Cardiac Failure in C5-Deficient A/J Mice after Candida albicans Infection

Abstract

The effect of a deficiency in the C5 component of complement on the pathophyisology of infection with the fungal pathogen Candida albicans was studied by using the A/J inbred mouse strain and the BcA17 congenic mouse strain. Acute infection caused by intravenous injection of C. albicans blastospores is associated with rapid fungal replication in the heart, brain, and, in particular, kidneys of C5-deficient mice. Histological studies and analysis of markers for tissue damage indicated that the heart is the organ that is most affected and that it ultimately fails in C5-deficient mice. In A/J and BcA17 mice, tissue damage is associated with (i) cellular infiltration in the heart, which is not seen in the kidney despite the higher fungal load in the latter organ, and (ii) a very strong inflammatory response, including elevated levels of many cytokines and chemokines. This results in cardiomyopathy, which is associated with elevated levels of creatine kinase and cardiac troponin I in the circulation. Damage to the cardiac muscle is associated with metabolic changes, including hypoglycemia, decreased lipid utilization resulting in elevated levels of cardiac triglycerides, and unproductive glucose utilization linked to a dramatic increase in the level of pyruvate dehydrogenase kinase 4 (Pdk4), a negative regulator of the pyruvate dehydrogenase complex.

Systemic infection with Candida albicans is a significant cause of morbidity in immunocompromised hosts, including transplant recipients and AIDS patients. The organs affected and the major sites of fungal replication during disseminated candidiasis are the kidney, heart, and brain, and death follows multiple-organ failure (24, 30). The mouse model of acute infection with C. albicans is a valuable experimental model for studying microbial pathogenesis, as it includes many of the clinical features of the human condition (2, 46). In addition, a genetic approach with mice can be used to identify host genes and proteins that regulate the onset of, response to, and ultimate outcome of infection. We previously studied the differential susceptibility of inbred strains A/J and C57BL/6J (B6) to acute infection with C. albicans (29). The A/J mouse strain is exquisitely sensitive to candidiasis, and there are high fungal loads in target tissues and rapid death occurs within 24 h following intravenous injection of 3 × 10⁵ C. albicans blastospores. On the other hand, B6 mice survive up to 2 weeks following infection and ultimately die of kidney failure, a consequence of continued fungal replication at that site. Using the extent of replication in the kidney, brain, and heart, as well as the survival time, as phenotypic markers of susceptibility, we determined that the differential susceptibility of A/J and B6 mice to C. albicans infection segregates as a single gene effect in A/J × B6 F2 mice which corresponds to a loss-of-function mutation in the C5 component of the complement pathway (45).

C5 is a component of the complement pathway, and it plays several critical roles in the innate and adaptive mechanisms of defense against many infections (17). It is cleaved upon activation of the complement cascade, giving rise to C5a and C5b. Whereas C5b participates in the formation of the membrane attack complex, C5a plays a crucial role in initiating and maintaining proinflammatory activity. Both C5a and C3a (a proteolytic product of C3) exhibit potent chemotactic activity for neutrophils, mast cells, and basophils through interactions with specific cell surface receptors expressed on these cells. These interactions trigger the release of additional cytokines by these cells, resulting in amplification of the inflammatory response (10, 18, 50). Other studies with mouse models of C5 deficiency (1, 4) have suggested that C5 may play an additional later role in dampening the inflammatory response to help reduce detrimental effects on the host. In agreement with the known proinflammatory functions of C5 (18, 50), our previous studies with C. albicans-infected A/J and B6 mice revealed that there was granulocyte infiltration in the spleens of C5-sufficient B6 mice, which does not occur in C5-deficient A/J mice. Intriguingly, the C5 deficiency in A/J mice was seemingly not associated with a dampened inflammatory response, as measured by the level of serum cytokines following C. albicans infection. Indeed, the levels of tumor necrosis factor alpha (TNF-α), interleukin-6 (IL-6), monocyte chemotactic protein 1 (MCP-1), MCP-5, and eotaxin were elevated in serum of A/J mice 24 h following infection with C. albicans, and the levels were higher than the levels detected in C5-sufficient B6 mice (29). Therefore, the C5 deficiency in A/J appears to have two effects on the host response to candidiasis, (i) reduced recruitment and infiltration of inflammatory cells in the spleen and (ii) an uncontrolled systemic allergic-type inflammatory response to the infectious stimulus occurring in the background of reduced cellular mobilization.

In both the A/J and B6 strains, as well as in segregating A/J × B6 F2 mice, the high levels of circulating cytokines, C5 deficiency, and mortality following infection with 3 × 10⁵ Candida blastospores are genetically linked. Although at the time of death (24 h postinfection) the fungal load in the heart, kidney, and brain is greater in A/J mice than in B6 mice, at the site of the greatest fungal replication, the A/J kidney, there is not impaired filtration function (as measured by serum blood urea nitrogen levels) and there are not signs of tissue damage or cellular infiltration. On the other hand, pathophysiological analyses of moribund B6 mice 2 weeks following infection with 3 × 10⁵ Candida blastospores revealed that there was massive kidney damage, as shown by a high fungal load, extensive neutrophil infiltration, the presence of necrotic foci, and impaired renal filtration (29). These results suggest that in contrast to the results for C5-sufficient B6 mice, kidney failure is not the cause of death of C. albicans-infected A/J mice.

In the present study, we aimed to identify the physiological processes that are adversely affected by the allergic-type host response and high levels of circulating proinflammatory cytokines detected in C5-deficient A/J mice infected with C. albicans. Since hyperinflammation is known to have pleiotropic effects on many host pathways (6, 43, 44), we wanted to identify which specific host response pathway ultimately causes death when it is dysregulated by C5 deficiency. To distinguish the specific contribution of C5 deficiency from additional genetic background effects fixed in A/J and B6 mice, we studied the recombinant congenic strain BcA17, which harbors the C5 mutation in an otherwise B6 genetic background (16). Our studies revealed that the heart is the organ that is most affected by candidiasis in C5-deficient A/J and BcA17 mice.

MATERIALS AND METHODS

Mice.

Eight- to 12-week-old A/J and C57BL/6J mice were purchased from The Jackson Laboratories (Bar Harbor, ME). The recombinant congenic strain BcA17 was purchased from Emerillon Therapeutics (Montreal, Québec, Canada). Mice were age and sex matched for all experiments. Housing and all experimental procedures were approved by the Biotechnology Research Institute Animal Care Committee, which operated under the guidelines of the Canadian Council of Animal Care.

