Quantification methods of Candida albicans are independent irrespective of fungal morphology

Abstract

The ability of Candida albicans to switch its morphology from yeast to filaments, known as polymorphism, may bias the methods used in microbial quantification. Here, we compared the quantification methods [cell/mL, colony forming units (CFU)/mL, and the number of nuclei estimated by viability polymerase chain reaction (vPCR)] of three strains of C. albicans (one reference strain and two clinical isolates) grown as yeast, filaments, and biofilms. Metabolic activity (XTT assay) was also used for biofilms. Comparisons between the methods were evaluated by agreement analyses [Intraclass and Concordance Correlation Coefficients (ICC and CCC, respectively) and Bland-Altman Plot] and Pearson Correlation (α = 0.05). Principal Component Analysis (PCA) was employed to visualize the similarities and differences between the methods. Results demonstrated a lack of agreement between all methods irrespective of fungal morphology/growth, even when a strong correlation was observed. Bland-Altman plot also demonstrated proportional bias between all methods for all morphologies/growth, except between CFU/mL X vPCR for yeasts and biofilms. For all morphologies, the correlation between the methods were strong, but without linear relationship between them, except for yeast where vPCR showed weak correlation with cells/mL and CFU/mL. XTT moderately correlated with CFU/mL and vPCR and weakly correlated with cells/mL. For all morphologies/growth, PCA showed that CFU/mL was similar to cells/mL and vPCR was distinct from them, but for biofilms vPCR became more similar to CFU/mL and cells/mL while XTT was the most distinct method. As conclusions, our investigation demonstrated that CFU/mL underestimated cells/mL, while vPCR overestimated both cells/mL and CFU/mL, and that the methods had poor agreement and lack of linear relationship, irrespective of C. albicans morphology/growth.1.

Keywords: Candida albicans; cell count; colony count; hyphae; microbial; polymerase chain reaction; yeasts.

Figure 1. Schematic representation of the morphological forms of yeast (A), hyphae (B), and pseudohyphae (C) of C. albicans.

Figure 2. Confocal images obtained for yeast growth of C. albicans SC5314. Bright-field (A) and fluorescence mode for Hoescht 33342 (nuclei in blue, (B)), propidium iodide (non-viable cells in red, (C)), and concanavalin A – Alexa Fluor 488 conjugate (viable cells in green, (D)). Red arrows show non-specific extracellular labeling of Hoescht 33342, white arrows show yeast cells and yellow arrows show filaments.

Figure 3. Bland-Altman plots and correlation analyses between the methods of cells/mL, CFU/mL, and vPCR (nuclei quantification) for yeast growth. The Bland-Altmand plots between cells/mL and CFU/mL (A), vPCR and cells/mL (B), and vPCR and CFU/mL (C) demonstrated lack of agreement, since dots are not evenly distributed around the mean line (black horizontal dotted line); red horizontal dotted lines: limits of agreement (standard deviation, SD), red whiskers: 95% CI of the limits of agreement; black continuos line: regression line showing the proportional bias in (A) and (B). Correlations between cells/mL and CFU/mL (D), vPCR and cells/mL (E), and vPCR and CFU/mL (F). The coeficient of determination (R²) shows that 77% of variation in cells/mL is explained by CFU/mL, while only 19% and 15% of variation in cells/mL and CFU/mL, respectively, are explained by vPCR data (nuclei quantification). Data were log₁₀-transformed for all assays.

Figure 4. Bland-Altman plots and correlation analyses between the methods of cells/mL, CFU/mL, and vPCR (nuclei quntification) for filament growth. The Bland-Altmand plots between cells/mL and CFU/mL (A), vPCR and cells/mL (B) and vPCR and CFU/mL (C) demonstrated lack of agreement, since dots are not evenly distributed around the mean line (black horizontal dotted line); red horizontal dotted lines: limits of agreement (standard deviation, SD), red whiskers: 95% CI of the limits of agreement; black continuos line: regression line showing the proportional bias with the differences between cells/mL and CFU/mL tend to zero as their averages (and values) increase in (A), while the differences between cells/mL and nuclei and also between CFU/mL and nuclei augment (in modulus) as their averages (and values) increase and this trend is more pronounced between cells/mL and nuclei. Correlations between cells/mL and CFU/mL (D), vPCR and cells/mL (E), and vPCR and CFU/mL (F). The coeficient of determination (R²) shows that 78% of variation in cells/mL is explained by CFU/mL, 66% and 69% of variation in cells/mL and CFU/mL, respectively, are explained by vPCR data (nuclei quantification). Data were log₁₀-transformed for all assays.

Figure 5. Bland-Altman plots and correlation analyses between the methods of cells/mL, CFU/mL, vPCR, and metabolic activity (XTT assay) for biofilms. Bland-Altmand plots between cells/mL and CFU/mL (A), vPCR and cells/mL (B), vPCR and CFU/mL (C), XTT and cells/mL (D), XTT and CFU/mL (E), and XTT and vPCR (F) demonstrated lack of agreement, since dots are not evenly distributed around the mean line (black horizontal dotted line); red horizontal dotted lines: limits of agreement (standard deviation, SD), red whiskers: 95% CI of the limits of agreement; black continuos line: regression line showing the proportional bias with the differences between cells/mL and CFU/mL tend to zero as their averages (and values) increase in (A), while the differences between cells/mL and nuclei and also between CFU/mL and nuclei augment (in modulus) as their averages (and values) increase and this trend is more pronounced between cells/mL and nuclei. Correlations between cells/mL and CFU/mL (G), vPCR and cells/mL (H), vPCR and CFU/mL (I), XTT and cells/mL (J), XTT and CFU/mL (K), and XTT and vPCR (L). The coeficient of determination (R²) shows that 62%, 69%, but only 23% of variation in cells/mL is explained by CFU/mL, vPCR, and XTT, respectively, 65% and 42% of variation in CFU/mL is explained by vPCR and XTT, respectively, and 55% of variation in vPCR is explained by XTT. Data were log₁₀-transformed (except for XTT) for (A), (B), (C), (G), (H), (I), (J), (K), and (L) and data were normalized for their maximum value for the concordance analysis of XTT (D, E, and F).

Figure 6. Principal Component Analysis (PCA) performed for all morphologies/growth (yeasts, filaments, and biofilms) involving the methods of cells/mL, CFU/mL, and vPCR (A) and also only for biofilms to include the XTT assay (B). After Oblimin rotation, PCA showed that cells/mL and CFU/mL were more similar to each other and loaded highly on component 1, while vPCR was a different method and loaded highly on component 2 for all morphologies/growths (C); for biofilms, vPCR became more similar to CFU/mL and cells/mL and loaded highly on component 1, while XTT was the most distinct method loading highly on component 2.