Candida glabrata Intra-Abdominal Candidiasis Is Characterized by Persistence within the Peritoneal Cavity and Abscesses

Abstract

The pathogenesis of Candida glabrata infections is poorly understood. We studied the pathogenesis of intra-abdominal candidiasis (IAC) in mice that were infected intraperitoneally with C. glabrata and sterile feces. C. glabrata BG2 (5 × 10⁸ CFU) caused a 100% mortality rate. Sublethal inocula of BG2 (1 × 10⁸ or 1 × 10⁷ CFU) caused peritonitis that progressed to abscesses. Three clinical C. glabrata strains (5 × 10⁸ CFU) caused 80 to 100% mortality rates, while a fourth (strain 346) caused a 29% mortality rate. Following sublethal inocula (1 × 10⁷ CFU), the intra-abscess burdens of virulent strain 356 were ∼1 log greater than those of strain 346. A C. glabrata Δplb1-2 mutant (phospholipase B genes disrupted) killed mice as well as BG2 did. When sublethal inocula were used, however, the Δplb1-2 mutant was associated with more rapid abscess resolution and lower intra-abscess burdens; these findings were reversed by PLB1 and PLB2 reinsertion. The Δplb1-2 mutant was also more susceptible than BG2 to killing by human neutrophils in vitro. BG2 and the Δplb1-2 mutant were indistinguishable during hematogenously disseminated candidiasis. C. albicans SC5314 was more virulent than C. glabrata BG2 during IAC, causing a 100% mortality rate following a challenge with 5 × 10⁷ CFU. In contrast, a sublethal inoculum (1 × 10⁷ CFU) of BG2 caused less neutrophil infiltration and greater burdens in peritoneal fluid than SC5314 did and abscesses that persisted longer and contained greater burdens. In conclusion, a mouse model of C. glabrata IAC mimics disease in humans and distinguishes the relative virulence of clinical and gene disruption strains. C. glabrata differed from C. albicans during IAC by being less lethal and eliciting dampened neutrophil responses but resulting in more persistent peritonitis and abscesses.

INTRODUCTION

Invasive candidiasis is the most common nonmucosal fungal disease among hospitalized patients in the United States. The predominant types of invasive candidiasis are candidemia and intra-abdominal candidiasis (IAC), infections associated with mortality rates of ∼20 to 40% (1–4). Candidemia is the fourth most frequent nosocomial bloodstream infection. IAC, which includes peritonitis and intra-abdominal abscesses, may occur in >40% of patients following repeat gastrointestinal (GI) surgery, GI perforation, or necrotizing pancreatitis. In cases of GI leakage following surgery or GI perforation, IAC develops after Candida cells and fecal matter are introduced into the normally sterile peritoneal cavity.

Much of our understanding of the molecular pathogenesis of invasive candidiasis comes from studies of Candida albicans. Pathogenesis research has been facilitated by a simple and reproducible mouse model of hematogenously disseminated C. albicans infection in which sepsis and organ invasion develop following lateral tail vein inoculation. C. glabrata is second to C. albicans as a leading cause of invasive candidiasis (5), but it is less well studied, and basic questions about pathogenicity and host adaptability are unanswered. Research has been hampered by the lack of an animal model that mimics invasive C. glabrata infection. C. glabrata does not consistently kill mice after intravenous (i.v.) infection in the absence of severe immunosuppression or extremely large inocula (>10⁸ CFU/mouse) (6–8). C. glabrata persists within mouse organs for weeks following an i.v. challenge but causes only mild inflammation and minimal tissue damage (9, 10). Indeed, the C. glabrata i.v. model of disseminated candidiasis (DC) is considered to be more reflective of persistent colonization than of invasive disease (9). Furthermore, it is limited for studying C. glabrata virulence, as tissue burdens and host responses are relatively insensitive in discriminating between the pathogenicities of different clinical strains or wild-type and gene mutant strains.

Our group and others have developed mouse models of IAC in which intraperitoneal (i.p.) inoculation of C. albicans mixed with sterile feces results in peritonitis and abscess formation (11, 12). IAC models have been used to evaluate C. albicans gene expression and virulence, host immune responses, and effectiveness of antifungal treatment. In this study, we adapted these models to study C. glabrata IAC pathogenesis.

(Preliminary data were presented at the Interscience Conference on Antimicrobial Agents and Chemotherapy in September 2013 in Denver, CO.)

MATERIALS AND METHODS

The Candida strains used in this study are listed in Table 1 (13, 14). Strains were routinely grown in yeast extract-peptone-dextrose (YPD) medium (1% yeast extract, 2% Bacto peptone, 2% α-d-glucose) at 30°C unless otherwise noted.

TABLE 1.