C. albicans infection.

C. albicans strain SC5314 was grown overnight in YPD medium at 30°C and was harvested by centrifugation. The blastospores were washed twice in phosphate-buffered saline (PBS) and resuspended in PBS at the required density. For experimental infections, A/J and C57BL/6J mice were inoculated via the tail vein with 200 μl of a suspension containing 3 × 10⁵ C. albicans blastospores in PBS. Mice were closely monitored for clinical signs of disease, such as lethargy, loss of appetite, hunched back, and ruffled fur. Mice exhibiting extreme lethargy were considered moribund and were euthanized.

Cytokine detection.

The levels of TNF-α and IL-6 in tissue extracts were determined using “two-site sandwich” assays with commercially available enzyme-linked immunosorbent assay (ELISA) kits (BD Biosciences). Tissue extracts were prepared by homogenization in lysis buffer (20 mM Tris [pH 7.5], 0.3 M NaCl, 2% sodium deoxycholate, 2% Triton X-100, protease inhibitor cocktail [Roche Diagnostics]). Additional determinations of the levels of 32 cytokines and chemokines were carried out by using cardiac tissue extracts and RayBio mouse cytokine array II membranes (RayBiotech Inc., Norcross, GA). Incubation, washing, and detection were performed according to the instructions of the manufacturer. All values were normalized to the protein content, which was determined using the Bradford reagent.

Markers of cardiac failure.

Levels of creatine kinase and cardiac troponin in the circulation were determined using commercially available kits obtained from Pointe Scientific Inc., Canton, Mich., and Life Diagnostics Inc., West Chester, PA (ELISA for cardiac troponin), respectively, according to the instructions of the manufacturer.

Measurement of inducible nitric oxide synthase activity.

Inducible nitric oxide synthase activity in cardiac tissue was measured as described by Szabolcs et al. (38). Briefly, an excised left ventricle was homogenized in ice-cold buffer containing 50 mM Tris-HCl (pH 7.4), 0.05 mM EDTA, 0.5 mM dithiothreitol (DTT), and a protease inhibitor cocktail (Roche Diagnostics), and homogenates were centrifuged at 10,000 × g for 60 min at 4°C. The inducible nitric oxide synthase activities in the supernatants were measured immediately by incubation with 15 mM HEPES (pH 7.4), 2 mM l-arginine, 0.2 mM NADPH, 2 mM magnesium acetate, and 0.5 mM DTT for 20 h at 37°C. The amount of NO₂⁻ produced was determined with the Greis reagent, which contained 2% (wt/vol) sulfanilamide in 5% HCl and 0.1% (wt/vol) N-(1-naphthyl)ethylenediamine dihydrochloride in water. NaNO₂ was used as a standard. Fifty microliters of tissue extract or plasma was mixed with 50 μl of the sulfanilamide solution, and the mixture was incubated in the dark for 10 min. At the end of the incubation, 50 μl of N-(1-naphthyl)ethylenediamine dihydrochloride was added to the mixture, and the preparation was incubated for another 30 min. Then the optical density at 540 nm was determined. The activated partial thromboplastin time was determined using a commercial kit obtained from Biopool International, Ventura, CA, according to the instructions of the manufacturer.

Transcription profiling studies.

Twenty micrograms of total RNA was converted into cDNA using reverse transcriptase (RT) (Super-Script II; Invitrogen, Burlington, ON, Canada) and either Cy5- or Cy3-labeled dCTP (1 mM; Perkin-Elmer, Boston, MA) in a reaction mixture containing 1.5 μl oligo(dT) (100 pmol/μl), 3 μl deoxynucleoside triphosphate (dNTP; 6.67 mM each), 1 μl dCTP (2 mM), 4 μl DTT (100 mM), and 8 μl 5× RT buffer (Invitrogen, Burlington, ON, Canada). The reaction mixtures were incubated at 42°C for 3 h, after which the RNA was degraded by addition of 0.5 μl RNase A (1 μg/μl) and 1.5 μl RNase H (5 U/μl). Labeled cDNA was separated from unincorporated nucleotides (QIAquick PCR purification columns; QIAGEN, Mississauga, ON, Canada) and then concentrated further by evaporation under a vacuum. Labeled cDNA was then used to hybridize mouse 15k v.4 cDNA spotted arrays purchased from the UHN Microarray Facility (Toronto, ON, Canada) containing 15,250 expressed sequence tags (NIA clone set) spotted in duplicate. Briefly, the arrays were prehybridized for 1 to 2 h with DIGEasy hybridization buffer (Roche, Indianapolis, IN) containing 10 μg/ml denatured salmon sperm DNA and 10 μg/ml yeast tRNA. The Cy5- and Cy3-labeled cDNAs were combined and hybridized in the same medium and then incubated with the arrays for 16 to 18 h at 37°C. Finally, the arrays were washed three times for 10 min in 0.1× SSC-0.1% sodium dodecyl sulfate at 50°C and four times for 3 min in 0.1× SSC at room temperature and then dried by centrifugation (1× SSC is 0.15 M NaCl plus 0.015 M sodium citrate). The slides were scanned, digitized images were acquired using a ScanArray Lite instrument (Perkin-Elmer, Boston, MA), and the intensities of individual spots were quantified from 16-bit TIFF images using the QuantArray software package (Perkin-Elmer, Boston, MA). The complete microarray data set can be accessed with NCBI GEO accession number GSE3381.

Microarray data analysis.

The raw data generated by the QuantArray (Perkin-Elmer, Boston, MA) analysis was normalized with the GeneSpring software package (Silicon Genetics, Redwood City, CA) by using the Lowess scatter smoothing algorithm. Genes with reproducible changes in transcript abundance were identified with the “one-class” algorithm in the SAM (Significance Analysis of Microarrays) application (47).

Quantitative RT-PCR.

The levels of Pparδ, Pparα, Lpl, CptIβ, Mcad, Glut1, Glut4, Pdk2, and Pdk4 were determined by a semiquantitative RT-PCR, using a LightCycler (Roche Diagnostics, Laval, Québec, Canada) with the DNA SYBR Green I reaction (Roche Diagnostics). cDNA was made from 1 μg of RNA by using Super-Script II (Gibco-Invitrogen, Burlington, Ontario, Canada) according to the instructions of the manufacturer. The sequences of the primers used in the analysis are shown in Table 1. An s-29 control PCR was performed in parallel in the same LightCycler run. Quantitation was performed by comparison of crossing points. Thus, the RNA level for gene X was the crossing point for X/crossing point for s-29.

