MCP Sign the guestbook
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

Originally published In Press as doi:10.1074/mcp.M500243-MCP200 on September 28, 2005.
This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow All Versions of this Article:
M500243-MCP200v1
5/1/79    most recent
Right arrow Submit a response
Right arrow Alert me when this article is cited
Right arrow Alert me when eLetters are posted
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow Glossary
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Pitarch, A.
Right arrow Articles by Gil, C.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Pitarch, A.
Right arrow Articles by Gil, C.
Social Bookmarking
Add to CiteULike   Add to Complore   Add to Connotea   Add to Del.icio.us   Add to Digg   Add to Reddit   Add to Technorati  
What's this?
Molecular & Cellular Proteomics 5:79-96, 2006.
© 2006 by The American Society for Biochemistry and Molecular Biology, Inc.

Research

Decoding Serological Response to Candida Cell Wall Immunome into Novel Diagnostic, Prognostic, and Therapeutic Candidates for Systemic Candidiasis by Proteomic and Bioinformatic Analyses*

Aida Pitarch{ddagger}, Antonio Jiménez§, César Nombela{ddagger}, and Concha Gil{ddagger},||

From the {ddagger} Department of Microbiology II, Faculty of Pharmacy, Complutense University of Madrid, 28040 Madrid, Spain and the § Department of Internal Medicine II, Salamanca University Hospital, 37007 Salamanca, Spain


   ABSTRACT
TOP
 ABSTRACT
EXPERIMENTAL PROCEDURES
 RESULTS
 DISCUSSION
 REFERENCES
 
In an effort to bring novel diagnostic and prognostic biomarkers or even potential targets for vaccine design for systemic candidiasis (SC) into the open, a systematic proteomic approach coupled with bioinformatic analysis was used to decode the serological response to Candida wall immunome in SC patients. Serum levels of IgG antibodies against Candida wall-associated proteins (proteins secreted from protoplasts in active wall regeneration, separated by two-dimensional gel electrophoresis, and identified by mass spectrometry) were measured in 45 SC patients, 57 non-SC patients, and 61 healthy subjects by Western blotting. Two-way hierarchical clustering and principal component analysis of their serum anti-Candida wall antibody expression patterns discriminated SC patients from controls and highlighted the heterogeneity of their expression profiles. Multivariate logistic regression models demonstrated that high levels of antibodies against glucan 1,3-ß-glucosidase (Bgl2p) and the anti-wall phosphoglycerate kinase antibody seropositivity were the only independent predictors of SC. Receiver operating characteristic curve analysis revealed no difference between their combined evaluation and measurement of anti-Bgl2p antibodies alone. In a logistic regression model adjusted for known prognostic factors for mortality, SC patients with high anti-Bgl2p antibody levels or a positive anti-wall enolase antibody status, which correlated with each other, had a reduced 2-month risk of death. After controlling for each other, only the seropositivity for anti-wall enolase antibodies was an independent predictor of a lower risk of fatality, supporting that these mediated the protective effect. No association between serum anti-cytoplasmic enolase antibody levels and outcomes was established, suggesting a specific mechanism of enolase processing during wall biogenesis. We conclude that serum anti-Bgl2p antibodies are a novel accurate diagnostic biomarker for SC and that, at high levels, they may provide protection by modulating the anti-wall enolase antibody response. Furthermore serum anti-wall enolase antibodies are a new prognostic indicator for SC and confer protection against it. Bgl2p and wall-associated enolase could be valuable candidates for future vaccine development.

Not only does systemic candidiasis (SC)1 continue to be significant in incidence (1–3), but it also remains a leading infectious cause of morbidity and mortality in intensive care, post-surgical, and cancer patients (3–5) and accounts for substantial healthcare costs (6, 7). Clinical outcomes might be improved by early initiation of antifungal therapy. SC diagnosis, however, is extremely difficult because signs and symptoms of invasive disease are nonspecific. In addition, its two gold standards, blood cultures and tissue biopsies, may lack sensitivity in the first stages of infection (8) and become excessively invasive in critically ill patients, respectively. As a result, SC diagnosis is often attained after a long delay or, unfortunately, following autopsy (9, 10).

This clinical setting has prompted the search for novel prompt and accurate disease markers or, instead, for immunoprophylactic strategies. Research efforts currently focus on screening both for Candida cell wall polysaccharides (mannans or ß-1,3-glucans), extracellular proteins (secreted aspartyl proteinase), cytoplasmic proteins (enolase or heat shock protein 90), metabolites (D-arabinitol), or nucleic acids (DNA or RNA) and for anti-Candida antibodies in body fluids of SC patients (11–15). The relevance of well defined Candida cell wall proteins or their related antibodies in SC diagnosis, prognosis, and therapy, however, has been barely examined (16–18). Given their privileged location within the cell (host-fungus interface), it is unsurprising that some of them may be major elicitors of a specific immune response, which could, intriguingly, be brought into play to establish prognosis and develop new diagnostic, prophylactic, and/or therapeutic procedures for SC (16, 18).

For many years, knowledge of the Candida cell wall proteome has, to a certain extent, been limited by a paucity of effective strategies for tackling its study especially because of its low abundance, low solubility, high heterogeneity, hydrophobicity, and interconnections with wall structural polysaccharides (mannan, glucan, and/or chitin) (19–21). Beyond encouraging chemical and/or enzymatic approaches for its extraction from intact cells or isolated walls (22–25) and by-passing their inherent troubles (glucan and/or chitin side-chain residues and protein modifications (19, 26)), an innovative stratagem, based on the analysis of proteins secreted from protoplasts in active cell wall regeneration, has also recently enabled this Achilles’ heel to be successfully profiled (26–28). This modus operandi draws on the observation that these wall biogenesis-related proteins are not retained into the nascent cell wall during the early stages of the regeneration process and are thus shed into the extracellular medium (26, 29).

On the other hand, there is no doubt either that biomarkers discovered in serum are highly sought after in critically ill and/or severely immunocompromised patients in view of the fact that they may serve as a keystone for the design of noninvasive and easy diagnostic tests and/or of therapeutic strategies. Among the currently available techniques, the combination of proteomics with serology (immunoproteomics (30–34)) is largely being used to screen for panels of biomarkers and therapeutic targets in patients with infectious diseases (including SC (27, 31, 35, 36)), cancer, or autoimmune disorders (32,37–41). This discipline permits characterization of the immunome of a (micro)organism, defined as the subset of the proteome of a (micro)organism that acts as a target for the immune system (42). Remarkably its application in SC research has made it possible to cut back further the false-negative rate associated with low antibody sensitivity in immunocompromised patients undergoing SC (36). Because of the huge amounts of data compiled in proteomic investigations, there is now increasing awareness that bioinformatics is certainly an essential tool for their handling and interpretation (43–45).

In the present study, we used both aforementioned proteomic approaches coupled with bioinformatic analyses (Fig. 1) to systematically explore the serological response to Candida cell wall-associated proteins in SC patients. We investigated their usefulness to diagnose SC, predict outcomes in these patients, and even to outline potential therapeutic targets for prevention and/or treatment of SC.