C. glabrata and C. albicans strains and primers used in this study

Strain(s) or primer	Genotype or sequence (5′–3′)a	Reference
Strains
BG2	Clinical isolate	14
Δplb1 mutant	plb1::FRT	This study
Δplb2 mutant	plb2::FRT	This study
Δplb1-2 mutant	plb1-2::FRT	This study
PLB1 PLB2 reinsertion	PLB1 PLB2::FRT	This study
SC5314	Clinical isolate	13
5, 8, 346, 356	Clinical isolates	This study
Primers
PLB1F1For (KpnI)	CGACGGGGTACCCTATCCCGCTTCTAATTTG
PLB1F1Rev (XhoI)	AACTTCTCGAGGATACTTTAGTATTTGCTCG
PLB1F2For (SacII)	TGGGCCATCTTCCGCGGTCTAATCCCTTGGGGATGAT
PLB1F2Rev (SacI)	CTCATAGATTACACGAGCTCTGCCCGAGTTGTATTTGTCA
PLB2F1For (KpnI)	CGACGGGGTACCGGGTCTAACACGAAGATTCT
PLB2F1Rev (XhoI)	AACTTCTCGAGGTTTCCCTTTTCACTTCGCA
PLB2F2For (SacII)	TGGGCCATCTTCCGCGGCCTAACGCTAATACTCGTGCA
PLB2F2Rev (SacI)	CTCATAGATTACACGAGCTCATGAAGAAGGTGGCAAAAA
Reins_For (ApaI)	CAAGTCAACGTAGGGCCCCTATCCCGCTTCTAATTTG
Reins_Rev (SgrDI)	AACTTGGTGCGTCGACGATGAAGAAGGTGGCAAAAA

^aUnderlined nucleotides are restriction sites that were introduced.

Mouse models of IAC and DC were prepared according to the University of Pittsburgh Institutional Animal Care and Use Committee guidelines. In the IAC model, mice (6 to 10 per group) were infected i.p. with a 100-μl Candida suspension grown overnight in Sabouraud dextrose broth at 30°C; inocula were prepared with and without steam-sterilized mouse feces (sterile feces) (12). Mouse feces was homogenized in a tissue grinder, suspended in normal saline (NS) to form a 5% (wt/vol) mixture, and sterilized in a steam autoclave (15 min, 2 × 10⁵ Pa, 120°C). Nonviable organisms were confirmed by culture in Luria broth (for aerobic bacteria), chopped meat broth (for anaerobic bacteria), and YPD medium (for fungi). To assess mice for peritonitis, the peritoneum was washed with 1 ml of phosphate-buffered saline and cell counts and tissue burdens were determined (12). Abdominal cavities were then explored, and abscesses >1 mm in diameter were counted, removed, and homogenized for colony enumeration. Thereafter, organs were homogenized for Candida cell enumeration. We did not evaluate burdens within kidneys, which are retroperitoneal and did not yield consistent results in previous studies (12). In the DC model, mice (six per group) were infected via the lateral tail vein with C. glabrata suspended in 100 μl of sterile saline. Tissue burdens are presented as the mean (± standard error) log₁₀ number of CFU per gram of tissue. The difference in tissue burdens between mice infected with different strains was determined by Wilcoxon test. A P value of <0.05 was considered statistically significant.

Construction of PLB1 and PLB2 mutants.

The PLB1 and PLB2 genes were disrupted in C. glabrata strain BG2 by the SAT1 flipper method (9, 15–17). The proximal and distal flanking regions of PLB1 or PLB2 were amplified by PCR with the primers described in Table 1. The KpnI and XhoI and the SacII and SacI restriction sites were introduced. Following amplification, PCR products were digested with appropriate restriction enzymes and ligated sequentially into plasmid pSFS2A (see Fig. S1 in the supplemental material). The C. glabrata PLB1 and PLB2 genes are located 1.5 kb apart in the same orientation on the same chromosome. To disrupt PLB1 and PLB2, PCR products containing the proximal flanking region of PLB1 and the distal flanking region of PLB2 (see Fig. S1) were ligated sequentially into plasmid pSFS2A. The PLB1, PLB2, and PLB1 PLB2 disruption cassettes were released by digestion with KpnI/SacI and transformed into C. glabrata strain BG2 by electroporation. Nourseothricin-resistant (Nou^r) transformants were selected on YPD plates containing 200 μg/ml nourseothricin at 30°C. After confirmation of disruption by PCR, Nou^r colonies were grown in YPMal (1% yeast extract, 1% Bacto peptone, 2% maltose) at 30°C for 6 h. The culture was then diluted and placed on YPD agar plates supplemented with 15 μg/ml nourseothricin. The smaller colonies were parallel streaked onto YPD agar with and without 100 μg/ml nourseothricin to verify nourseothricin sensitivity. Genomic DNA was isolated from nourseothricin-sensitive (Nou^s) colonies, and excision of the cassette was verified by Southern blotting.

To reinsert PLB1 and PLB2 at their own locus, PLB1, PLB2, and their flanking regions were amplified with primers Reins_For_ApaI and Reins_Rev_SgrDI (Table 1; see Fig. S1 in the supplemental material), which introduced ApaI and SgrD1 sites, respectively. The amplified fragment and the PLB2 distal flanking fragment (PLB2F2) were sequentially cloned into pSFS2A. The reinsertion cassette was digested by ApaI digestion and used to transform a Nou^s PLB1 and PLB2 null mutant as described above. The success of the reinsertion was confirmed by Southern blotting.