TABLE 1.

Primers

Gene	Forward primer	Reverse primer
Pparδ	GAACACACGCTTCCTTCCAG	CCGACATTCCATGTTGAGG
Pparα	TCGGCGAACTATTCGGCTG	GCACTTGTGAAAACGGCAGT
Lpl	GGATGAGCGACTCCTACTT	TCCCTAGCACAGAAGATGACCT
CptIβ	GTACCGCCTAGCCATGACA	GGCTCCAGGGTTCAGAAAGT
Mcad	CCGAAGAGTTGGCGTATGGG	GGGCTCTGTCACACAGTAAGC
Glut1	GTATCCTGTTGCCCTTCTGC	TCGAAGCTTCTTCAGCACAC
Glut4	CTGCTGCCCTTCTGTCCT	CGGTCAGGGGCTTTAGACT
Pdk2	CACCGGACTCTAAGCCAGTT	ACGGGGTCATCTCCATAGGT
Pdk4	CGTCTTGGGAAAAGAAGACCT	TGTAACTAAAGAGGCGGTCAGTAA
s-29	GTCTGATCCGCAAATACGGG	AGCCTATGTCCTTCGCGTACT

The levels of cardiac triglyceride and glucose in the circulation were determined by using commercially available kits (Sigma).

Serum transfer.

BcA17 mice were inoculated intraperitoneally with 350 μl of B6 or BcA17 serum just prior to intravenous administration of 3 × 10⁵ C. albicans blastospores in 200 μl PBS. The sera from equal numbers of male and female B6 or BcA17 mice were pooled. Three hundred fifty microliters of serum should bring the C5 levels to normal values, with a half-time of 21 h (49). Animals were euthanized when mice that were infected with C. albicans and were not treated with serum were moribund. The levels of cardiac markers (creatine kinase and cardiac troponin) and glucose in the circulation were determined as described above.

Statistical analysis.

The statistical significance of the differences between the various experimental and control samples was assessed using a two-tailed t test.

RESULTS

Cytokine levels in infected tissues.

We hypothesized that the high levels of circulating proinflammatory cytokines detected in A/J serum 24 h after infection with C. albicans must have originated at least in part from a focus of fungal replication in the mice. Therefore, we measured the levels of the key inflammatory cytokines, IL-6 and TNF-α, in tissue extracts from the kidney, brain, and heart, the three principal sites of fungal replication in the mouse model of systemic infection. Mice were inoculated intravenously with 3 × 10⁵ C. albicans blastospores. Twenty-four hours later the A/J mice were extremely lethargic and had ruffled fur and a hunched back, whereas the B6 mice exhibited no clinical signs. At this point, these mice, together with noninfected controls, were euthanized, the organs were harvested and homogenized, and the IL-6 and TNF-α levels were determined by ELISA (Fig. 1). Although C. albicans infection stimulated TNF-α production in A/J mice in all three organs tested, the most striking induction was seen in the heart (P < 0.005). In addition, the robust TNF-α response in A/J hearts was completely absent from B6 mice (P < 0.005), in which only modest induction in the brain was seen (Fig. 1A). Even more pronounced differences were noted for IL-6; A/J hearts exhibited dramatic levels of induction of IL-6 expression compared to other infected tissues from the same strain or from C5-sufficient B6 mice. These results indicate that there is a unique heart-specific IL-6 and TNF-α response in A/J mice 24 h following systemic C. albicans infection. After these results were obtained, a broader evaluation of cytokine expression in the hearts of C. albicans-infected A/J mice was performed. The levels of expression of 32 cytokines and chemokines in C. albicans-infected B6 and A/J hearts were determined using a commercially available cytokine array (www.Raybiotech.com). The results were quantified by densitometry and are shown in Fig. 2. 6Ckine, MCP-5, IL-12-p70, IL-9, and MIP-1α were expressed at comparable levels, but in the A/J heart tissue there were much higher levels of the remaining 27 cytokines. Thus, C. albicans infection in C5-deficient A/J mice results in pleiotropic cytokine and chemokine production in the heart muscle, which indicates that there is a strong inflammatory response in this tissue.

FIG. 1.

TNF-α and IL-6 induction in A/J and B6 mice following C. albicans infection. TNF-α (A) and IL-6 (B) levels were determined in cytoplasmic extracts of the tissues from six C. albicans-infected A/J mice (A/Jinf.), six C. albicans-infected B6 mice (B6inf.), six uninfected control A/J mice (A/J), and six uninfected control B6 mice (B6). The data were normalized to the protein content of the extract. The data represent data from at least two independent experiments. The error bars indicate the standard errors of the means. Statistically significant differences are indicated by three asterisks (P < 0.005).

FIG. 2.

Cytokine and chemokine levels in C. albicans-infected A/J and B6 mice. Cytokine levels were determined in cytoplasmic extracts of cardiac tissue from C. albicans-infected A/J and B6 mice using a commercially available cytokine array. Quantitation was performed by densitometric scanning and normalization using the protein content of extracts. G-CSF, granulocyte colony-stimulating factor; GM-CSF, granulocyte-macrophage colony-stimulating factor; IFN-gamma, gamma interferon; CTACK, cutaneous T-cell-attracting chemokine; MCP-1, monocyte chemoattractant protein 1; MCP-1-alpha, macrophage inflammatory protein 1 alpha; RANTES, regulated upon activation, normal T-cell expressed and presumably secreted; SCF, Sertoli cell factor; sTNF RI, soluble tumor necrosis factor receptor I; TARC, thymus and activation-regulated chemokine; TIMP-1, tissue inhibitor of metalloproteinases 1; TPO thrombopoietin; VEGF, vasular endothelial growth factor; IL, interleukin.

Histological evaluation of the cardiac tissue.