View larger version (44K):
[in this window]
[in a new window]
  		FIG. 1. Overview of the proteomic strategy used to identify serum antibodies against Candida wall-associated proteins as potential clinical biomarkers and/or therapeutic targets for SC. The Candida cell wall is located outside the plasma membrane and basically consists of polysaccharides (ß-glucans, mannans, and chitin) and proteins interconnected through covalent and non-covalent linkages in various ways, resulting in an elevated complexity (17) (top left). The proteome of this intricate cellular compartment may be profiled using an effective approach (27) based on the secretion of cell wall biogenesis-related proteins in the medium during the early stages of the regeneration process of protoplast walls. This stratagem takes advantage of the fact that their covalent incorporation into the nascent cell wall is delayed at that juncture, and consequently they are secreted into the extracellular medium (top right). As shown in the flow chart, these proteins are thereafter separated by two-dimensional gel electrophoresis and then subjected either to silver staining and mass spectrometry for protein identification or to Western blotting with serum specimens from SC and non-SC subjects following by computer-assisted analysis of the densitometric profiles of antibody reactivities. These patterns are subsequently assessed by statistical analyses to translate them into potential diagnostic and/or prognostic biomarkers for SC and/or candidates for therapeutic intervention. (M + 2H)2+, doubly charged precursor ion.

 

   EXPERIMENTAL PROCEDURES
TOP
 ABSTRACT
 EXPERIMENTAL PROCEDURES
RESULTS
 DISCUSSION
 REFERENCES
 
Study Population and Serum Specimens—
Between December 1997 and March 2003, serum specimens from 48 patients with laboratory-confirmed SC belonging to different risk groups were obtained on the day of culture sampling at the Salamanca University Hospital (Spain) after informed consent was given. SC was defined as isolation of the same Candida species in one or more blood cultures and/or in culture from at least three noncontiguous sites from patients who manifested clinical signs of infection or sepsis. Three patients who had received antifungal drug prophylaxis before diagnosing SC were excluded from the study. To provide data on assay specificity, serum samples from 118 individuals without clinical or microbiological evidence of SC and with a similar age and sex distribution to cases were evaluated as controls. This control group consisted of 57 hospitalized patients with the same primary diagnosis as cases and of 61 healthy subjects. All sera were stored at –80 °C and analyzed in a blinded fashion. The outcome of hospital stay (death or discharge) was also recorded for each patient. Base-line characteristics of the study patients and controls are shown in Table I. By design, no substantial differences were found either in the demographic characteristics among the study groups or in the clinical characteristics between SC and non-SC patients. Not surprisingly, SC patients, however, had a 3 times greater mortality rate than non-SC patients (27 versus 9%; p = 0.02).


View this table:
[in this window]
[in a new window]
  		TABLE I Base-line characteristics of the 163 subjects included in the screening study

 
Preparation of Candida Cell Wall-associated Proteins—
Proteins secreted into the culture medium during the early stages of the regeneration process of protoplast walls from a clinical Candida albicans isolate (strain SC5314 (46)) were exploited as a source of Candida wall-associated antigens and prepared basically as reported previously (27). Briefly yeast cells were grown in YED medium (1% Difco yeast extract and 2% D-glucose) (47) and incubated at 28 °C first in a pretreatment solution (10 mM Tris-HCl, pH 9.0, 5 mM EDTA, 1% 2-mercaptoethanol) for 30 min and then in a solution containing 30 µg/ml Glusulase (DuPont) and 1 M sorbitol (up to 5 x 108 cells/ml) until over 90% protoplasts were obtained. After three washes with 1 M sorbitol, protoplasts were induced to regenerate their cell walls in Lee medium (48) containing 1 M sorbitol at 28 °C for 2 h. Following centrifugation, protease inhibitors were added to the supernatant. This was filtered and ultrafiltered using a pore size of 0.22 µm and 10,000 Da, respectively, and then diluted with water, concentrated a further three times, and eventually lyophilized. All steps were performed with very gentle shaking. Cell lysis was controlled by quantitative determination of alkaline phosphatase (49). Protein concentration was measured with the Bradford assay (Bio-Rad).

Two-dimensional Polyacrylamide Gel Electrophoresis (2-DE)—
Candida wall-associated proteins were separated by 2-DE as described elsewhere (22) using immobilized, nonlinear pH 3–10 gradient strips (18 cm; Amersham Biosciences) for isoelectric focusing and 10% SDS-polyacrylamide gels (10% T, 1.6% C) for the second dimension separation. Thereafter proteins were either visualized with silver nitrate (22, 50) or electrotransferred to nitrocellulose membranes (51). Mr and pI values were assigned after calibration of gels with internal 2-DE standards (Bio-Rad) using Melanie 3.0 software (GeneBio, Geneva, Switzerland).

Lectin Blotting—
After electroblotting, 2-D blots were blocked with 3% BSA in TBS for 2 h and then rinsed with TBS. They were subsequently incubated with peroxidase-labeled concanavalin A (Sigma) in blocking solution for 1 h and washed again. Concanavalin A (ConA)-binding proteins were detected using an enhanced chemiluminescence reagent (ECL, Amersham Biosciences) and high performance films (Hyperfilm ECL, Amersham Biosciences).

Western Blot Analysis—
Western blotting was performed essentially as described previously (36). Serum anti-Candida wall IgG antibody levels were indirectly measured in each tested specimen by computer-assisted densitometric analysis of the corresponding Western blotting results using Melanie 3.0 software and expressed as arbitrary units relative to the integrated optical density of their related spot area after background subtraction and normalization to the loading control (i.e. both to membranes reproved with polyclonal antibodies against glyceraldehyde-3-phosphate dehydrogenase and to silver-stained gels run in parallel).

In-gel Digestion and Mass Spectrometry Analysis—
Immunoreactive protein spots of interest were excised from silver-stained preparative 2-DE gels, in-gel destained, reduced, alkylated, and digested with trypsin as reported previously (22). The resulting peptides were analyzed using a MALDI-TOF mass spectrometer (Voyager-DE STR, PerSeptive Biosystems, Framingham, MA) (22). The peptide mass fingerprinting data were used to search for protein candidates in the Swiss-Prot database (www.expasy.ch/sprot). Proteins ambiguously identified using this procedure were subjected to nanoelectrospray ionization tandem quadrupole-TOF mass spectrometry analysis (nano-ESI Q-TOF MS/MS; Micromass, Manchester, UK) to obtain peptide sequences for database search as described elsewhere (28, 35). The identified proteins and their related 2-DE map are currently available on our electronic COMPLUYEAST-2DPAGE database, which can be found on the ExPASy server (www.expasy.ch/ch2d/2d-index.html) or at the URL address babbage.csc.ucm.es/2d/2d.html (19, 52).

Statistical Analysis—
Data distribution was assessed by the Kolmogorov-Smirnov test. Categorical variables were compared using the {chi}2 test with Yates’ correction or Fisher’s exact test as appropriate. Distributions of continuous data were analyzed by the Kruskal-Wallis test or Mann-Whitney test for multiple or pairwise group comparisons, respectively. Correlations were determined by the Spearman’s rank correlation coefficient, and linear regression was calculated using the least squares method. Associations of antibody levels with base-line variables were estimated using simple and multiple linear regression analyses. Distances among antibody responses were measured by one-way hierarchical cluster analysis.

Two-way (antibodies against specimens) hierarchical clustering and principal component analysis were applied to group the samples of the study population based on the similarity of their anti-Candida wall antibody expression profiles (53, 54). Anti-Candida wall IgG antibody levels were normalized by median centering antibodies for each specimen and then by median centering each antibody across all specimens. The homology and homogeneity of these patterns among groups were evaluated on principal component analysis data by the Mann-Whitney test and analysis of variance, respectively.