Neutrophil killing assays were performed with human neutrophils that were incubated with opsonized C. glabrata in RPMI 1640 medium with 10% human serum (18). The neutrophil/C. glabrata ratios used were 5:1 and 10:1. After 2 h of incubation at 37°C with shaking at 75 rpm, sterile water was added to 10 ml. The percentage of organisms killed was calculated as follows: [1 − (number of CFU in tubes containing neutrophils/number of CFU in tubes without neutrophils)] × 100. Each experiment was performed in triplicate with neutrophils from different healthy volunteers.

RESULTS

A mouse model of C. glabrata IAC.

We first performed a dose-ranging survival study with ICR mice infected i.p. with C. glabrata BG2. No mice died in the 28 days following inoculation with 1 × 10⁷ or 1 × 10⁸ CFU of BG2 in the presence or absence of sterile feces or with 5 × 10⁸ CFU in the absence of sterile feces. All of the mice infected with 5 × 10⁸ CFU mixed with sterile feces died by day 8 (Fig. 1A). On necropsy, numerous abscesses were affixed to the peritoneal surface, bowel, and intra-abdominal organs. The C. glabrata burdens within the parenchyma of the liver, spleen, and pancreas were >8 log₁₀ CFU/g of tissue, indicating that the mice died with overwhelming DC.

FIG 1.

Outcomes of i.p. infection of mice with C. glabrata. (Top panel) Survival of mice infected with various inocula of BG2 with or without sterile feces. Ten mice were included in each group, and experiments were repeated once. All of the mice infected with 5 × 10⁸ CFU of BG2 mixed with feces died by day 8 after infection. In contrast, all of the mice infected with 1 × 10⁸ CFU of BG2 with or without feces or with 5 × 10⁸ CFU of BG2 without feces survived to at least day 28 after infection. (Second panel from top) Tissue burdens in the peritoneal fluid of mice infected with 1 × 10⁷ CFU of BG2 with or without sterile feces. The values on the y axis are the mean log₁₀ numbers of CFU/ml of peritoneal fluid ± the standard deviation from at least eight mice per group. Note that mice infected with BG2 mixed with sterile feces had more prolonged positive cultures within their peritoneal fluid than did mice infected with BG2 alone. In the presence of sterile feces, C. glabrata persisted within peritoneal fluid through day 14. C. glabrata was cleared from the peritoneal fluid of all of the mice by day 7 in the absence of sterile feces. (Third panel from top) Tissue burdens within abscesses of mice infected with 1 × 10⁷ CFU of BG2/mouse with or without sterile feces. Mice infected with BG2 alone did not form intra-abdominal abscesses. The values on the y axis are the mean log₁₀ numbers of CFU/g of abscess ± the standard deviation from at least six mice per group per time point. (Fourth panel from top) Necropsy and histopathology findings on mice infected with 1 × 10⁷ CFU of BG2/mouse. (A) Necropsy of a mouse 7 days after intraperitoneal infection with BG2 mixed with sterile feces showing multiple coalescing abscesses. (B, C) Hematoxylin-and-eosin staining of a pancreatic abscess (magnifications: B, ×100; C, ×400). Note the outer abscess border and dense polymorphonuclear neutrophil (PMN) infiltrate surrounding a necrotic center. (D) Methamine silver staining (magnification, ×600) showing clusters of budding yeast cells ∼2 μm in size. (Bottom panel) Survival of mice infected with clinical strains of C. glabrata. At least eight mice per strain were infected i.p. with 5 × 10⁸ CFU of each strain. Strain 5 was associated with candidemia complicated by endophthalmitis and endocarditis, strain 356 was associated with candidemia complicated by endophthalmitis, strain 346 was associated with i.v. catheter infection without positive peripheral blood cultures or organ infection, and strain 8 was associated with urinary tract infection. The strains exhibited similar growth rates in YPD medium at 30°C.

We next studied mice over the course of 14 days following sublethal i.p. infection with 1 × 10⁷ CFU of BG2. In the presence of sterile feces, the mean log₁₀ C. glabrata burdens within peritoneal fluid progressively decreased from day 1 through day 7 (Fig. 1B). Beginning on day 1, the C. glabrata burdens in mice infected with BG2 mixed with sterile feces were significantly greater and persisted longer than those in mice infected with BG2 alone. Neutrophil counts in the peritoneal fluid increased from 6 h to 1 day and were greater for C. glabrata mixed with sterile feces than for C. glabrata or sterile feces alone (Table 2). Beginning on day 3, the leukocyte response was predominantly monocytic. The peritoneal fluid at each time point was alkaline (mean pH, 7.8).

TABLE 2.

Neutrophil responses within the peritoneal fluid of mice infected with sterile feces alone, BG2 alone, or BG2 mixed with sterile feces^a

Time after infection	Mean no. of PMNs/mm3 ± SE			P value for:
	Feces alone	BG2 alone	BG2 + feces	Feces vs BG2	Feces vs BG2 + feces	BG2 vs BG2 + feces
6 h	1,068 ± 199	336 ± 111	1,770 ± 194	0.018	0.045	0.001
Day 1	1,391 ± 582	66 ± 23	2,607 ± 506	0.029	0.037	0.009
Day 3	390 ± 175	52 ± 52	513 ± 200	0.115b	0.66b	0.07b
Day 7	193 ± 71	133 ± 123	464 ± 130	0.46b	0.24b	0.133b
Day 14	22.7 ± 9.5	0.0 ± 0.0	84.6 ± 43.9	<0.0001c	0.27b	<0.0001c

^aCell counts were determined by the University of Pittsburgh Medical Center Clinical Hematology Laboratory with a Coulter HMX Hematology Analyzer. The values are the numbers of polymorphonuclear neutrophils (PMNs)/mm³ measured in the peritoneal fluid of at least four mice. Note that neutrophils were the predominant type of leukocyte at 6 h and 1 day. Thereafter, neutrophil counts decreased progressively through day 14 and did not differ significantly between mice infected with C. glabrata mixed with sterile feces and mice infected with C. glabrata or sterile feces alone.