To further investigate the A/J-specific cytokine response in the heart, A/J mice were infected with Candida, and 24 h later histological sections of formalin-fixed hearts and kidneys were stained with periodic acid-Schiff (PAS) reagent (Fig. 3A). Microscopic examination revealed that fungal replication was associated with foci of PAS-positive C. albicans mycelial strands in the ventricular myocardium of A/J mice. Granulocyte infiltration (Fig. 3A) was evident at such Candida foci. Although fungal infection was not evident in the atrium, it was evident at the origins of the atrioventricular valves, suggesting that there was myocardial deficiency and also possible valvular deficiency (data not shown). Similar analysis of the B6 heart tissue did not reveal Candida foci, overt infiltration, or tissue damage (data not shown). On the other hand, PAS staining of kidney sections from Candida-infected A/J mice revealed dense fungal growth but no cellular infiltration and tissue damage. These results showed that the unique cytokine response detected in the Candida-infected A/J hearts was associated with in situ fungal replication and cellular infiltration which was specific to this organ (and not seen in kidneys). The histological changes in Candida-infected A/J hearts were linked to elevated levels of the two cardiac proteins, creatine kinase (P < 0.005) and cardiac troponin I (P < 0.05), in the serum of these mice (Fig. 3B). These changes were induced by infection and were not seen in control uninfected A/J mice; Candida-induced leakage of creatine kinase and cardiac troponin I from the A/J heart muscle was a direct result of loss of C5 function, as these changes were not detected in wild-type B6 mice but were present in BcA17 recombinant congenic mice that carried the C5 mutation from A/J introgressed in a B6 background (P < 0.005) (Fig. 3B).

FIG. 3.

Effect of C5 deficiency on the integrity of the cardiac muscle. (A) Histological evaluation of C. albicans-infected heart and kidney. Formalin-fixed tissues were sectioned and stained with periodic acid-Schiff reagent using standard procedures. Magnification, ×20. F, Candida hyphae; sCM, striated cardiac muscle; wbc, granulocytic infiltration. (B) Markers of necrotic damage of the cardiac tissue. Creatine kinase and cardiac troponin I levels were determined in the sera of Candida-infected A/J, B6, and BcA17 mice (A/Jinf., B6inf., and BcA17inf., respectively) and in the sera of noninfected controls (A/J, B6, and BcA17). The data represent data from at least two independent experiments. The numbers in parentheses indicate the sample size for each experimental group. The error bars indicate the standard errors of the means. Statistical significance is indicated by asterisks (one asterisk, P < 0.05; three asterisks, P < 0.005).

Together, these results suggest that C5 deficiency in the A/J heart impairs host resistance to Candida, resulting in fungal replication, cellular infiltration, and consequential tissue damage. Interestingly, this altered response is specific to the heart and is not seen in the kidney, another major site of fungal replication in A/J mice.

Transcriptional profiling studies.

In order to obtain further insight into the molecular basis of the unique effect of C5 deficiency on the heart response to Candida infection, transcriptional profiling using cDNA microarrays was performed with the hearts of infected BcA17 and B6 mice. Briefly, RNA samples from infected heart tissue of four susceptible (BcA17) mice and four resistant (B6) mice were converted to Cy5- or Cy3-labeled cDNA and hybridized directly to cDNA microarrays. To ensure that potential strain-specific differences in transcript levels were the result of responses to infection and not intrinsic to the hearts of the two strains, control hybridizations between RNAs from uninfected BcA17 and B6 hearts were performed. In these BcA17-versus-B6 controls, very few significant differences in transcript profiles were observed (data not shown). Differences in transcript levels detected in hearts of C. albicans-infected BcA17 and B6 mice were identified using the SAM statistical package set; the false discovery rate was 0.8%, and the fold change was >1.5 (significant gene lists). Examination of significant gene lists revealed that up-regulation of several genes was associated with the inflammatory response in the BcA17 mice compared to the B6 mice (Table 2 and Fig. 4). These included genes such as Socs3, Hmox1, Ctla2a, and Gro-1, the gene coding for the KC chemokine. There was a striking difference (BcA17 > B6) in the expression of genes encoding metalloproteases, such as a disintegrin and metalloproteinase-8 (Adam8), a disintegrin, and a metalloproteinase with thrombospondin motifs 1 and 4 (Adamts1 and -4), as well as genes involved in fibrosis and modeling of the extracellular matrix (Spp1, Nppb, Sdc1, and Cspg2). Finally, reduced expression of mRNAs coding for cardioprotective molecules, including selenoprotein P (Sepp1), sestrin 1 (Sesn1), connexin 43 (Gja1), heat shock proteins (Hspd1, Hspa5, and Serpinh1), and targets of the hypoxia-inducible factor (ferrochelatase [Fech] and EGL nine homologue 1 [Egln1]), was observed in BcA17 hearts compared to B6 hearts.

TABLE 2.

Microarray analysis of gene expression in C. albicans-infected B6 and BcA17 hearts