Multivariate logistic regression models containing variables significantly associated either with the risk of SC or with the outcome of SC patients in univariate analyses were developed to identify independent predictors of SC or death, respectively, after controlling for potential confounding factors. Antibody levels entered in the models were dichotomized as detectable versus undetectable and as high versus low or medium on the basis of the 95th percentile of the distribution of values in the control group. Model calibration was estimated by the Hosmer-Lemeshow goodness-of-fit test, and model discrimination was examined using the area under receiver operating characteristic (ROC) curve. A nonparametric asymptotic method was used to compare ROC areas (55). Differences in sensitivity, specificity, and accuracy between univariate and multivariate predictors of SC were assessed with the McNemar’s test. The exact 95% confidence intervals (CIs) for the test operating characteristics of selected indicators were computed on the basis of a binomial approximation to the normal distribution. Statistical significance was set at p < 0.05 (two-sided). A flow chart of the experimental procedure followed is depicted in Fig. 1.


   RESULTS
TOP
 ABSTRACT
 EXPERIMENTAL PROCEDURES
 RESULTS
DISCUSSION
 REFERENCES
 
Seroprevalence of IgG Antibodies against Candida Cell Wall-associated Proteins in SC Patients—
Serum samples from SC patients (n = 45), non-SC patients (n = 57), and healthy subjects (n = 61) were screened individually by Western blotting for circulating IgG antibodies against 2-DE-separated Candida wall-associated proteins, proteins secreted from protoplasts in active wall regeneration (Fig. 1). Immunoreactive protein spots showing significantly different recognition by SC patients’ sera compared with specimens from either of the control groups (p < 0.05) were characterized by mass spectrometry (Table II). In all, seven wall-associated proteins were identified of which five displayed series of seroreactive spots on 2-D patterns and one (four rod-like 33–37-kDa spots characterized as Bgl2p) was further stained with ConA-peroxidase treatment (data not shown).


View this table:
[in this window]
[in a new window]
  		TABLE II Seroprevalence of anti-Candida wall IgG antibodies in the study population

 
The median number of immunogenic proteins detected on each blot was significantly higher with SC patients’ sera (4; interquartile range, 2–5) than with samples from non-SC patients (2; interquartile range, 1–4; p < 0.001) or healthy individuals (1; interquartile range, 1–2; p < 0.001), but it did not differ between specimens from both control groups (p = 0.1). Interestingly all SC patients were seropositive for anti-Bgl2p IgG antibodies albeit with a mere 36% (95% CI, 27–44%; p < 0.001) higher frequency rate than controls (Table II). Although the anti-BgI2p IgG antibody seroprevalence had a sensitivity (the likelihood of correctly identifying SC patients) of 100%, its positive predictive value (the likelihood that a person with this positive test has SC) was only 37%. Likewise SC patients had a 40% (95% CI, 24–56%; p < 0.001), a 32% (95% CI, 16–47%; p < 0.001), and a 26% (95% CI, 10–41%; p = 0.002) greater prevalence of seropositive anti-Pgk1p, anti-Met6p, and anti-Eno1p IgG antibodies, respectively, than the control group.

Combinations of the seven types of identified anti-Candida wall IgG antibodies were often observed (Table II). Remarkably 31% (14 of 45) of SC patients’ serum specimens were simultaneously positive for IgG antibodies against Bgl2p, Eno1p, Pgk1p, and Met6p, whereas only 3% (4 of 118) of control samples displayed this recognition pattern (p < 0.001).

Circulating Levels of Anti-Candida Wall IgG Antibodies in SC Patients—
The estimated coefficients of variation of the recognition intensities generated for this set of antigens ranged from 5.3 to 10.5%, suggesting a good reproducibility of our assay. Overall median serum levels of the defined anti-Candida surface antibodies proved to be consistently lower in either of the control groups than in the SC group (Fig. 2) except for anti-Tpi1p antibodies, which, similar to their seroprevalence, did not differ between SC and non-SC patients (p = 0.3). No statistical evidence of variation in serum levels of this panel of antibodies was established between the two control groups, emphasizing that they do not increase as a result of underlying patient pathology.



View larger version (40K):
[in this window]
[in a new window]
  		FIG. 2. Distribution of serum levels of IgG antibodies against Candida wall-associated proteins among the study groups. The central panel depicts a silver-stained 2-DE map of the Candida cell wall proteome corresponding to that obtained from proteins secreted from protoplasts in active cell wall regeneration (see Fig. 1 for further details). Labeled spots represent the Candida wall-associated antigenic proteins (cell wall immunome) that were detected with serum specimens from the study population, some of which display several protein species. Protein names refer to those in Table II. The remaining panels show box-and-whisker plots of serum levels of their related IgG antibodies in the 163 subjects included in the screening study. Their distribution differed significantly between SC patients and controls except for anti-Tpi1p IgG antibody levels from SC and non-SC patients (p = 0.3). The boxes indicate the interquartile ranges (25–75th percentiles), the horizontal thick lines denote the medians, the black squares portray the means, the whiskers extend to 1.5 times the interquartile range, and the circles illustrate the outliers. The scales on the y axes in the left middle panel (that of anti-Bgl2p antibodies) and the rest of panels differ. AU, Western blotting arbitrary units.

 
A positive correlation was found between serum concentrations of any of these antibodies and the number of antigens immunodetected (p < 0.001; Fig. 3A). In the SC group, circulating levels of anti-Eno1p IgG antibodies were associated with those of IgG antibodies against Pgk1p (r = 0.59; p < 0.001) and Met6p (r = 0.65; p < 0.001; Fig. 3B). In turn, the recognition intensities for Pgk1p also correlated with those for Met6p (r = 0.54; p < 0.001) and Gap1p (r = 0.31; p = 0.04). Clustering of the characterized anti-Candida envelope antibodies resulted in a hierarchical tree that led to assessment of the proximity among them in terms of immune response to SC (Fig. 3C). As expected, IgG antibodies against Eno1p, Pgk1p, Met6p, and Gap1p were clustered together, forming two subgroups, whereas those against Bgl2p or Tpi1p were not associated with specific branches.



View larger version (19K):
[in this window]
[in a new window]
  		FIG. 3. Relations among serum levels of anti-Candida wall IgG antibodies. A, correlation between serum levels of anti-Candida wall IgG antibodies (anti-Pgk1p IgG antibodies in the illustration) and recognition frequency of Candida wall-associated antigenic proteins in the study population. The linear regression line was established by the least squares method. The scale on the y axis is logarithmic. B, three-dimensional scatter plot showing serum anti-Eno1p IgG antibody levels versus serum anti-Pgk1p and anti-Met6p IgG antibody levels in SC patients. The regression equation, defined by a linear hyperplane in the multidimensional space, was determined by multiple linear regression analysis. The scales are logarithmic. C, hierarchical clustering tree of the serological biomarkers tested in SC patients. Serum IgG antibodies with similar expression patterns in SC patients were joined close to each other and by short branches (marked with dashed circles) in the resulting dendrogram, whereas longer branches clustered those with lower correlation. The antibody response to the mannoprotein Bgl2p in SC patients was placed far away from that to the stress-induced proteins (61–63) Met6p, Gap1p, and Fba1p (subgroup "a") and Eno1p and Pgk1p (subgroup "b"). Protein names refer to those in Table II. AU, Western blotting arbitrary units.

 
Multiple linear regression models revealed that the anti-Candida wall IgG antibody levels were not related to age, sex, or coexisting disease. In SC patients subjected to chemotherapy, median serum levels of IgG antibodies against Eno1p (p = 0.01) or Met6p (p = 0.02) as well as the median number of antigens immunorecognized with their sera (p = 0.02) were slightly but significantly lower than in those who were not, but these differences did not reach statistical significance among non-SC patients (Fig. 4).