^bNo statistically significant difference.

^cThe number of neutrophils within the peritoneal fluid all of the mice tested that were infected with BG2 alone was 0. To calculate P values, we assigned a neutrophil count of 1.1 cells/mm³ to one sample.

Intra-abdominal abscesses were apparent upon gross pathologic examination on day 3 among mice infected with C. glabrata BG2 (1 × 10⁷ CFU) mixed with sterile feces. They were larger and more common on day 7 (median number, 10/mouse) and persisted through day 14 (Fig. 1C). Abscesses were evident on the peritoneal surface, bowel, and intra-abdominal organs, but the predominant site of involvement was the pancreas. Histopathologic examination showed that abscesses were composed of abundant yeast forms and cellular necrosis, which were surrounded by an acute inflammatory response characterized by brisk neutrophil infiltration (Fig. 1D). Mice infected with C. glabrata alone did not develop abscesses. Mice infected with sterile feces alone developed sterile abscesses, which were small and not well defined and resolved by day 7. Subsequent experiments were performed in the presence of sterile feces to mimic intra-abdominal infections that occur in humans following GI perforation.

Relative virulence of C. glabrata clinical strains during IAC.

To determine if C. glabrata clinical strains manifest differing levels of virulence during IAC, we performed survival studies with mice (eight or more per group) infected with strains in the presence of sterile feces. The strains exhibited comparable growth rates in YPD at 30 and 37°C (see Table S1 in the supplemental material). The strains used were associated with various types of disease in humans (Fig. 1E). The mortality rates following i.p. inoculation of strains 5, 8, and 356 were 80 to 100%. In contrast, the mortality rate among mice infected with strain 346 was only 29% (2/7; P = 0.02 versus strain 356). Necropsies demonstrated numerous intra-abdominal abscesses and organ tissue burdens that were similar to those caused by strain BG2 (data not shown).

We next compared the progression of IAC in mice infected with a sublethal inoculum (1 × 10⁷ CFU in the presence of sterile feces) of strain 356 or 346 (Table 3). C. glabrata burdens did not differ within peritoneal fluid over 14 days. There was no significant difference in the mean burdens of the strains within abscesses at 3 days. On days 7 and 14, however, the intra-abscess burdens due to strain 356 were ∼1 log greater than those due to strain 346. There were also greater burdens of strain 356 within the liver, spleen, and pancreas on days 7 and 14.

TABLE 3.

Tissue burdens of mice infected with strains 346 and 356

Sample and strain or parameter	Mean tissue burden ± SE or P valuea on:
	Day 1	Day 3	Day 7	Day 14
Peritoneal fluid
346	3.85 ± 0.45	3.48 ± 0.10	2.09 ± 0.13	1.78 ± 0.21
356	4.46 ± 0.04	3.72 ± 0.16	2.09 ± 0.12	1.57 ± 0.18
P value	0.21b	0.24b	0.96b	0.50b
Abscess
346	NA	7.79 ± 0.14	6.11 ± 0.40	4.95 ± 0.34
356	NA	7.10 ± 1.36	7.03 ± 0.12	6.54 ± 0.59
P value		0.30b	0.05	0.04
Liver
346	4.92 ± 0.17	2.93 ± 0.16	1.67 ± 0.17	0.65 ± 0.38
356	4.08 ± 0.17	3.74 ± 0.26	2.48 ± 0.31	2.40 ± 0.56
P value	0.12b	0.04	0.046	0.04
Spleen
346	5.40 ± 0.15	4.08 ± 0.10	3.30 ± 0.13	1.22 ± 0.58
356	5.10 ± 0.14	4.58 ± 0.30	3.85 ± 0.14	3.06 ± 0.53
P value	0.20b	0.15b	0.02	0.04
Pancreas
346	6.17 ± 0.46	4.26 ± 0.10	2.79 ± 0.16	1.41 ± 0.39
356	6.14 ± 0.34	4.56 ± 0.21	3.56 ± 0.18	2.46 ± 0.54
P value	0.98b	0.22b	0.01	0.03

^aAt least six mice per group per time point were infected with 1 × 10⁷ CFU of C. glabrata strains mixed with sterile feces. The values are the mean log₁₀ numbers of CFU/ml of peritoneal fluid or mean log₁₀ numbers of CFU/g of abscess ± the standard error. P values denote the statistical significance of differences between the tissue burdens of mice infected with isolate 346 and mice infected with isolate 356.

^bNo statistically significant difference.

Contribution of C. glabrata PLB1 and PLB2 to pathogenesis.

