Genomic Variation across a Clinical Cryptococcus Population Linked to Disease Outcome

Abstract

Cryptococcus neoformans is the causative agent of cryptococcosis, a disease with poor patient outcomes that accounts for approximately 180,000 deaths each year. Patient outcomes may be impacted by the underlying genetics of the infecting isolate; however, our current understanding of how genetic diversity contributes to clinical outcomes is limited. Here, we leverage clinical, in vitro growth and genomic data for 284 C. neoformans isolates to identify clinically relevant pathogen variants within a population of clinical isolates from patients with human immunodeficiency virus (HIV)-associated cryptococcosis in Malawi. Through a genome-wide association study (GWAS) approach, we identify variants associated with the fungal burden and the growth rate. We also find both small and large-scale variation, including aneuploidy, associated with alternate growth phenotypes, which may impact the course of infection. Genes impacted by these variants are involved in transcriptional regulation, signal transduction, glycosylation, sugar transport, and glycolysis. We show that growth within the central nervous system (CNS) is reliant upon glycolysis in an animal model and likely impacts patient mortality, as the CNS yeast burden likely modulates patient outcome. Additionally, we find that genes with roles in sugar transport are enriched in regions under selection in specific lineages of this clinical population. Further, we demonstrate that genomic variants in two genes identified by GWAS impact virulence in animal models. Our approach identifies links between the genetic variation in C. neoformans and clinically relevant phenotypes and animal model pathogenesis, thereby shedding light on specific survival mechanisms within the CNS and identifying the pathways involved in yeast persistence. IMPORTANCE Infection outcomes for cryptococcosis, most commonly caused by C. neoformans, are influenced by host immune responses as well as by host and pathogen genetics. Infecting yeast isolates are genetically diverse; however, we lack a deep understanding of how this diversity impacts patient outcomes. To better understand both clinical isolate diversity and how diversity contributes to infection outcomes, we utilize a large collection of clinical C. neoformans samples that were isolated from patients enrolled in a clinical trial across 3 hospitals in Malawi. By combining whole-genome sequence data, clinical data, and in vitro growth data, we utilize genome-wide association approaches to examine the genetic basis of virulence. Genes with significant associations display virulence attributes in both murine and rabbit models, demonstrating that our approach can identify potential links between genetic variants and patho-biologically significant phenotypes.

Keywords: Cryptococcus; GWAS; aneuploidy; genome sequencing; sugar transporters; virulence.

FIG 1

Maximum likelihood phylogeny of patient isolates, estimated from 72,258 segregating SNP sites, rooted to VNII. Isolates separate distinctly into VNI, VNB, and VNII, with all lineages having 100% bootstrap support. All isolates possess MAT α, except for ACTA3523 (highlighted). The colored circles correspond to patient survival. The grayscale squares indicate the patient fungal burden of cerebrospinal fluid prior to treatment, log₁₀ CFU/mL.

FIG 2

Clinical outcomes by lineage. (A) Correlation coefficient between clinical phenotypes, P < 0.0001 in orange. Asterisk indicates the phi coefficient. Days survived indicates the total number of days survived for all individuals. Days survived (Died) indicates the number of days survived only for individuals that died during the ACTA clinical trial. (B) Deaths (top, gray) and survival (bottom) of patients by the infecting isolate lineage. (C) Probability of survival, by lineage of infecting isolate. (D) CSF log₁₀ CFU/mL (fungal burden) by infecting isolate lineage. Asterisk indicates P < 0.05 via a Wilcoxon test. (E) Rate of clearance (EFA) by the infecting isolate lineage. Displayed as −1(gradient). Asterisk indicates P < 0.05 and 0.01 via a Wilcoxon test.

FIG 3

Impact on the virulence of genes containing variants significantly associated with fungal burden. (A) Manhattan plot displaying variants associated with a high fungal burden. Variants with an association score < 0.000005 (score test) are labeled in red. (B) Survival of mice infected with closely related ACTA isolates 2129 and 3544. 15 CD-1 mice were infected with approximately 5 × 10⁴ CFU by oropharyngeal aspiration. (C) Rabbit CSF CFU for the parental strain (H99) and for two independent CNAG_06033 mutant strains extracted on days 3, 7, and 10 postinfection. Three rabbits were infected per strain, with n = 3 per strain at days 3 and 7, and n = 2 per strain at day 10. (D) Rabbit log₁₀ CFU/g for parental (H99) and one CNAG_06033 mutant strain, extracted from brain tissue (right lobe) at 10 days postinfection. (E) Survival of mice infected with the parental strain (H99) and four independent CNAG_06033 mutant strains. 5 CD-1 mice were infected with approximately 5 × 10⁴ CFU by oropharyngeal aspiration.

FIG 4

Isolate growth and phenotype correlation. Colony size (area, px) of isolates grown on YPD at (A) 30°C, (B) 37°C, and (C) 39°C. Correlation of isolate growth (area, px) on YPD, with the axes corresponding to the colony size when grown at (D) 30°C and 37°C or at (E) 37°C and 39°C. Colors correspond to the isolate lineage: VNIa (dark blue), VNIb (light blue), VNII (purple), VNB (orange).

FIG 5

Impact of aneuploidy on growth phenotypes. (A) Whole (aneuploid) and partial chromosomal duplication frequency throughout the population, by chromosome. Colony size (growth) on YPD, by ploidy state at (B) 30°C, (C) 37°C, (D) 39°C.