View larger version (24K):
[in this window]
[in a new window]
  		FIG. 4. Effect of immunosuppressive therapy both on serum levels of IgG antibodies against Candida wall-associated Eno1p (A) and Met6p (B) and on recognition frequency of Candida surface-associated antigenic proteins (C) in SC and non-SC patients. The boxes represent the interquartile ranges (25–75th percentiles), the horizontal thick lines indicate the medians, and the whiskers extend to 1.5 times the interquartile range. Chemotherapy had no significant effect on serum levels of the remaining anti-Candida wall IgG antibodies identified. A and B depict representative 2-D immunorecognition patterns. AU, Western blotting arbitrary units.

 
Profiling of Serum Proteomic Signatures in SC Patients—
To determine whether the antibody response to Candida surface-associated antigens in SC patients forms a characteristic serum proteomic signature, a two-way hierarchical cluster analysis was performed using the serological biomarkers identified differentially in SC and non-SC patients. Intriguingly serum specimens separated into two distinct groups, one mainly containing SC patients and the other predominantly consisting of non-SC patients (Fig. 5A, top), resulting in 87% (95% CI, 80–93%) sensitivity, 77% (95% CI, 69–85%) specificity, and 81% (95% CI, 74–89%) accuracy for SC detection (Table III). Similar results were also evidenced between SC patients and healthy subjects (Fig. 5A, bottom, and Table III). In both analyses, high anti-Bgl2p IgG antibody levels mainly defined the proteomic signature for SC specimens. True to form, samples from the two control groups did not segregate into separate clusters (data not shown).



View larger version (49K):
[in this window]
[in a new window]
  		FIG. 5. Discrimination of serum specimens from the study population according to their anti-Candida wall IgG antibody expression patterns. A, two-way hierarchical cluster analyses of differentially expressed anti-Candida wall IgG antibodies (rows) and serum specimens (columns) from SC and non-SC patients (top) or SC patients and healthy subjects (bottom). Because anti-Tpi1p IgG antibody levels did not differ between SC and non-SC patients (see Fig. 2), they were excluded from the analyses. Red or green rectangles correspond to IgG antibody expression levels above or below, respectively, the median value (black rectangles). As distinguished on both x axis dendrograms, SC patients and controls had largely distinct serum anti-Candida wall IgG antibody expression patterns and thus different immune responses (see Table III for further details on SC prediction results). SC, systemic candidiasis patients; N, non-sc patients; H, healthy subjects. B, principal component analyses of the anti-Candida wall IgG antibody expression profiles in serum specimens from SC and non-SC patients (left), SC patients and healthy subjects (center), or all controls (right) within a three-dimension vector space. The percentages of variance explained using the first three principal components are displayed on the corresponding axes. Each circle denotes the anti-Candida wall IgG antibody expression pattern of a single serum specimen. Samples are color-coded as shown. The color-shaded areas represent clustering of specimens. Asterisks and daggers indicate the degree of homology and homogeneity, respectively, of these antibody patterns among the specified study groups. Protein names refer to those in Table II.

 

View this table:
[in this window]
[in a new window]
  		TABLE III Test operating characteristics of the clinical biomarkers for SC identified in this trial

 
Principal component analyses of serum anti-Candida wall IgG antibody expression profiles also discriminated between SC and non-SC individuals (p < 0.001) and highlighted their heterogeneity among these subjects (p < 0.001; Fig. 5B, left and center). The first three principal components explained 82–86% of variability in their antibody expression profiles and yielded distinct clusters for each of the study groups, close to those acquired from two-way hierarchical clustering. In contrast, the counterparts in both control groups, accounting for 88% of variance, gave rise to no successful separation of the samples (Fig. 5B, right). Their antibody expression profiles were similar (p = 0.5) and homogeneous (p = 0.8).

Independent Predictors of SC—
The ability of these serological biomarkers to predict the risk of SC either individually or in combination, before Candida cultures became positive, was further investigated. Given that no significant differences were found in anti-Candida wall IgG antibody levels between non-SC patients and healthy subjects (Figs. 2 and 5), pooled data were used. In univariate analyses, an elevated anti-Bgl2p IgG antibody concentration was the strongest predictor of SC (odds ratio (OR), 16.5; 95% CI, 5.64–48.31; p < 0.001; Fig. 6A). Multivariate logistic regression models showed that a high anti-Bgl2p IgG antibody level (OR, 12.5; 95% CI, 3.96–39.55; p < 0.001) and a positive anti-Pgk1p IgG antibody status (OR, 7.9; 95% CI, 3.05–20.71; p < 0.001) were the only variables that remained independently associated with a significant risk of SC after adjusting for the other base-line predictors (Fig. 6B). Further adjustment for the matching factors of age, sex, and coexisting illness did not substantially affect the multivariate model nor did additional control for the potentially confounding effects of known predisposing factors for SC (listed in Table I). The Hosmer-Lemeshow test demonstrated that the fit of this model was good (p = 0.9).



View larger version (50K):
[in this window]
[in a new window]
  		FIG. 6. Risk of SC according to serum levels of anti-Candida wall IgG antibodies. Antibody levels were dichotomized as positive versus negative and as high (defined as a value above the 95th percentile of the distribution of values in the control group) versus medium or low. The floating odds ratios (symbols) and corresponding 95% floated confidence intervals (horizontal lines) for SC are presented on a logarithmic scale. A, unadjusted odds ratios (squares) for potential predictors of SC. Because no anti-Bgl2p IgG antibody-seronegative patients had SC, the odds ratio for any positive anti-Bgl2p IgG antibody level was undefined. For the analysis of antigen recognition frequency, specimens with low frequencies served as the reference group (p < 0.001 for the trend). B, odds ratios for variables that retained significance in independently predicting the risk of SC after adjustment for all significant variables indicated in A (diamonds). The analysis was further adjusted for matching factors of age, sex, and coexisting disease (triangles) as well as known predisposing factors for SC listed in Table I (circles). The fit of the final model was good (p = 0.9). On the left, in both panels representative expanded sections of 2-D immunoblots obtained using SC patients’ serum samples display the number of protein species detected with their related IgG antibodies. Protein names refer to those in Table II. AU, Western blotting arbitrary units.

 
Based on the final model, the multivariate adjusted odds ratios of rising anti-Bgl2p IgG antibody concentrations combined with the anti-Pgk1p IgG antibody seroprevalence were estimated (Fig. 7A). Although at each of the serum anti-Bgl2p IgG antibody levels, anti-Pgk1p IgG antibody-seropositive subjects had a higher risk of SC than those who were seronegative; these levels were, however, similarly predictive of SC in both groups (p < 0.001 for trend). In fact, the seroprevalence of anti-Pgk1p IgG antibodies had no considerable effect on the serum anti-Bgl2p IgG antibody concentrations (Fig. 7B).



