B, monospecific antiserum to rhwp1. C and D, monospecific antiserum to rhwp1 incubated with rhwp1 at concentrations of 8.5 g/ml (C) or 4.25 g/ml (D) or with enolase 1.25 mg/ml (E). F-J, sections from an 11-month-old female bg/bg-nu/ϩ mouse that had been colonized for 9 months with C. albicans were treated with various antisera (G-J) or histologically stained with the periodic acid-Schiff reagent (F). G, preimmune serum. H, anti-C. albicans serum I, preimmune serum. J, monospecific antiserum to rhwp1. The large arrowhead points to positive-staining hyphae, and the small arrowhead points to negative-staining yeast. The bars represent 5 m. that hwp1 is not specific to SC5314 but is a feature of C. albicans strains in general. Although hwp1 was not detected on four strains of C. tropicalis, more experiments are needed to determine if hwp1 is present in other species of Candida.
To gain further information about native hwp1, Western blots of cell wall digests using monospecific antiserum from rhwp1 were performed. The Western blots showed a polydisperse pattern of antibody binding that was concentrated at a M r of approximately 34,000 (Fig. 5A, arrow). Because the enzyme used to prepare cell wall digests, Zymolyase, contains a protease and because cell wall proteins are highly processed, the M r of the material detected on the Western blot does not represent that of the intact native protein. The same pattern of antibody binding was seen in antisera from two rabbits that and been immunized with rhwp1 (Fig. 5A, lanes 4 and 8).
Although the pattern of background bands differed, the polydisperse pattern was similar for both sera and was absent in preimmune sera. The specificity for hwp1 was demonstrated by the ability of rhwp1 to block antibody binding, whereas higher concentrations of an irrelevant protein, enolase, had no effect (Fig. 5A, lanes 5 and 9 versus lanes 6, 7, 10, and 11). Native hwp1 was not found in yeast cell wall digests (Fig. 5B), consistent with the developmental expression seen by immunofluorescence and by Northern blotting. Polydisperse patterns of binding are typical of glycoproteins and may reflect O-glycosy-lation of hwp1.
Immunofluorescence of Colonized Tissue-To determine if hwp1 was produced by C. albicans growing in mammalian hosts, immunofluorescence assays were performed on paraffinembedded tissues from beige mice that were heavily colonized with C. albicans. Large numbers of organisms were seen in the lumen and keratinized superficial layers of the stomach following staining with the periodic acid-Schiff stain (Fig. 4F) or when indirect immunofluorescence using polyvalent rabbit antiserum to C. albicans was performed (Fig. 4H). Monospecific antiserum to rhwp1 stained the filamentous but not yeast forms of C. albicans in the tissue sections (Fig. 4J), indicating that hwp1 is specific to hyphal forms during growth in the host as well as during growth in laboratory medium. DISCUSSION Our use of recombinant techniques to identify surface antigens of C. albicans hyphae has led to the discovery of a novel amino acid segment that is both developmentally regulated and expressed on hyphal surfaces. Given the high antigen density required for visualization by immunofluorescence assays, the positive immunofluorescence of hyphae when tested with monospecific antiserum to rhwp1 indicates that hwp1 is an abundant antigen. The same monospecific antiserum was also able to detect "native" hwp1 in Western blots of hyphal walls digested with Zymolyase. Coupled with the strong signal intensity on Northern blots probed with HWP1 cDNA, the results suggest that HWP1 encodes a major, immunodominant hyphal surface protein.
Messenger RNA levels of HWP1 were clearly correlated with C. albicans morphology and not with a particular medium component or temperature. This pattern of expression is similar to that found for ECE1 (46) but is in contrast to other genes that have been correlated with hyphal growth (47,48). The presence of HWP1 mRNA correlated with hwp1 protein expression as well. The absence of hwp1 mRNA in yeast forms under three different environmental conditions compared with the abundant message in hyphal forms suggests that putative cisacting gene elements play a role in maintaining transcription control until hyphae-specific signals allow for HWP1 expression. Similar regulatory mechanisms have been demonstrated for gene expression during switching between white and opaque forms of C. albicans (49), although the relationship between expression of genes controlled by switching and genes controlled by the bud-hypha transition is unknown. The presence of hwp1 on hyphal and not yeast surfaces in murine tissues indicated that the developmental regulation of hwp1 is also operative in vivo and thus hwp1 could be an important determinant of pathogenicity.
Several mechanisms have been proposed to contribute to the changes in surface composition of C. albicans during morphogenesis. Ultrastructural studies support the occurrence of rearrangements or losses of wall components during morphogenesis (50,51), whereas unmasking of cryptic antigenic determinants has been suggested using monoclonal antibodies to localize specific antigens (6,52,53). Although we showed in previous work that hyphal surfaces contained unique surface proteins (8,39), the mechanisms contributing to the expression of these proteins were not determined. The present finding that control of expression of hwp1 was mediated by mRNA levels proves that synthesis of new proteins is an important process in controlling hyphal surface gene expression. However, other mechanisms cannot be ruled out and are supported by the changes in carbohydrate structures on hyphal forms compared with yeast forms (10).