Phospholipase B (PLB) is a secretory enzyme that hydrolyzes phospholipids, plays key roles in cellular processes like signal transduction and membrane modeling, and mediates microbial virulence by disrupting host cell membranes (19). C. albicans PLB1 has been implicated in the pathogenesis of DC and translocation across GI mucosa in mouse models (20). C. glabrata PLB1 and PBL2 are 41.8 and 39.3% similar to C. albicans PLB1 and PLB2, respectively, and they encode proteins with 51.3% similarity to each other. To determine if C. glabrata PLB1 and/or PLB2 contribute to the pathogenesis of IAC, we created single-gene (Δplb1 and Δplb2) mutant strains and a double-gene (Δplb1-2) mutant by the SAT1 flipper method (15) (Table 1; see Fig. S1 in the supplemental material). We also reintroduced full copies of PLB1 and PLB2 into the Δplb1-2 mutant to create a reinsertion strain. The single- and double-gene mutant strains and the reinsertion strain were viable and demonstrated growth rates in YPD medium at 30 and 37°C that were indistinguishable from those of parental strain BG2 (the doubling times at 30 and 37°C of BG2 and the single- and double-gene mutant strains were 1.47 ± 0.2 and 1.04 ± 0.01 h, respectively) (see Table S1 in the supplemental material). Following i.p. infection (5 × 10⁸ CFU in the presence of sterile feces), the Δplb1, Δplb2, and Δplb1-2 mutants resembled parental strain BG2 by killing mice over the course of 7 days.

To compare strains during sublethal infections (1 × 10⁷ CFU in the presence of sterile feces), we elected to monitor mice for 28 days since our previous experiments showed persistence of C. glabrata within abscesses through day 14. Burdens of Δplb1 or Δplb2 mutant cells were not attenuated in peritoneal fluid or within abscesses (data not shown). The Δplb1-2 mutant was not attenuated in peritoneal fluid, but it was impaired in persistence within abscesses (Table 4). On day 14, 28% (4/14) and 86% (12/14) of the mice infected with the Δplb1-2 mutant and BG2 had abscesses, respectively (P = 0.006). The findings were similar on day 21 (33% [4/12] versus 83% [10/12]; P = 0.04) and day 28 (0% [0/12] versus 75% [6/8]; P = 0.0007). Significantly lighter burdens of the mutant were also evident within abscesses at these time points (Table 4).

TABLE 4.

Tissue burdens within peritoneal fluid and intra-abdominal abscesses of mice infected i.p. with BG2 or the Δplb1-2 mutant mixed with sterile feces

Sample and strain or parameter	Mean tissue burden ± SE or P valuea on:
	Day 3	Day 7	Day 14	Day 21	Day 28
Peritoneal fluid
BG2	3.42 ± 0.20	2.24 ± 0.18	1.36 + 0.16	0.16 ± 0.16	0.00 ± 0.00
Δplb1-2 mutant	3.50 ± 0.13	2.04 ± 0.12	0.00 ± 0.00c	0.00 ± 0.00c	0.00 ± 0.00
Reinsertion strain	3.58 ± 0.27	2.03 ± 0.18	2.03 ± 0.17	2.03 ± 0.50	0.00 ± 0.00
P value	0.80b	0.17b	<0.0001c	0.38b,c	NDd
Abscess
BG2	7.74 ± 0.524	7.18 ± 0.16	4.84 ± 0.59	4.43 ± 0.97	4.14 ± 0.93
Δplb1-2 mutant	5.12 ± 1.63	5.21 ± 1.50	1.42 ± 0.93	1.15 ± 0.75	0.00 ± 0.00c
Reinsertion strain	6.39 ± 1.61	7.08 ± 0.08	3.57 ± 1.17	3.23 ± 1.11	2.38 ± 1.02
P value	0.18b	0.13b	0.004	0.02	<0.0001c
Liver
BG2	3.29 ± 0.84	2.06 ± 0.17	1.20 ± 0.33	0.34 ± 0.21	0.15 ± 0.15
Δplb1-2 mutant	3.52 ± 0.28	1.77 ± 0.07	0.75 ± 0.39	0.26 ± 0.26	0.09 ± 0.09
Reinsertion strain	3.23 ± 0.72	1.74 ± 0.17	1.36 ± 0.36	0.84 ± 0.46	0.28 ± 0.18
P value	0.78b	0.15b	0.39b	0.81b	0.74b
Spleen
BG2	4.46 ± 0.31	3.95 ± 0.14	3.06 ± 0.26	2.81 ± 0.57	1.81 ± 0.55
Δplb1-2 mutant	4.60 ± 0.21	3.66 ± 0.18	2.07 ± 0.70	2.74 ± 0.56	1.78 ± 0.39
Reinsertion strain	4.56 ± 0.17	4.29 ± 0.10	3.41 ± 0.21	3.34 ± 0.10	2.24 ± 0.50
P value	0.70b	0.24b	0.11b	0.94b	0.96b
Pancreas
BG2	4.41 ± 0.31	3.13 ± 0.15	2.20 ± 0.22	2.39 ± 0.33	0.00 ± 0.00
Δplb1-2 mutant	4.55 ± 0.25	3.05 ± 0.13	1.23 ± 0.49	2.28 ± 0.51	0.00 ± 0.00
Reinsertion strain	4.22 ± 0.35	3.43 ± 0.15	2.41 ± 0.13	2.30 ± 0.06	0.00 ± 0.00
P value	0.75b	0.67b	0.06b	0.85b	ND