View larger version (26K):
[in this window]
[in a new window]
  		FIG. 7. Assessment of the interaction between the two independent predictors of SC determined by multivariate logistic regression analysis, and of their SC diagnostic performance. A, odds ratios for SC according to anti-Bgl2p IgG antibody level and anti-Pgk1p IgG antibody status in serum adjusted for other significant anti-Candida wall IgG antibody levels in univariate analysis, age, sex, coexisting illness, and known predisposing factors for SC (listed in Table I). Anti-Bgl2p IgG antibody levels were categorized as low (below the 75th percentile), medium (ranging from the 76th to 95th percentile), or high (above the 95th percentile) relative to their values in the control group. p values are for the comparison with the values in individuals who were seronegative for anti-Pgk1p IgG antibodies and who had low anti-Bgl2p IgG antibody levels (the reference group; OR = 1.0). The increased risk of SC associated with the seropositivity for anti-Pgk1p IgG antibodies was observed at all anti-Bgl2p IgG antibody levels (p < 0.001 for the trend). B, dot diagram showing the distribution of serum anti-Bgl2p IgG antibody levels on the basis of the anti-Pgk1p IgG antibody seroprevalence in SC patients and controls. C, receiver operating characteristic curves for the serum anti-Bgl2p IgG antibody levels (Univariate model) and in conjunction with the anti-Pgk1p IgG antibody seroprevalence (Multivariate model) in 45 SC patients and 118 controls. No significant difference was found between the areas under their curves (p = 0.4). The combined evaluation did not prove, therefore, to be better as a SC diagnostic indicator than the measurement of anti-Bgl2p IgG antibody levels alone. The optimum cutoff values for both models are labeled (see Table III). An asterisk indicates a serum level ≥1927 Western blotting arbitrary units of anti-Bgl2p IgG antibodies in conjunction with seronegativity for anti-Pgk1p IgG antibodies or a serum level >150 Western blotting arbitrary units of anti-Bgl2p IgG antibodies together with anti-Pgk1p IgG antibody seropositivity. In A and C, representative 2-D immunorecognition patterns are illustrated. Protein names refer to those in Table II. AU, Western blotting arbitrary units.

 
Both the univariate model containing anti-Bgl2p IgG antibody levels and the multivariate model based on the two aforementioned independent predictors were able to discriminate properly between SC and non-SC individuals (mean ± S.E. ROC area, 0.90 ± 0.026 versus 0.91 ± 0.024; 95% CI, 0.85–0.95 versus 0.86–0.96, respectively; Fig. 7C). However, the slight increase found in area under the ROC curve was not statistically significant (p = 0.4) nor was the difference in diagnostic accuracy between both models (p = 0.5; Table III), suggesting that the predictive model based on those two items did not contribute to a better discrimination beyond the anti-Bgl2p antibody level-based model. The optimal anti-Bgl2p IgG antibody threshold (1900 arbitrary units) derived from the ROC curve led to 78% (95% CI, 71–84%) sensitivity, 93% (95% CI, 89–97%) specificity, and 89% (95% CI, 84–94%) accuracy for identifying SC patients (Fig. 7C and Table III). Thereby assessing the serum levels of this diagnostic biomarker in lieu of its seroprevalence reduced the sensitivity of the measures (from 100 to 78%) but considerably improved their specificity (from 36 to 93%; p < 0.001).

Predictive Factors of 2-month Mortality in SC Patients—
No substantial associations between anti-Candida wall IgG antibody levels and outcomes were found within the non-SC patient group. On the contrary, SC patients with a high anti-Bgl2p IgG antibody level (OR, 0.08; 95% CI, 0.009–0.79; p = 0.03) or a positive anti-Eno1p IgG antibody status (OR, 0.007; 95% CI, 0.001–0.086; p < 0.001) in blood at the time of culture collection had a decreased risk of death in the ensuing 2-month period after controlling other prognostic factors for mortality (Fig. 8A). Although elevated anti-Bgl2p IgG antibody concentrations were significantly related both to a lower risk of fatality and to the anti-Eno1p IgG antibody seropositivity within the SC group in univariate analysis (p < 0.02; Fig. 8B), their association with outcomes, however, was no longer significant once the anti-Eno1p IgG antibody seroprevalence was forced into the model (OR, 0.18; 95% CI, 0.01–3.42; p = 0.3). Indeed the seropositivity for anti-Eno1p IgG antibodies was the only strong and independent predictor of a reduced 2-month risk of an adverse outcome (OR, 0.009; 95% CI, 0.001–0.13; p < 0.001).



View larger version (34K):
[in this window]
[in a new window]
  		FIG. 8. Evaluation of serum levels of anti-Candida wall IgG antibodies as predictors of 2-month mortality in SC patients. A, 2-month risk of death in SC patients according to the serum anti-Candida wall IgG antibody levels that approached statistical significance in univariate models. Antibody levels were dichotomized as positive versus negative and as high (defined as a value above the 95th percentile of the distribution of values in the control group) versus medium or low. The squares represent unadjusted odds ratios. The circles indicate odds ratios adjusted for other potential prognostic factors for mortality (age up to 65 years, malignancy, neutropenia, chemotherapy, surgical procedure, renal failure, and intensive care unit stay). The diamonds correspond to odds ratios further adjusted for each other (anti-Eno1p IgG antibody seroprevalence or anti-Bgl2p IgG antibody levels). Horizontal lines denote 95% confidence intervals, which were wide because of the relative small number of SC patients. B, scatter plot representing the distribution of serum anti-Bgl2p IgG antibody levels in relation to the seroprevalence of anti-Eno1p IgG antibodies and SC patient outcome. Horizontal lines represent medians, and vertical lines correspond to interquartile ranges. The dashed line indicates the 95th percentile of anti-Bgl2p IgG antibody levels in the control group and separates high levels from medium or low levels (shaded rectangle) of this serological biomarker. C, receiver operating characteristic curves of serum anti-Eno1p IgG antibody levels in 45 SC patients for an unfavorable (thick line) or favorable (thin line) outcome in the ensuing 2-month period. An undetectable level of anti-Eno1p IgG antibodies (0 Western blotting arbitrary units) had the highest combined sensitivity and specificity for the prediction of an adverse outcome within 2 months (see Table III). In A and C, representative 2-D immunorecognition patterns are illustrated. Protein names refer to those in Table II. AU, Western blotting arbitrary units.

 
This final model discriminated accurately SC patients who died from those who survived (mean ± S.E. ROC area, 0.88 ± 0.07; 95% CI, 0.73–1.03; p < 0.001; Fig. 8C). ROC curve analysis further indicated that 91% (95% CI, 82–99%) of SC patients with undetectable levels of IgG antibodies against Eno1p (the wall-associated form of Eno1p) died, whereas 94% (95% CI, 87–100%) of those with any positive signal survived (Table III). It is noteworthy that using the same batch of serum specimens as that used in this trial, none of the SC patients, including those with a fatal outcome, was seronegative for IgG antibodies against the cytoplasmic form of Eno1p according to analogous 2-D Western blot assays nor was there any significant association between their circulating levels and clinical outcomes.2


   DISCUSSION
TOP
 ABSTRACT
 EXPERIMENTAL PROCEDURES
 RESULTS
 DISCUSSION
REFERENCES
 
The Serological Response to the Candida Cell Wall Immunome in SC Patients Forms a Distinct Proteomic Signature—
We found that SC was associated with a typical serum anti-Candida wall IgG antibody expression pattern, specially defined by high anti-Bgl2p antibody levels, which differed significantly from that (issued from the commensal nature of Candida) of non-SC individuals and expressed a high degree of heterogeneity among SC and non-SC subjects. This uneven distribution of anti-Candida wall IgG antibody levels between SC and non-SC individuals, revealed by two-way hierarchical clustering and principal component analysis, upholds the assertion that distinct antibody responses to the Candida cell wall immunome lie behind Candida pathogenic and commensal behavior or, more exactly, behind invasive disease and colonization. These findings are not only in line with recent investigations showing the value of gene or protein expression profiling of biological samples to identify diagnostic or prognostic (bio)markers for cancers (56–59) but also extend this potential and the gene expression or proteomic signature concept (56, 57) to the proteomic study of serological response to infectious agents as well as to data derived from 2-DE analyses.