The amino acid sequence provided plausible explanations for the unusual physical characteristics of rhwp1. The lack of FIG. 5. Western blot and immunodetection of hwp1 in the hyphal wall of C. albicans. Cell wall proteins were subjected to SDS-PAGE, and the separated proteins were transferred to an Immobilon-P membrane. Immunodetection of hwp1 was performed in an Immunetics manifold (see under "Experimental Procedures"). Rabbit anti-rhwp1 sera (321 and 326) were used alone, preincubated with of rhwp1 (7.08 g/ml, lanes 6 and 10 and 14.2 g/ml, lanes 7 and 11) or preincubated with enolase (31.3 g/ml, lanes 5 and 9) as indicated above each lane. Lanes 1 and 2 were incubated with preimmune sera 321 and 326, respectively. Lane 3 was incubated with rabbit antiserum to whole C. albicans organisms bearing germ tubes; the antiserum was not adsorbed to yeast forms. The membrane was subsequently incubated with goat anti-rabbit horse radish peroxidase conjugate and developed with ECL reagents (Amersham Corp.). A, hyphal wall proteins. B, yeast wall proteins. Protein mass standards (Rainbow Markers, Amersham Corp.) are shown on the left. The arrows indicate the position of the immunodominant hwp1 protein fragment recognized by anti-rhwp1 antiserum that is present in cell wall digests of hyphae but not yeast. agreement between the predicted molecular weight and the SDS-polycrylamide gel-determined molecular weight is typical for proteins with high proline content (54 -57), although Oglycosylation of serine residues located near the C terminus might also contribute to the discrepancy. A feature in common with acidic proline-rich salivary proteins was the poor binding of Coomassie Blue (57). Recombinant hwp1 stained blue with a cationic carbocyanine dye (58) and migrated to the anode following isoelectric focusing (data not shown), features that were correlated with the low isoelectric point of the amino acid sequence.
The deduced amino acid sequence of hwp1 suggested several potential functional properties in addition to antigenicity. First, the high percentage of proline residues and the length of the repeat (10 amino acids) place hwp1 within a varied group of proline-rich proteins in which the proline residues are proposed to function in maintaining the polypeptide chains in extended conformations and to mediate noncovalent interactions between protein chains or, in the case of salivary proteins, to bind toxic plant polyphenols (59). Of particular interest is the presence of acidic salivary proline-rich proteins (aprp) in this group because of the findings by Bradway et al. (60,61) that aprp are substrates for buccal epithelial cell transglutaminase and that an aprp-like, transglutaminase substrate might also be present on surfaces of C. albicans. Like hwp1, aprp contain glutamine residues within proline-rich repeats, a conformational arrangement that may be favorable for formation of ⑀ (␥ glutamyl) lysine cross-links by transglutaminase, given the properties of other known substrates for epithelial cell transglutaminases (62,63). The similarity of hwp1 to aprp could be important for interactions between C. albicans and the oral mucosa. The presence of a polyphenol-binding extracellular proline-rich wall protein of the plant fungal pathogen Colletotrichum graminicola (64) lends support to the idea that the presence of proteins with aprp-like properties might be a common feature of fungal cell walls. A cell wall protein containing abundant proline residues has also been reported from S. cerevisiae (65). The primary sequences of these proteins are unknown.
Other features of the hwp1 repeats are also likely to be functionally important. The presence of a cysteine residue in each repeat probably leads to the formation of regularly spaced extracellular disulfide bonds that may be important for the surface conformation of hwp1 and possibly for intermolecular cross-linking of proteins on the cell surface. Given the specificity of hwp1 for hyphal surfaces, the presence of cysteines may be related to the enhanced sensitivity and increased protein release from hyphal forms following treatment with dithiothreitol (4,66,67). An additional feature of the repeats is the presence of acidic amino acids that would confer a negative charge to hyphal surfaces at physiological pH. The presence of anionic proteins on hyphal surfaces has been demonstrated by others (68). The serine and threonines near the C-terminal end of hwp1 are likely to be sites for O-glycosylation and serve as wall spanning domains as has been proposed for other yeast surface proteins (69 -71). Finally, the presence of amino acid repeats on hyphal surfaces is not surprising given that tandem amino acid repeats are widely distributed on surfaces of broad groups of microorganisms including fungi, parasites, viruses, and bacteria. The functions of repetitive surface proteins have frequently been found to involve the host and include attachment sites to host cells, evasion of phagocytosis, invasion of host cells, and neutralization epitopes (54,(72)(73)(74)(75)(76)(77)(78)(79)(80). Thus it is likely that hwp1 plays a significant role in interaction with the oral mucosa and with components of the host immune system. Future molecular and biochemical characterization of hwp1 will provide insights into the pathogenesis of candidiasis and will be important for developing new strategies for the medical management of candidiasis.