^aAt least eight mice per group per time point were infected with 1 × 10⁷ CFU of each strain. The values are the mean log₁₀ numbers of CFU/ml ± the standard of error of peritoneal fluid or the mean log₁₀ numbers of CFU/g of abscess ± the standard of error. The P values describe the statistical significance of differences between the tissue burdens of mice infected with BG2 and mice infected with the Δplb1-2 mutant. Note that mice infected with the Δplb1-2 mutant had lighter Candida burdens within abscesses than mice infected with BG2. The intra-abscess C. glabrata burdens of mice infected with the PLB1 PLB2 reinsertion strain did not differ from those of mice infected with BG2.

^bNo statistically significant difference.

^cThe tissue burden within samples was 0 for all of the mice tested. To calculate the P value, we assumed that the tissue burden of one sample was 1.1 CFU/ml of fluid or CFU/g of tissue.

^dND, no P value could be determined because the tissues from all of the groups were sterile.

Since neutrophils play a key role in the clearance of Candida from intra-abdominal abscesses, we evaluated the susceptibility of BG2 and the Δplb mutant strains to killing by human neutrophils. For C. glabrata cells in stationary phase, the mean killing of the Δplb1-2 mutant (25.3% ± 6.4%) was greater than that of BG2 (12.3% ± 1.4%; P = 0.03), the Δplb1 mutant (12.1% ± 3.6%; P = 0.04), the Δplb2 mutant (13.0% ± 6.8%; P = 0.08), and the PLB1 PLB2 reinsertion strain (9.5% ± 5.0%; P = 0.06). With mid-log-phase cells, the killing percentages of the Δplb1-2 mutant and BG2 were 59.8% ± 5.7% and 34.2% ± 7.5%, respectively (P = 0.02).

Following lateral tail vein injection, neither the Δplb1-2 mutant nor BG2 (1 × 10⁸ CFU) killed mice during DC. There were no differences in tissue burdens within the kidneys, liver, or spleen (data not shown).

Comparison of C. glabrata and C. albicans during IAC.

To compare C. glabrata and C. albicans during IAC, we performed a dose-ranging survival study of mice infected i.p. with C. albicans SC5314 and sterile feces. The doubling times of BG2 and SC5314 in YPD at 30°C were 1.47 and 1.30 h, respectively. No mice died following infection with 1 × 10⁷ CFU, whereas 100% of the mice died in the 7 days following infection with 5 × 10⁷ CFU. Next, we studied the progression of IAC by infecting mice i.p. with 1 × 10⁷ CFU of strain BG2 or SC5314 (Table 5). From day 1 through day 21, the burdens of C. glabrata BG2 in peritoneal fluid were significantly greater than those of C. albicans SC5314. Whereas BG2 persisted in peritoneal fluid through day 21, SC5314 was last recovered on day 7. The number of neutrophils within peritoneal fluid was greater in response to SC5314 than in response to BG2 at days 1 and 3 (Table 6). SC5314 caused small abscesses on day 1, which were larger on day 3 and distributed randomly in intra-abdominal organs. Abscesses were not apparent following BG2 infection until day 3, and they were smaller, often coalesced, and found disproportionately in the pancreas. As in the peritoneal fluid, BG2 persisted longer and achieved greater tissue burdens within abscesses than SC5314 did (Table 5). The burdens of SC5314 and BG2 within intra-abdominal organs differed by site and time point.

TABLE 5.

Tissue burdens within peritoneal fluid and intra-abdominal abscesses of mice infected i.p. with C. albicans SC5314 and C. glabrata BG2 mixed with sterile feces

Sample and strain or parameter	Mean tissue burden ± SE or P valuea on:
	Day 1	Day 3	Day 7	Day 14	Day 21	Day 28
Peritoneal fluid
SC5314	5.02 ± 0.15	3.26 ± 0.12	1.83 ± 0.14	0.00 ± 0.00b	0.00 ± 0.00b	0.00 ± 0.00
BG2	5.54 ± 0.05	4.20 ± 0.09	2.80 ± 0.12	1.93 ± 0.14	1.60 ± 0.02	0.00 ± 0.00
P value	0.017	0.001	0.002	<0.0001b	<0.0001b	NDd
Abscess
SC5314	NA	6.52 ± 0.14	5.08 ± 0.23	3.15 ± 0.46	0.73 ± 0.72	0.00 ± 0.00b
BG2	NA	7.74 ± 0.13	6.77 ± 0.08	6.04 ± 0.26	5.41 ± 0.36	4.14 ± 2.64
P value	NA	0.002	0.0005	0.002	0.003	<0.0001b
Liver
SC5314	5.44 ± 0.13	4.70 ± 0.08	2.9 ± 0.42	1.64 ± 0.41	0.00 ± 0.00	0.00 ± 0.00b
BG2	4.45 ± 0.24	4.14 ± 0.21	3.45 ± 0.37	3.16 ± 0.58	0.00 ± 0.00	0.15 ± 0.42
P value	0.01	0.046	0.37c	0.69c	ND	0.0001c
Spleen
SC5314	5.23 ± 0.06	4.65 ± 0.40	2.17 ± 0.50	0.00 ± 0.00b	0.00 ± 0.00b	0.00 ± 0.00b
BG2	5.50 ± 0.06	4.92 ± 0.01	4.23 ± 0.28	3.38 ± 0.19	2.92 ± 0.10	1.81 ± 1.56
P value	0.02	0.04	0.05	<0.0001b	<0.0001b	0.23c
Pancreas
SC5314	6.34 ± 0.10	4.79 ± 0.22	2.3 ± 0.17	0.00 ± 0.00b	0.00 ± 0.00	0.00 ± 0.00
BG2	5.76 ± 0.30	5.40 ± 0.20	3.36 ± 0.18	1.81 ± 0.01	0.00 ± 0.00	0.00 ± 0.00
P value	0.11c	0.09c	0.0003	<0.0001	ND	ND