Our results confirm a previous report (60) of a close relationship between immunosuppression and low or undetectable serum anti-Candida antibody levels in SC patients. This may come from a reduced, delayed, or absent antibody response, which is archetypal of patients with an impaired immune system.

Serum Anti-Bgl2p IgG Antibodies Are a Novel Independent Diagnostic Biomarker for SC—
Elevated serum anti-Bgl2p IgG antibody levels appeared to increase the risk of SC by no confounding factors and, through the immune response other than that directed against Pgk1p, Eno1p, Met6p, Gap1p, and Fba1p, stress-induced proteins (61–65). The correlation observed among circulating levels of antibodies against these stress-related proteins, also noted along the course of SC (36, 66), may reflect the stressful environment resulting from the Candida invasive process (64, 67). Plausibly this might stimulate the overexpression of these proteins and facilitate their concurrent visibility on the Candida surface and targeting by immune system cells during SC in a distinct and unrelated way to that directed against Bgl2p, a cell wall remodeling-involved mannoprotein (17, 68). Accordingly both types of antibody responses could display separate effects to SC. This may explain our observation that the risk of serum anti-Bgl2p IgG antibody level-associated SC was unaltered with the concomitant anti-Pgk1p IgG antibody seropositivity and was thus independent and not a consequence of this additive contribution. Still the reason for the synergistic effect of their combination on SC risk remains unknown.

The strong immune response generated against Bgl2p may be attributable, at least in part, to its great abundance (22, 27), loose association with cell wall (22, 68,69), and secretory nature (27, 70,71) because it could, therefore, be shed easily into the bloodstream and processed early by immunocompetent cells resulting in potent immune responses. Although no antigenic properties for this glucan 1,3-ß-glucosidase (Bgl2p) have been described so far in any Candida species, consistent evidence that antibodies against other fungal cell wall-bound and extracellular ß-glucosidases, like Histoplasma capsulatum H antigen (Hag1p), Coccidioides immitis tube-precipitin antigen (TP/Bgl2p), or Paracoccidioides brasiliensis 43-kDa glycoprotein (gp43), are largely used in the serodiagnosis of active histoplasmosis (72), early coccidioidomycosis (73), or paracoccidioidomycosis (72, 74), respectively, provides additional support for our finding that serum anti-Candida Bgl2p IgG antibodies may likewise be useful and complement Candida blood cultures for the early identification (seeing that these require several days to become positive (8)) for SC in high risk patients.

As gathered from its 2-D patterns and ConA binding, this immunodominant antigen was found to be a glycoprotein (in keeping with previous reports (17, 69,75)) with {alpha}-D-mannose and/or {alpha}-D-glucose residues. Remarkably the corresponding glycan moieties of the aforementioned fungal ß-glucosidases seem to retain antigenicity (76, 77) and/or play an essential role in protein conformation, required for the maintenance of their serological reactivity and thus implying the existence of conformational peptide epitopes (76, 78). In view of these observations, further biochemical research is now warranted to shed light on the relative contributions of the carbohydrate and protein moieties of Candida Bgl2p to its antigenicity and also on whether anti-Candida Bgl2p antibodies cross-react with those ß-glucosidases.

High Anti-Bgl2p IgG Antibody Levels May Provide Indirect Protection against SC—
In addition to their diagnostic implications, high anti-Bgl2p IgG antibody levels in SC patients also correlated with a reduced frequency of fatal outcomes, indicating a protective role. Given that loss of Bgl2p function attenuates Candida virulence (68), it is feasible that Bgl2p-neutralizing antibodies may represent an important defense mechanism against SC and be relevant to SC progression. Interestingly vaccines using the fungal ß-glucosidases H antigen and gp43 have proved to be effective in animal models of pulmonary histoplasmosis (79) and paracoccidioidomycosis (80, 81), respectively. These studies, in conjunction with ours, offer the prospect of Candida Bgl2p also being considered, in future research, as a potential vaccine candidate against SC.

Multivariate logistic regression analyses pointed out that these anti-Bgl2p antibody levels were not an independent predictor of mortality. Our results bring to light the inference that high anti-Bgl2p antibody levels may be indirectly related to protection through their association with the anti-Eno1p IgG antibody seropositivity. Hence their protective effect might be the result of modulating the serological response to Eno1p, which conveys protection against SC. The mechanism by which anti-Bgl2p IgG antibodies modify anti-Eno1p IgG antibody levels and therefore provide indirect protection remains speculative. A negative correlation has been found between Bgl2p and wall-associated Eno1p in fluconazole-treated Candida cells (69). Furthermore BGL2 deletion results in an increased chitin deposition into the wall (68) that in turn has been associated with up-regulation of cell surface stress-related proteins like Eno1p (31, 63). Taken together, it is conceivable that high anti-Bgl2p antibody concentrations may neutralize Bgl2p activity and somehow stimulate the Eno1p overexpression in compensatory response, thus driving an amplified production of protective Eno1p-neutralizing antibodies.

Serum IgG Antibodies against the Candida Wall-associated Form of Enolase Are a New Independent Prognostic Biomarker for SC and Convey Protection against SC—
Our findings further highlight the importance of detecting serum anti-Eno1p IgG antibodies in SC patients for the early detection of those at lower risk of death beyond that provided by known prognostic indicators. Clues as to the prognostic significance of these antibodies may stem from a non-glycolytic property of enolase (Eno1p), a moonlighting protein (82, 83), related to its expression on the cell surface. Intriguingly Candida envelope-associated enolase may interact with human plasminogen and mediate its activation, endowing Candida cells with an extra extracellular proteolytic activity to invade and degrade tissues (19, 84). Therefore, the development of antibodies that neutralize this cell surface plasminogen-binding protein could serve as an effective defense barrier and play a key role in SC pathogenesis. This may to some extent explain why the mortality rate was lower in SC patients who were seropositive for antibodies against wall-associated enolase (plasminogen receptor) and why serum antibodies against its cytoplasmic form (intracellular glycolytic enzyme) exhibited no prognostic relevance.

Consequently our data suggest a differentiation of the immune response to these two forms of enolase during SC. This raises the hypothesis of a specific mechanism of Candida enolase processing, as reported in other eukaryotic enolases (85–87), during cell wall biogenesis and prompts the proposal that the antibodies involved in protection against SC might be directed against specific epitopes or antigenic motifs of the Candida surface-associated form of enolase. Their identification, which is now under way, could provide a rationale for the design of novel immunotherapy- or vaccine-based strategies to prevent and control SC.

On the other hand, it is worth emphasizing that two SC patients who were anti-Eno1p IgG antibody-seropositive died, and one seronegative patient survived. These results are consistent with those of earlier studies in which anti-enolase antibodies were found to be partially protective in experimental SC (88, 89) and supports the notion that other protective mechanisms against SC are also implicated (13–15,90).

Future Challenges—
Our proteomic approach has led to the identification of new biomarkers for early detection of SC and prompt assessment of patient prognosis as well as novel candidates for vaccine design. However, although our findings are promising, they must be validated in future prospective multicenter studies using a larger serial and independent set of serum specimens before their routine implementation in clinical practice. If confirmed, they could therefore have future repercussions not only for diagnostic and therapeutic decision making in high risk populations and to contribute to minimizing the mortality rate or hospitalization period but also for the prevention and treatment of SC. Furthermore the strategies used in this research to search for clinical biomarkers and therapeutic targets for SC (in which 2-DE, a time-consuming and labor-intense procedure, could be replaced by other high throughput proteomic technologies) can hopefully be extended to other specimens such as urine or cerebrospinal fluid and even to other infectious diseases, cancers, or autoimmune disorders.