Gene	BcA17/B6 ratio	B6/BcA17 ratio	Description
Inflammation
Gro1	23.745		GRO1 oncogene (Gro1), mRNA
Idb2	3.481		Id2 protein (Id-2) mRNA, 3′ end
Socs3, Cish3	3.470		Cytokine-inducible SH2-containing protein 3 (Cish3), mRNA
Hmox1	3.382		AC005290, complete sequence (Mus musculus)
S100a9	3.310		S100A9 gene for S100A9 protein exons 1 to 3
Hdc	3.073		l-Histidine decarboxylase
Gp49a	2.869		gp49A gene, complete cds
Adamts1	2.719		ADAMTS-1, complete cds
Ctla2b	2.441		ctla-2-beta mRNA, homolog of cysteine protease proregion
Adam8	2.307		Disintegrin and metalloprotease domain 8 (Adam8), mRNA
Pde4b	2.184		3′,5′-cyclic AMP phosphodiesterase mRNA, 3′ end
Egr1	2.154		Early growth response 1 (Egr1), mRNA
Ctla2a	2.142		Cytotoxic T-lymphocyte-associated protein 2 alpha (Ctla2a)
Mt2	2.126		Metallothionein II (MT-II) gene, disintegrin-like and metalloprotease (reprolysin type) with thrombospondin type 1
Adamts4	2.086		Motif 4 (ADAMTS4), mRNA
Mlp	1.910		MARCKS-like protein (Mlp), mRNA
Ifitm3	1.839		Interferon-induced transmembrane protein 3
Mlp	1.789		MARCKS-like protein (Mlp), mRNA
Clec4d	1.668		Macrophage C-type lectin (Mpcl), mRNA
Cardiac fibrosis
Spp1	4.776		Secreted phosphoprotein 1
Nppb	3.550		BNP gene for brain natriuretic peptide
Sdc1	1.797		Mus musculus syndecan 1 (Sdc1), mRNA
Cspg2	1.517		Chondroitin sulfate proteoglycan 2 (versican), mRNA
Apoptosis
Gadd45g	3.811		Growth arrest and DNA damage inducible, gamma (Gadd45g), mRNA
Ier3	2.607		gly96 mRNA
Myc	2.059		Normal c-myc gene and translocated homologue from J558 plasmocytoma cells
Itgb3bp	1.834		Integrin beta 3 binding protein (beta3-endonexin)
Btg1	1.573		B-cell translocation gene 1 protein (BTG1) mRNA, complete cds
Mcl1	1.412		Myeloid cell leukemia sequence 1 (Mcl1), mRNA
Metabolism
Odc1	1.906		Kidney ornithine decarboxylase mRNA, clone pODC16, 3′ end
Srm	1.736		Spermidine synthase gene
Sgk	1.647		Serum and glucocorticoid-dependent protein kinase (Sgk) mRNA
Oxidative stress
Glrx1	1.865		Glutaredoxin, complete cds
Gsr	1.525		Glutathione reductase
Atf4	1.577		mATF4 (mTR67) mRNA, complete cds
Translation/transcription
Rbm3	4.301		RNA binding motif protein 3 (Rbm3), mRNA
Rbm18	2.532		RNA binding motif protein 18
Rrs1	2.339		RRS1 ribosome biogenesis regulator homolog (Saccharomyces cerevisiae)
Rab20	2.216		rab20 mRNA
Xpo4	2.091		Exportin 4 (Xpo4-pending), mRNA
H3f3b	2.016		H3 histone, family 3B (H3f3b), mRNA
H3f3b	1.880		H3 histone, family 3B (H3f3b), mRNA
Sfrs5	1.850		HRS gene, complete cds
Ssbp2	1.682
Snrpd1	1.556		Small nuclear ribonucleoprotein D1 (Snrpd1), mRNA
Growth factor binding
Fin14	1.863		Fibroblast growth factor inducible 14 (Fin14), mRNA
Fin14	1.745		Fibroblast growth factor inducible 14 (Fin14), mRNA
Igfbp3	1.744		Insulin-like growth factor-binding protein (IGF-BP3) (Igfbp3), mRNA
Other
2210403K04Rik	3.325
C79468	3.004
Nr5a1	2.988		ELP3, complete cds
3930401B19Rik	2.950
1200015M12Rik	2.707
1200015M12Rik	2.428
C86813	2.054		DKFZp564C2063 (from clone DKFZp564C2063)
Rnf149	1.982		Ring finger protein 149
AU018180	1.897
Cap1	1.893		Adenylyl cyclase-associated protein (CAP), mRNA
AU044980	1.815
D1Ertd218e	1.762
	1.732		DKFZp434J214 (from clone DKFZp434J214), partial cds
D1Ertd692e	1.607		FLJ23280 fis, clone HEP07194
Ifrd1	1.540		Interferon-related developmental regulator 1 (Ifrd1), mRNA
Basp1	1.520		Brain abundant, membrane-attached signal protein 1
Cardiac development
Gja1		2.542	Connexin 43
Mglap		2.288	Matrix gamma-carboxyglutamate (gla) protein
Zfhx1b		2.250	Zinc finger homeobox 1b
Calr		1.890	Calreticulin (Calr), mRNA
Nrp1		1.836	Neuropilin (Nrp), mRNA
Fibrotic response
Col3a1		3.787	COL3A1 gene for collagen alpha-I
Col1a2		1.904	Collagen, type I, alpha 2 (COL1A2), mRNA
Sparcl1		1.820	SPARC-like 1 (mast9, hevin), extracellular matrix protein 2 (Ecm2), mRNA
Inflammation
Tra1		3.532	Tumor rejection antigen gp96 (Tra1), mRNA
Tra1 T Tra1		3.354	Tumor rejection antigen gp96 (Tra1), mRNA
Lyzs		1.856	Lysozyme M gene, exon 4
Cd81		1.641	Kcnq1, Ltrpc5, Mash2, Tapa-1, Tssc4, and Tssc6 genes
Ghitm		1.517	Growth hormone-inducible transmembrane protein
Metabolism
Sdh1		1.989	Sorbitol dehydrogenase precursor mRNA, partial cds
Sdha		1.851	Succinate dehydrogenase Fp subunit mRNA, partial cds
Acat1		1.744	Acetyl-CoA acetyltransferase 1
Ak3		1.692	Adenylate kinase 3
Acadm		1.619	Medium-chain acyl-CoA dehydrogenase (Acadm) mRNA
Mat2b		1.589	Methionine adenosyltransferase II, beta
Idh3a		1.546	Isocitrate dehydrogenase 3 (NAD+) alpha (IDH3A), mRNA
Bcat2		1.515	Branched-chain aminotransferase 2, mitochondrial
Stress or hypoxia induced
Hspa5		3.328	BiP
Serpinh1		2.876	47-kDa heat shock protein (HSP47)
Hspa5		2.590	70-kDa heat shock protein 5 (Hspa5), mRNA
Txndc7		2.298	Thioredoxin domain containing 7, mRNA
Sesn1		1.891	Sestrin 1, mRNA
Hspd1		1.790	Heat shock protein 1 (chaperonin)
Sepp1		1.733	Mus musculus selenoprotein P, plasma 1 (Sepp1), mRNA
Egln1		1.566	EGL nine homolog 1 (Caenorhabditis elegans)
Hspd1		1.544	Heat shock protein 1 (chaperonin)
Fech		1.514	Ferrochelatase (Fech), mRNA
Cytoskeleton
Actn4		3.205	Actinin alpha 4
Sorbs1		2.209	Sorbin and SH3 domain containing 1
Lmo7		2.014	LIM domain only 7, mRNA
Ivns1abp		2.002	Influenza virus NS1A binding protein
Dag1		1.639	Dystroglycan 1 (DAG1) gene, exons 1 and 2 and complete cds
Other
C85340		3.020
		2.701	Ribosomal protein S6 (Rps6), mRNA
6720475J19Rik		2.451	RIKEN
Armet		2.132	Arginine-rich, mutated in early-stage tumors
AA407887		1.845
Zwint		1.826	ZW10 interactor
Nedd4a		1.696	Neural precursor cell expressed, developmentally down-regulated gene 4
Ptp4a3		1.625	Protein tyrosine phosphatase 4a3 (Ptp4a3), mRNA
		1.527	DKFZp586L0518 (from clone DKFZp586L0518)

FIG. 4.