^aAt least eight mice per group per time point were infected with 1 × 10⁷ CFU of each strain. The values are the mean log₁₀ numbers of CFU/ml of peritoneal fluid or mean log₁₀ numbers of CFU/g of abscess ± the standard of error. The P values describe the statistical significance of differences between the tissue burdens of mice infected with SC5314 and mice infected withBG2. Note that SC5314 caused significantly greater tissue burdens than BG2 within the liver at early time points (days 1 and 3) but not later. In contrast, BG2 caused significantly greater burdens within the pancreas at later time points (days 7 and 14) but not early. Within the spleen, BG2 achieved significantly higher concentrations at each time point from day 1 though day 28.

^bThe tissue burdens within samples were 0 for all of the mice tested. To calculate P values, we assumed that the tissue burden of one sample was 1.1 CFU/ml of fluid or CFU/g of tissue.

^cNo statistically significant difference.

^dND, no P value could be determined because the tissue burdens of all of the mice in both groups were 0.

TABLE 6.

Neutrophil responses within the peritoneal fluid of mice infected with C. glabrata BG2 or C. albicans SC5314 mixed with sterile feces

Day	Median no. of neutrophils/mm3 ± SEa		P value
	SC5314	BG2
1	2,657 ± 254	1,955 ± 71	0.032
3	1,629 ± 430	551 ± 57	0.048
7	765 ± 378	1,019 ± 244	0.60b
14	182 ± 60	94 ± 48	0.30b
21	93 ± 32	88 ± 15	0.48b

^aThe values are the numbers of neutrophils measured in the peritoneal fluid of at least four mice. For mice infected with either BG2 or SC5314, a prominent neutrophil response within the peritoneal cavity was seen on day 1, which switched to mononuclear cell predominance by day 7.

^bNo statistically significant difference.

DISCUSSION

To our knowledge, this is the first report of a mouse model of C. glabrata invasive infection that mimics disease in humans and distinguishes the relative virulence of clinical and gene disruption strains. C. glabrata-host interactions in this model afford insights into the pathogenesis of IAC. Large inocula of several C. glabrata strains mixed with sterile feces and administered i.p. resulted in deaths due to candidiasis. With smaller inocula, IAC progressed from peritonitis to intra-abdominal abscesses, as is observed in patients who do not die of acute disease following GI perforation or leakage. C. glabrata peritonitis in mice was characterized by a rapid influx of neutrophils into the peritoneal cavity. C. glabrata the peritoneal cavity formed abscesses in contiguous organs by day 3 that were composed of yeast cells that were walled off by neutrophils. C. glabrata was slowly eliminated from both peritoneal fluid and abscesses, persisting in most of the mice for at least 21 and 28 days, respectively. Taken together, the data indicate that C. glabrata can cause acute disease and death if the host does not gain control of an intra-abdominal infection rapidly. In contrast, C. glabrata causes persistent infections that are tolerated by the host if the progression of IAC is arrested within the first few days.

The pathogenesis of IAC due to C. glabrata differed from that of IAC due to C. albicans in important ways. As expected, C. albicans SC5314 was clearly more virulent than C. glabrata BG2 by several measures, including the ability to cause death with a smaller inoculum, a more brisk neutrophil response within the peritoneal cavity, abscesses in the absence of sterile feces, and abscesses at earlier time points in the presence of sterile feces. At the same time, C. glabrata BG2 achieved significantly higher concentrations within peritoneal fluid and abscesses and persisted longer at these sites. The seemingly paradoxical findings are best understood in the context of the damage response framework of pathogenesis, which proposes that the outcome of infection depends upon the interplay between the pathogen and the host (21). In this regard, each of the outcomes we observed may be explained by C. albicans causing a more robust inflammatory response than C. glabrata. At one extreme, an overexuberant inflammatory response may contribute to the death of mice receiving a lethal i.p. challenge, as has been demonstrated in mouse models of sepsis induced by cecal ligation and puncture (22). With smaller inocula, the stronger neutrophil response engendered by C. albicans may contribute to more aggressive abscess formation, lighter tissue burdens, and shorter durations of infection than those seen with C. glabrata.