   ACKNOWLEDGMENTS
 
We are indebted to the patients recruited to this clinical trial for participation in this study. We also thank Dr. W. Blackstock (from Functional Proteomics, University of Sheffield, Sheffield, UK), Dr. M. Ward (from Proteome Sciences, Kings College, London, UK), and M. L. Hernáez and M. D. Gutiérrez (from Genomics and Proteomics Center, Complutense University of Madrid, Madrid, Spain) for technical assistance and Dr. M. L. Gil (from the Department of Microbiology and Ecology, Faculty of Pharmacy, University of Valencia, Valencia, Spain) for providing anti-Gapp antibodies.


  FOOTNOTES
 
Received, August 1, 2005, and in revised form, September 13, 2005.

Published, MCP Papers in Press, September 28, 2005, DOI 10.1074/mcp.M500243-MCP200

1 The abbreviations used are: SC, systemic candidiasis; 2-D, two-dimensional; 2-DE, two-dimensional gel electrophoresis; Bgl2p, glucan 1,3-ß-glucosidase or ß-1,3-glucosyltransferase; CI, confidence interval; ConA, concanavalin A; Eno1p, enolase; Fba1p, fructose-bisphosphate aldolase; Gap1p, glyceraldehyde-3-phosphate dehydrogenase; Met6p, methionine synthase; OR, odds ratio; Pgk1p, phosphoglycerate kinase; ROC, receiver operating characteristic; Tpi1p, triose-phosphate isomerase. Back

2 A. Pitarch, C. Nombela, and C. Gil, unpublished observations. Back

* This work was supported by Project Grants CPGE 1010/2000 from Strategic Groups of Comunidad Autónoma de Madrid and BIO-2003-00030 from Comisión Interministerial de Ciencia y Tecnologí and by grants from the Merck, Sharp & Dohme Special Chair in Genomics and Proteomics and from Fundación Ramón Areces, Spain. Back

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

Director of the Merck Sharp & Dohme Special Chair in Genomics and Proteomics, Spain. Back

|| To whom correspondence should be addressed. Tel.: 34-91-394-1748; Fax: 34-91-394-1745; E-mail: conchagil{at}farm.ucm.es


   REFERENCES
TOP
 ABSTRACT
 EXPERIMENTAL PROCEDURES
 RESULTS
 DISCUSSION
 REFERENCES
 

  1. Pfaller, M. A., Diekema, D. J., Jones, R. N., Sader, H. S., Fluit, A. C., Hollis, R. J., and Messer, S. A. (2001) International surveillance of bloodstream infections due to Candida species: frequency of occurrence and in vitro susceptibilities to fluconazole, ravuconazole, and voriconazole of isolates collected from 1997 through 1999 in the SENTRY antimicrobial surveillance program. J. Clin. Microbiol. 39, 3254 –3259[Abstract/Free Full Text]

  2. Almirante, B., Rodriguez, D., Park, B. J., Cuenca-Estrella, M., Planes, A. M., Almela, M., Mensa, J., Sanchez, F., Ayats, J., Gimenez, M., Saballs, P., Fridkin, S. K., Morgan, J., Rodriguez-Tudela, J. L., Warnock, D. W., Pahissa, A., Barcelona Candidemia Project Study Group (2005) Epidemiology and predictors of mortality in cases of Candida bloodstream infection: results from population-based surveillance, Barcelona, Spain, from 2002 to 2003. J. Clin. Microbiol. 43, 1829 –1835

  3. Eggimann, P., Garbino, J., and Pittet, D. (2003) Epidemiology of Candida species infections in critically ill non-immunosuppressed patients. Lancet Infect. Dis. 3, 685 –702[CrossRef][Medline]

  4. Leleu, G., Aegerter, P., and Guidet, B. (2002) Systemic candidiasis in intensive care units: a multicenter, matched-cohort study. J. Crit. Care 17, 168 –175[CrossRef][Medline]

  5. Gudlaugsson, O., Gillespie, S., Lee, K., Vande, B. J., Hu, J., Messer, S., Herwaldt, L., Pfaller, M., and Diekema, D. (2003) Attributable mortality of nosocomial candidemia, revisited. Clin. Infect. Dis. 37, 1172 –1177[CrossRef][Medline]

  6. Miller, L. G., Hajjeh, R. A., and Edwards, J. E., Jr. (2001) Estimating the cost of nosocomial candidemia in the United States. Clin. Infect. Dis. 32, 1110[Medline]

  7. Wilson, L. S., Reyes, C. M., Stolpman, M., Speckman, J., Allen, K., and Beney, J. (2002) The direct cost and incidence of systemic fungal infections. Value Health 5, 26 –34[CrossRef][Medline]

  8. Horvath, L. L., Hospenthal, D. R., Murray, C. K., and Dooley, D. P. (2003) Detection of simulated candidemia by the BACTEC 9240 system with plus aerobic/F and anaerobic/F blood culture bottles. J. Clin. Microbiol. 41, 4714 –4717[Abstract/Free Full Text]

  9. Kami, M., Machida, U., Okuzumi, K., Matsumura, T., Mori, S. S., Hori, A., Kashima, T., Kanda, Y., Takaue, Y., Sakamaki, H., Hirai, H., Yoneyama, A., and Mutou, Y. (2002) Effect of fluconazole prophylaxis on fungal blood cultures: an autopsy-based study involving 720 patients with haematological malignancy. Br. J. Haematol. 117, 40 –46[CrossRef][Medline]

  10. Schwesinger, G., Junghans, D., Schroder, G., Bernhardt, H., and Knoke, M. (2005) Candidosis and aspergillosis as autopsy findings from 1994 to 2003. Mycoses 48, 176 –180[CrossRef][Medline]

  11. Reiss, E., and Morrison, C. J. (1993) Nonculture methods for diagnosis of disseminated candidiasis. Clin. Microbiol. Rev. 6, 311 –323[Abstract/Free Full Text]

  12. Ellepola, A. N., and Morrison, C. J. (2005) Laboratory diagnosis of invasive candidiasis. J. Microbiol. 43, 65 –84

  13. Matthews, R. C., Rigg, G., Hodgetts, S., Carter, T., Chapman, C., Gregory, C., Illidge, C., and Burnie, J. (2003) Preclinical assessment of the efficacy of mycograb, a human recombinant antibody against fungal HSP90. Antimicrob. Agents Chemother. 47, 2208 –2216[Abstract/Free Full Text]

  14. Han, Y., Ulrich, M. A., and Cutler, J. E. (1999) Candida albicans mannan extract-protein conjugates induce a protective immune response against experimental candidiasis. J. Infect. Dis. 179, 1477 –1484[CrossRef][Medline]

  15. Vilanova, M., Teixeira, L., Caramalho, I., Torrado, E., Marques, A., Madureira, P., Ribeiro, A., Ferreira, P., Gama, M., and Demengeot, J. (2004) Protection against systemic candidiasis in mice immunized with secreted aspartic proteinase 2. Immunology 111, 334 –342[CrossRef][Medline]

  16. Martinez, J. P., Gil, M. L., Lopez-Ribot, J. L., and Chaffin, W. L. (1998) Serologic response to cell wall mannoproteins and proteins of Candida albicans. Clin. Microbiol. Rev. 11, 121 –141[Abstract/Free Full Text]

  17. Chaffin, W. L., Lopez-Ribot, J. L., Casanova, M., Gozalbo, D., and Martinez, J. P. (1998) Cell wall and secreted proteins of Candida albicans: identification, function, and expression. Microbiol. Mol. Biol. Rev. 62, 130 –180[Abstract/Free Full Text]