Comparison of transcript profiles of control and C. albicans-infected BcA17 and B6 hearts. Transcript profiles from C. albicans-infected BcA17 and B6 hearts were compared by microarray analysis. The four columns of green (higher in BcA17 mice than in B6 mice) and red (higher in B6 mice than in BcA17 mice) represent the four samples used for the analysis, while each row represents a gene whose level is being compared and whose designation is indicated and color coded based on function (red, inflammation, apoptosis, and immune response; blue, extracellular matrix and tissue remodeling; pink, hypoxia-induced stress and cardioprotective; green, translation and transcription). The genes shown are a subset of the genes that were identified by SAM (false discovery rate, 0.8%; fold change, >1.5) as genes that were significantly different in the two strains of mice. A similar analysis of uninfected control BcA17 and B6 mice (three mice each) revealed no significant strain-dependent differences (black).

These differences in transcript profiles are indicative of the increased inflammatory response in BcA17 mice. This response appears to be associated with cardiomyopathy and a concomitant fibrotic response and remodeling of the extracellular matrix. At the same time, the noticeable lack of expression of cardioprotective molecules in BcA17 mice suggests that there was a failure to adapt to the debilitating effects of inflammatory activity. Finally, the differences in transcript profiles were the result of a direct response to Candida infection (Fig. 4) and were not observed for parallel comparisons of RNAs from control uninfected BcA17 and B6 hearts (Fig. 4). The complete microarray data set can be accessed with NCBI GEO accession number GSE3381.

Biochemical studies.

Additional biochemical studies were performed to investigate the cellular processes that are impaired by C5 deficiency and that are required for a protective host response to C. albicans infection in the heart. Since several of the cytokines that are differentially expressed in the hearts of permissive mice (A/J and BcA17 mice) and nonpermissive mice (B6 mice) (Fig. 1 and 2) can modulate NO production (20, 23, 35) and consequently cardiac function, NO production in mice was monitored prior to and following C. albicans infection. No differences were detected in cardiac NO production in either uninfected hearts (A/J mice, 3 ± 0.25 nmol NO₂/mg tissue; B6 mice, 2.96 ± 0.16 nmol NO₂/mg tissue; n = 4) or C. albicans-infected hearts (A/J mice, 3.48 ± 0.23 nmol NO₂/mg tissue; B6 mice, 3.12 ± 0.17 nmol NO₂/mg tissue; n = 18). Similar NO levels were also found in the circulation (A/J mice, 0.66 ± 0.5 μM NO₂ [n = 10]; B6 mice, 0.73 ± 0.3 μM NO₂ [n = 14]), and the levels of inducible nitric oxide synthase in cardiac extracts were also similar in the three strains (A/J mice, 3.76 ± 0.35 nmol NO₂/mg tissue/20 h; B6 mice, 3.81 ± 0.35 nmol NO₂/mg tissue/20 h; BcA17 mice, 3.51 ± 0.60 nmol NO₂/mg tissue/20 h; n = 6). High-level circulating cytokines can trigger disseminated intravascular coagulation, leading to formation of microthrombi, obstruction of the blood flow, and anoxic cell death (9, 11). Consequently, the effect of C5 deficiency on activated partial thromboplastin time was determined. The efficacies of coagulation were found to be similar in uninfected mice (A/J mice, 38.5 ± 3.75 s; B6 mice, 42.4 ± 1.97 s) and in C. albicans-infected A/J and B6 mice (A/J mice, 56 ± 3.35 s; B6 mice, 49.82 ± 3.44 s). Thus, dysregulated inflammation in the hearts of C5-deficient strains had no effect on the local or systemic NO levels and did not affect the blood flow or time of coagulation.

Data in Table 2 suggest that there are differences in the metabolic activities of the cardiac muscle of Candida-infected BcA17 and B6 mice (higher levels of Sdh1, Sdha, Acat1, Ak3, Acadm, Mat2b, Idh3a, and Bcat2 in B6 mice than in BcA17 mice), including the levels of genes associated with lipid and glucose metabolism. Therefore, we examined the levels of molecules implicated in fatty acid oxidation in the hearts of B6 and BcA17 mice. Pparα and Pparδ (3, 7, 14) are key regulators of cardiac lipid and carbohydrate metabolism. Figure 5A shows that C. albicans infection did not affect the Pparδ transcript levels in either B6 or BcA17 mice (Fig. 5A and C). On the other hand, while there was only a modest (20%) reduction in the level of Pparα transcripts in the hearts of infected B6 mice, a dramatic fivefold decrease in Pparα mRNA was detected in BcA17 mice (Fig. 5C) (P < 0.001). The transcriptional targets of Pparα include lipoprotein lipase (Lpl), fatty acid transport protein (Fatp), fatty acid translocase (Fat/CD36), and heart-type fatty acid binding protein (H-Fabp), all of which are involved in fatty acid transport into the cell and within the cytoplasm. Cytoplasmic fatty acids are then esterified to fatty acyl-coenzyme A (fatty acyl-CoA) by fatty acyl-CoA synthetase (Facs). Transport across the mitochondrial membrane requires transesterification by carnitine palitoyltransferase I (CptI), and CptIβ is the isoform that is the predominant form in the adult heart (5). Oxidation takes place in the mitochondria by a family of acyl-CoA dehydrogenases specific for very-long-chain (Vlcad), long-chain (Lcad), medium-chain (Mcad), and short-chain (Csad) molecules (48). Consistent with the reduction in Pparα transcript levels following infection (Fig. 5A), the levels of Lpl, CptIβ, and Mcad mRNAs were also decreased in the hearts of C. albicans-infected BcA17 mice (P < 0.001) and B6 mice (P < 0.001 for Mcad and CptIβ; P < 0.05 for Pparα and Lpl); however, the decreases were much more pronounced in BcA17 mice than in B6 mice. Finally, decreases in the Pparα, Lpl, Mcad, and CptIβ mRNA levels were linked to net accumulation of triglycerides in the hearts of C5-deficient A/J mice but not in the hearts of B6 mice (Fig. 5D).

FIG. 5.