In some regards, our findings on C. glabrata IAC were similar to previous descriptions of DC. Most notably, C. glabrata inoculated i.v. persisted for at least 21 days within mouse organs like the kidney, liver, and spleen (9). Therefore, the ability to maintain long-term infections at different sites within the host is a general property of C. glabrata that is relevant to diverse types of deeply invasive disease. It has been proposed that persistence during DC reflects a particular capacity of C. glabrata for immune evasion (23). Indeed, C. glabrata is known to survive an attack by phagocytes and replicate within macrophages after engulfment (23–26). Along these lines, neutrophil activation has been shown to be critical to the elimination of C. albicans from intra-abdominal abscesses (11). In other aspects of pathogenesis, C. glabrata IAC differed dramatically from DC. Most importantly, IAC caused both death and tissue damage (in the form of neutrophil infiltration and abscess formation), endpoints that were not evident during DC.

In part, the fecal contamination that is a key element of the pathogenesis of IAC may account for differences from DC. Sterile feces was necessary for death and sustained abscess formation in the IAC model and resulted in greater early neutrophil influx, greater tissue burdens, and longer persistence within the peritoneal cavity. In human IAC, perforation or leakage of the GI tract is commonly associated with abscess formation but abscesses rarely result from infected peritoneal catheters. Sterile feces potentiates intra-abdominal infections by depleting complement-derived opsonins and impairing phagocytic killing (27, 28). These effects likely account for our observation that tissue burdens and persistence of C. glabrata were increased in the presence of sterile feces, despite greater neutrophil counts. Of note, the inocula used in our model are relevant to the pathogenesis of human IAC, as they fall within the reported range of C. glabrata concentrations in the feces of persons receiving short courses of broad-spectrum antibiotics (10⁶ to 10¹⁰ CFU/g) (29, 30).

There were clear differences between C. glabrata clinical strains in the ability to kill immunocompetent mice and persist within abscesses and tissues, which suggests that strain-specific factors may contribute to the course of disease in patients. The relative importance of strain versus host contributions to the manifestations, severity, and outcomes of C. glabrata infections has been a matter of debate. In general, impairments of host defenses and immunologic responses are taken to be major drivers of disease, and C. glabrata strain virulence is considered less important (31). In part, this sentiment is shaped by an acceptance that C. glabrata is intrinsically less virulent than C. albicans. Moreover, it has been difficult to show differences among C. glabrata clinical strains or gene disruption mutants by using existing in vivo models without immunosuppressing mice or using inoculum concentrations that greatly exceed those of human infections. As shown in this study and others, however, the “virulence” of C. glabrata may be underestimated if only acute endpoints like death are considered. Furthermore, it is reasonable to hypothesize that strain-specific factors may be particularly relevant to the pathogenesis of IAC since the majority of patients who develop these infections are not immunosuppressed by traditional criteria. The four strains in the present study were recovered from patients with extreme clinical manifestations ranging from an uncomplicated i.v. catheter infection to bloodstream infections associated with severe organ complications like endophthalmitis and endocarditis. It is not possible to establish conclusive links between the virulence of strains during mouse IAC and human infections. Nevertheless, it is intriguing that the most attenuated strain, as judged by both death and tissue burden endpoints, in mice was associated with the least severe disease state among patients. In the future, clinical strains defined as virulent or hypovirulent in our model may be valuable resources for comparative genomic and transcriptomic studies.

As for C. albicans, the full virulence of C. glabrata strains depends upon the interplay of multiple factors. This fact was highlighted by our demonstration that C. glabrata PLB1 and PLB2 only contributed to the pathogenesis of IAC at certain stages of disease. A Δplb1-2 null mutant strain showed decreased tissue burdens and persistence within abscesses, but it was not impaired in the ability to cause death or peritonitis. PLB activity is an important determinant of the virulence of several fungi, including C. albicans, Cryptococcus neoformans, and Aspergillus fumigatus (19). The mechanisms by which PLB contributes to infections by these fungi are not fully defined, but it has been proposed that the secreted enzyme hydrolyzes host cell membrane phospholipids, thereby facilitating tissue damage and invasion (19). C. glabrata PLB1 and PLB2 have some degree of functional redundancy, as Δplb1 and Δplb2 single-gene mutants were not attenuated during IAC or more susceptible to phagocytosis. Indeed, basic steps in the pathogenesis of C. glabrata infections like tissue adhesion and resistance to phagocytosis are complex processes mediated by multiple genes, as best exemplified by the EPA adhesin gene family of >20 paralogues (32). It is notable that the Δplb1-2 mutant was indistinguishable from parental strain BG2 following i.v. infection, which indicates either that PLB does not play a role in this form of disease or that the DC model is insufficiently sensitive to detect a contribution.

In conclusion, our mouse model of C. glabrata IAC advances the field in several ways. It is simple and reproducible, uses clinically relevant inocula, and does not require immunosuppressive medications or expensive inbred mice with underlying immune deficiencies. C. glabrata IAC in mice mimics disease in humans, as it progresses from peritonitis to abscess formation following perforation or leakage of the GI tract. Most importantly, the model is suitable for characterizing the relative virulence of C. glabrata clinical strains and gene mutants through measurements of mortality rates or tissue burdens. In each of these aspects, the IAC model offers significant advantages over the C. glabrata DC model. As we have demonstrated, the IAC model reveals unique insights into pathogenesis, C. glabrata virulence, and differences between C. glabrata and C. albicans.