  18. Lopez-Ribot, J. L., Casanova, M., Murgui, A., and Martinez, J. P. (2004) Antibody response to Candida albicans cell wall antigens. FEMS Immunol. Med. Microbiol. 41, 187 –196[CrossRef][Medline]

  19. Pitarch, A., Nombela, C., and Gil, C. (2006) in Microbial Proteomes (Humphery-Smith, I., and Hecker, M., eds) Greene Publishing Associates and Wiley Interscience, New York, in press

  20. Pitarch, A., Sanchez, M., Nombela, C., and Gil, C. (2003) Analysis of the Candida albicans proteome. I. Strategies and applications. J. Chromatogr. B Anal. Technol. Biomed. Life Sci. 787, 101 –128[Medline]

  21. Niimi, M., Cannon, R. D., and Monk, B. C. (1999) Candida albicans pathogenicity: a proteomic perspective. Electrophoresis 20, 2299 –2308[CrossRef][Medline]

  22. Pitarch, A., Sanchez, M., Nombela, C., and Gil, C. (2002) Sequential fractionation and two-dimensional gel analysis unravels the complexity of the dimorphic fungus Candida albicans cell wall proteome. Mol. Cell. Proteomics 1, 967 –982[Abstract/Free Full Text]

  23. Urban, C., Sohn, K., Lottspeich, F., Brunner, H., and Rupp, S. (2003) Identification of cell surface determinants in Candida albicans reveals Tsa1p, a protein differentially localized in the cell. FEBS Lett. 544, 228 –235[CrossRef][Medline]

  24. de Groot, P. W., de Boer, A. D., Cunningham, J., Dekker, H. L., de Jong, L., Hellingwerf, K. J., de Koster, C., and Klis, F. M. (2004) Proteomic analysis of Candida albicans cell walls reveals covalently bound carbohydrate-active enzymes and adhesins. Eukaryot. Cell 3, 955 –965[Abstract/Free Full Text]

  25. Casanova, M., Lopez-Ribot, J. L., Martinez, J. P., and Sentandreu, R. (1992) Characterization of cell wall proteins from yeast and mycelial cells of Candida albicans by labelling with biotin: comparison with other techniques. Infect. Immun. 60, 4898 –4906[Abstract/Free Full Text]

  26. Pardo, M., Monteoliva, L., Pla, J., Sanchez, M., Gil, C., and Nombela, C. (1999) Two-dimensional analysis of proteins secreted by Saccharomyces cerevisiae regenerating protoplasts: a novel approach to study the cell wall. Yeast 15, 459 –472[CrossRef][Medline]

  27. Pitarch, A., Pardo, M., Jimenez, A., Pla, J., Gil, C., Sanchez, M., and Nombela, C. (1999) Two-dimensional gel electrophoresis as analytical tool for identifying Candida albicans immunogenic proteins. Electrophoresis 20, 1001 –1010[CrossRef][Medline]

  28. Pardo, M., Ward, M., Bains, S., Molina, M., Blackstock, W., Gil, C., and Nombela, C. (2000) A proteomic approach for the study of Saccharomyces cerevisiae cell wall biogenesis. Electrophoresis 21, 3396 –3410[CrossRef][Medline]

  29. Elorza, M. V., Marcilla, A., Sanjuan, R., Mormeneo, S., and Sentandreu, R. (1994) Incorporation of specific wall proteins during yeast and mycelial protoplast regeneration in Candida albicans. Arch. Microbiol. 161, 145 –151[Medline]

  30. Krah, A., and Jungblut, P. R. (2004) Immunoproteomics. Methods Mol. Med. 94, 19 –32[Medline]

  31. Pitarch, A., Nombela, C., and Gil, C. (2006) in Microbial Proteomes (Humphery-Smith, I., and Hecker, M., eds) Greene Publishing Associates and Wiley Interscience, New York, in press

  32. Haas, G., Karaali, G., Ebermayer, K., Metzger, W. G., Lamer, S., Zimny-Arndt, U., Diescher, S., Goebel, U. B., Vogt, K., Roznowski, A. B., Wiedenmann, B. J., Meyer, T. F., Aebischer, T., and Jungblut, P. R. (2002) Immunoproteomics of Helicobacter pylori infection and relation to gastric disease. Proteomics 2, 313 –324[CrossRef][Medline]

  33. Hess, J. L., Blazer, L., Romer, T., Faber, L., Buller, R. M., and Boyle, M. D. (2005) Immunoproteomics. J. Chromatogr. B Anal. Technol. Biomed. Life Sci. 815, 65 –75

  34. Pedersen, S. K., Sloane, A. J., Prasad, S. S., Sebastian, L. T., Lindner, R. A., Hsu, M., Robinson, M., Bye, P. T., Weinberger, R. P., and Harry, J. L. (2005) An immunoproteomic approach for identification of clinical biomarkers for monitoring disease. Application to cystic fibrosis. Mol. Cell. Proteomics 4, 1052 –1060[Abstract/Free Full Text]

  35. Pardo, M., Ward, M., Pitarch, A., Sanchez, M., Nombela, C., Blackstock, W., and Gil, C. (2000) Cross-species identification of novel Candida albicans immunogenic proteins by combination of two-dimensional polyacrylamide gel electrophoresis and mass spectrometry. Electrophoresis 21, 2651 –2659[CrossRef][Medline]

  36. Pitarch, A., Abian, J., Carrascal, M., Sanchez, M., Nombela, C., and Gil, C. (2004) Proteomics-based identification of novel Candida albicans antigens for diagnosis of systemic candidiasis in patients with underlying hematological malignancies. Proteomics 4, 3084 –3106[CrossRef][Medline]

  37. Seliger, B., and Kellner, R. (2002) Design of proteome-based studies in combination with serology for the identification of biomarkers and novel targets. Proteomics 2, 1641 –1651[CrossRef][Medline]

  38. Le Naour, F., Brichory, F., Misek, D. E., Brechot, C., Hanash, S., and Beretta, L. (2002) A distinct repertoire of autoantibodies in hepatocellular carcinoma identified by proteomic analysis. Mol. Cell. Proteomics 1, 197 –203[Abstract/Free Full Text]

  39. Fathman, C. G., Soares, L., Chan, S. M., and Utz, P. J. (2005) An array of possibilities for the study of autoimmunity. Nature 435, 605 –611[CrossRef][Medline]

  40. Huguet, S., Labas, V., Duclos-Vallee, J. C., Bruneel, A., Vinh, J., Samuel, D., Johanet, C., and Ballot, E. (2004) Heterogeneous nuclear ribonucleoprotein A2/B1 identified as an autoantigen in autoimmune hepatitis by proteome analysis. Proteomics 4, 1341 –1345[CrossRef][Medline]

  41. Imafuku, Y., Omenn, G. S., and Hanash, S. (2004) Proteomics approaches to identify tumor antigen directed autoantibodies as cancer biomarkers. Dis. Markers 20, 149 –153[Medline]

  42. Wilson, R. A., Curwen, R. S., Braschi, S., Hall, S. L., Coulson, P. S., and Ashton, P. D. (2004) From genomes to vaccines via the proteome. Mem. Inst. Oswaldo Cruz 99, 45 –50

  43. Pollard, K. S., and van der Laan, M. J. (2002) Statistical inference for simultaneous clustering of gene expression data. Math. Biosci. 176, 99 –121[CrossRef][Medline]

  44. Grigoriev, A. (2004) Understanding the yeast proteome: a bioinformatics perspective. Expert Rev. Proteomics 1, 193 –205[CrossRef]