Fatty acid oxidation in C. albicans-infected A/J and B6 mice. The transcript levels of Pparδ, Pparα, Lpl, CptIβ, and Mcad mRNAs in C. albicans-infected B6 hearts (B6inf.) (A), C. albicans-infected A/J hearts (A/Jinf.) (B), or C. albicans-infected BcA17 hearts (BcA17inf.) (C) were compared to the levels in the corresponding control uninfected hearts using quantitative PCR. The difference in crossing points between the infected and uninfected samples was calculated for a gene of interest and normalized to differences in the corresponding crossing points of an internal control, s-29. The data rrepresent data from at least two independent experiments. Each experimental group consisted of six males and six females. (D) Levels of triglyceride in hearts of Candida-infected A/J mice (A/Jinf.) and Candida-infected B6 mice (B6inf.) and uninfected controls. The error bars indicate the standard errors of the means. Statistical significance is indicated by asterisks (one asterisk, P < 0.05; three asterisks, P < 0.005).

Glucose metabolism.

In the heart, lipid metabolism and glucose metabolism are tightly coupled to respond efficiently to changing energy needs and variations in nutrient availability in the cardiac muscle. Under depressed fatty acid oxidation conditions, the heart muscle switches to glucose utilization (31) for energy production. To determine how changes in fatty acid and glucose metabolism may have affected glucose levels in the circulation, sera from healthy and C. albicans-infected A/J and B6 mice were analyzed. As shown in Fig. 6D, Candida-infected C5-deficient A/J and BcA17 mice became hypoglycemic, and there was a 70% reduction in the glucose levels compared to their healthy counterparts (P < 0.005). Glucose metabolism is strictly regulated by glucose transport across the membrane by Glut1 and Glut4. After infection (Fig. 6A and C), all three strains up-regulated Glut1 transcripts and down-regulated Glut4 transcripts in a similar fashion. However, C5 deficiency was associated with a dramatic up-regulation of the level of mRNA for pyruvate dehydrogenase kinase 4 (Pdk4) (P < 0.005) (Fig. 6C and D), a potent inhibitor of the pyruvate dehydrogenase complex and thus glucose oxidation (54). Therefore, increased glucose uptake by the cardiac muscle of C5-deficient mice would be unlikely to result in productive glucose oxidation.

FIG. 6.

Glucose oxidation in C. albicans-infected A/J and B6 mice. The transcript levels of Glut1, Glut4, Pdk2, and Pdk4 mRNAs in C. albicans-infected B6 hearts (B6inf.) (A), C. albicans-infected A/J hearts (A/Jinf.) (B), or C. albicans-infected BcA17 hearts (BcA17inf.) (C) were compared to the levels in the corresponding control uninfected hearts using quantitative PCR. The difference in crossing points between the infected and uninfected samples was calculated for a gene of interest and normalized to differences in the corresponding crossing points of an internal control, s-29. The data represent data from at least two independent experiments. Each experimental group consisted of six males and six females. (D) Glucose concentrations in the sera of Candida-infected A/J mice (A/Jinf.), Candida-infected B6 mice (B6inf.), and Candida-infected BcA17 mice (BcA17inf.) and uninfected controls. The numbers in parentheses indicate the number of mice analyzed in each experimental group. The error bars indicate the standard errors of the means. Statistical significance is indicated by three asterisks (P < 0.005).

Serum transfer.

Having identified the pathogenic events that led to morbidity in C5-deficient BcA17 and A/J mice following C. albicans infection, we determined whether the mice could be protected from these events by treatment with serum from a C5-sufficient strain. To do this, six BcA17 mice (three males and three females) were each inoculated intraperitoneally with either 350 μl of B6 (C5-sufficient) or BcA17 (C5-deficient) serum just prior to intravenous administration of C. albicans. One group of six BcA17 mice was inoculated with only C. albicans, and another group served as a control for serum injection and was inoculated with B6 serum alone. All animals were euthanized when the mice inoculated with only C. albicans were moribund. At that point, the only other infected experimental group in which animals were succumbing to infection was the group treated with C5-deficient serum. No mortality was seen in any of the other groups. These observations indicated that adaptive transfer of C5 from B6 serum could protect BcA17 mice against lethal C. albicans infection. Figure 7 shows the protective effect of administration of B6 serum on necrotic damage to the hearts of BcA17 mice. As expected, there were increases in the circulating levels of creatine kinase and cardiac troponin in BcA17 mice infected with C. albicans with or without pretreatment with C5-deficient serum. These increases were not seen in BcA17 mice pretreated with C5-sufficient serum (Fig. 7). Circulating glucose levels were also restored in BcA17 females treated with B6 serum, but this effect was not seen in males. Nevertheless, these results demonstrate that the major pathological consequences of C. albicans infection in C5-deficient mice can be corrected by passive transfer of C5.

FIG. 7.

Effect of pretreatment with serum from C5-sufficient B6 mice on C. albicans infection in C5-deficient mice. The levels of creatine kinase, cardiac troponin, and glucose were determined in the sera of six (three male and three female) uninfected BcA17 mice (BcA17), six C. albicans-infected BcA17 mice (BcA17inf), six mice that were infected with C. albicans but were pretreated with B6 serum (BcA17inf + S_B6), six mice that were infected with C. albicans but were pretreated with BcA17 serum (BcA17inf + S_BcA17), and six uninfected BcA17 mice treated with B6 serum (BcA17 + S_B6). The error bars indicate the standard errors of the means. Statistical significance is indicated by three asterisks (P < 0.005).

DISCUSSION

An early response to infection is a crucial component of a host's ability to control the growth of the invading microorganisms and eradicate them. In particular, proinflammatory cytokines, such as TNF-α, form part of an early detection system, and they provide immediate signals that allow the host to discriminate between self and nonself. The recognition of invading pathogens triggers a series of signal transduction cascades and associated inflammatory and early immune responses. Under normal conditions, these responses are strictly regulated to produce optimal antimicrobial effects while minimizing collateral damage to the host. However, under certain conditions, either due to the strength of the danger signal or due to failure of negative regulatory mechanisms, the inflammatory response overwhelms the host and causes significant damage, including morbidity. The studies of the effect of the C5 deficiency in A/J and congenic BcA17 mice on the pathogenesis of acute infection with C. albicans provide a striking example of a deleterious inflammatory response and highlight a previously suspected but poorly understood fundamental role of C5 in dampening this process (29). The key role of the inflammatory response in generating clinical symptoms was revealed in a study performed by Fidel et al. (13), who studied vulvovaginal candidiasis. These authors reported that it was not the fungal load per se, but the presence of polymorphonuclear leukocyte infiltration, that determined whether a patient was symptomatic.

In the systemic candidiasis model that we used, fungal replication takes place in a number of organs, including the kidney, brain, and heart, and C5 deficiency is associated with increased replication at all these sites, with the kidney having the largest fungal loads (29). Histological and biochemical analyses have indicated that it is the heart and not the kidney that suffers the most damage following Candida infection in C5-deficient mice (Fig. 3) (29). The damage does not appear to be a direct consequence of fungal replication per se since the fungal load is considerably higher in the kidney, which nevertheless remains histologically and functionally intact. In our investigation, therefore, we focused on the molecular mechanisms underlying cardiomyopathy and organ failure in C5-deficient hearts following Candida infection. A clue to the aberrant response came from microscopic evaluation of C. albicans-infected A/J hearts and kidneys. Significant cellular infiltration in the vicinity of Candida hyphae was detected in the heart, while no such infiltration was detected in the kidney despite a much greater prevalence of fungal colonies. Although the lack of granulocyte recruitment to the C5-deficient A/J kidney can be accounted for by the known chemoattractant function of C5a (51), leukocyte infiltration in Candida-infected A/J hearts in the absence of functional C5 was unexpected. These results suggest that there is another source of chemoattractant in the C5-deficient heart, and we propose that cytokines and/or chemokines may be secreted by the infected muscle or vascular (endothelial) tissue itself. In this scenario, in the early stages of infection, when C5a-dependent pathways are normally active, C5-sufficient C57BL/6J tissues respond to the infectious stimulus, control infection, and have small fungal loads. By contrast, C5-deficient mice (A/J and BcA17 mice) do not mount such a response, and fungal replication in the heart triggers an intrinsic compensatory cytokine response. Cardiac tissue, in which virtually all cell types can participate in proinflammatory activity (28), responds to relatively minor stress by activation of cytokines and chemokines. This response results in cellular infiltration, resulting in global increases in in situ production of many chemokines and cytokines. Some chemokines, such as MIP-2, KC, and granulocyte colony-stimulating factor, are known for their ability to recruit granulocytes. Similarly, IL-6 and MCP-1 are expected to attract mononuclear leukocytes, further increasing the inflammatory activity in the heart. This exacerbated cytokine-chemokine response is not seen in the control C5-sufficient mice and is the initial insult, which is followed by pleiotropic consequences for metabolism of the cardiac muscle.

The expression of proinflammatory cytokines in the heart is a warning signal which triggers various protective mechanisms that help the tissue regain homeostasis once the initial stimulus is eliminated (27, 53). A second conclusion of our study is that expression of protective mechanisms, including stress and the hypoxic response program, is impaired in C5-deficient mice (A/J and BcA17 mice). For instance, expression of several hypoxia-inducible factor 1α targets, including ferrochelatase (26) (1.5×), EGL nine homologue 1 (8) (1.5×), and adenylate kinase 3 (8) (1.7×), is more robust in B6 Candida-infected hearts than in BcA17 Candida-infected hearts, and this supports the hypothesis that there is a more significant hypoxia-induced stress response in the C5-sufficient strain. This response is required to maintain metabolic homeostasis by regulating the expression of genes participating in glycolysis and glucose transport, Glut-1/3 (8). In addition, hypoxia-inducible factor 1α activation is implicated in a number of cardiac functions, including vascularity, energy generation, calcium homeostasis, and contractility (21). The intensity of the stress response (12, 19, 52) is also different in C5-deficient and C5-sufficient hearts infected with Candida, as shown by differential expression of Hsp47 (36) (2.8×). The expression of Hsp47 has been linked to procollagen processing and secretion during myocardial infarction (41). Likewise, Hspa5 (also known as Grp78 or Bip), a member of the Hsp70 family, is expressed at threefold-higher levels in the B6 heart than that in the BcA17 heart. The roles of several other Hsps in cardioprotection by interference with apoptosis or protection of cytoskeletal integrity are well recognized (25, 34). Another cardioprotective protein expressed at higher levels in B6 mice (3.9×) than in BcA17 mice is connexin 43, a protein that plays an essential role in ischemic preconditioning (37).

The up-regulation of a number of corrective measures in the Candida-infected B6 hearts is consistent with higher-level metabolic activity (genes from the microarray and the enzymes involved in fatty acid oxidation and glucose oxidation). The C5-deficient heart, on the other hand, appears to be unable to activate adequate protective mechanisms in spite of elevated levels of cytokines that provide the danger signal and therefore should trigger these responses. Clearly, key physiological processes affected by inflammation, such as metabolism and energy generation, continue to deteriorate in the absence of adaptive changes. Using a quantitative PCR analysis, we detected down-regulation of the levels of a number of enzymes involved in fatty acid and glucose oxidation. Consistent with the relative decrease in the potential to oxidize fatty acids, we observed accumulation of triglycerides only in the Candida-infected A/J and BcA17 hearts. Accumulation of intracardiac lipids is known to cause contractile dysfunction (42). Other conditions that favor lipid accumulation, such as obesity and diabetes (15, 32, 55), as well as genetic defects in fatty acid oxidation (22), are linked to cardiac failure, suggesting that the imbalance between fatty acid uptake and utilization contributes to tissue damage in the A/J heart. Although the cardiac tissue can use either fatty acids, glucose, or lactate as a substrate, fatty acids are the fuel of choice since they produce the most energy per molecule (39, 40). With a decrease in fatty acid oxidation and a consequent decrease in acetyl-CoA production, gluconeogenesis is compromised, and hypoglycemia occurs (33). Thus, a failure to meet energy requirements due to a depressed rate of fatty acid oxidation and an inability to compensate with glucose metabolism leads to cardiac failure in C5-deficient mice after C. albicans infection.

Finally, we were able to correct all pathological effects of the C5 deficiency in BcA17 mice by passive transfer of C5-competent serum from B6 mice.

In conclusion, in this paper we describe a model of infection in which inflammation-induced down-regulation of cardiac metabolism is not compensated for and therefore results in cardiac failure in BcA17 mice. C5-sufficient B6 mice mount a protective stress response that does not occur in the C5-deficient mice and that appears to induce homeostatic metabolic pathways. Identification of the adaptive pathways that are not present in Candida-infected A/J and BcA17 mice is important for preventing cardiac failure linked to stress, such as infection. Here we describe an ideal model with which to dissect the underlying molecular mechanisms.