FTIR and Raman Spectroscopy-Based Biochemical Profiling Reflects Genomic Diversity of Clinical Candida Isolates That May Be Useful for Diagnosis and Targeted Therapy of Candidiasis

Despite the fact that Candida albicans is documented to be the main cause of human candidiasis, non-C. albicans Candida (NCAC) species, such as Candida glabrata and Candida tropicalis, are also suggested to be implicated in the etiopathogenesis of opportunistic fungal infections. As biology, epidemiology, pathogenicity, and antifungal resistance of NCAC species may be affected as a result of genomic diversity and plasticity, rapid and unambiguous identification of Candida species in clinical samples is essential for proper diagnosis and therapy. In the present study, 25 clinical isolates of C. albicans, C. glabrata, and C. tropicalis species were characterized in terms of their karyotype patterns, DNA content, and biochemical features. Fourier transform infrared (FTIR) spectra- and Raman spectra-based molecular fingerprints corresponded to the diversity of chromosomal traits and DNA levels that provided correct species identification. Moreover, Raman spectroscopy was documented to be useful for the evaluation of ergosterol content that may be associated with azole resistance. Taken together, we found that vibrational spectroscopy-based biochemical profiling reflects the variability of chromosome patterns and DNA content of clinical Candida species isolates and may facilitate the diagnosis and targeted therapy of candidiasis.


Introduction
Fungal infections caused by Candida species are either mucosal or systemic, in which the fungus invades and penetrates internal organs or tissues and/or reaches the bloodstream and spreads throughout the body (candidemia) [1,2]. Invasive candidiasis, which accounts for approximately three-fourths of systemic fungal infections, may be a life threatening condition, especially in a case of immunocompromised patients [2][3][4]. The Candida genus is composed of more than 150 heterogeneous species, but just a few of them have been implicated in human candidiasis [1]. While C. albicans causes the majority of human infections [1], the number of fungal infections caused by non-C. albicans Table 1. Clinical Candida species isolates used in the present study. Three reference strains, namely haploid, diploid, and tetraploid strains were also considered.  , C. tropicalis group (20,21), C. glabrata group (3 and 22-25), 302 haploid reference strain (26), SC5314 diploid strain (27), and T15 tetraploid strain (28).
In general, typical shapes (spherical to oval) and sizes (2-5 × 3-7 µm) of Candida cell isolates with the ability to form buds and/or hyphae/pseudohyphae were compared to reference strains ( Figure 1a). The ability to grow in yeast, pseudohyphal and hyphal forms is a characteristic feature of C. albicans biology [34]. As expected [1, 11,35], C. glabrata budding cells (blastoconidia) (1-4 µm) were smaller then C. albicans (4-6 µm) and C. tropicalis cells (4-8 µm) that is due to the fact that C. glabrata is generally considered haploid while C. albicans and C. tropicalis are diploid and C. glabrata did not form hyphae/pseudohyphae (Figure 1a).
Of course, clinical species cannot be determined solely based upon morphological features (Figure 1a). CHEF-PFGE was used for karyotype analysis (Figure 1b). C. albicans cells have eight pairs of chromosomal homologs, ranging in size from 0.95 to 3.3 Mb and comprising 16 Mb in total [36], but we were able to observe from four to nine chromosomes in clinical isolates assigned to C. albicans species (Figure 1b). The reference strains of C. albicans, namely haploid (302), diploid (SC5314) and tetraploid (T15) were characterized by seven, eight, and six distinguishable chromosomes, respectively ( Figure 1b). This confirms a high genomic diversity of C. albicans species [18,36]. It has been suggested that genomic diversity of C. albicans is due to chromosome length polymorphism (CLP) that results from expansion and contraction of subrepeats RPS; reciprocal translocation at the major repeat sequence (MRS) loci; chromosomal deletion and trisomy of individual chromosomes [36]. The karyotypes of C. tropicalis and C. glabrata isolates were found to be much more consistent (Figure 1b). Two C. tropicalis isolates had four chromosomes and five C. glabrata isolates had from six to twelve distinguishable chromosome bands (Figure 1b). The genome of C. glabrata clinical isolates was found to be very plastic with the variations in the number and size of chromosomes and the occurrence of intra-and interchromosomal segmental duplications [37]. For example, it has been reported that C. glabrata CBS 138 strain has 13 chromosomes with the genome size of 12.3 Mb [38]. CHEF-PFGE analysis found a minimum of 10 chromosome bands in C. glabrata [39]. Moreover, rapid changes in C. glabrata genomic organization have been comprehensively documented in numerous clinical studies [40][41][42][43]. Interestingly, isolates from one patient may exhibit 2 or 3 different karyotypes and during infection the chromosome pattern may change within a few days [41].
Phylogenetically, C. glabrata is more closely related to the model yeast Saccharomyces cerevisiae than to other Candida pathogens [38], as C. glabrata belongs to post-WGD (whole genome duplication) yeasts [37]. Chromosome similarity between 25 clinical Candida isolates was also further evaluated using NJ clustering (this study). Three considered species, namely C. albicans, C. tropicalis, and C. glabrata were characterized by clearly separate three clusters ( Figure 1c). The most accented chromosome polymorphism was observed among the C. albicans group. Nevertheless, due to high genomic diversity and plasticity [15,36,44], it is difficult to discriminate between Candida species based on karyotype profiling only.
We also analyzed DNA content of Candida isolates using fluorescent measurements and compared them to the reference strains used (n, 2n, and 4n) ( Figure 2). However, one should remember that haploid strains of C. albicans are considered to be unstable, often autodiploidize, and that genomic features often vary among tetraploid strains as well [23,45]. glabrata isolates had from six to twelve distinguishable chromosome bands (Figure 1b). The genome of C. glabrata clinical isolates was found to be very plastic with the variations in the number and size of chromosomes and the occurrence of intra-and interchromosomal segmental duplications [37]. For example, it has been reported that C. glabrata CBS 138 strain has 13 chromosomes with the genome size of 12.3 Mb [38]. CHEF-PFGE analysis found a minimum of 10 chromosome bands in C. glabrata [39]. Moreover, rapid changes in C. glabrata genomic organization have been comprehensively documented in numerous clinical studies [40][41][42][43]. Interestingly, isolates from one patient may exhibit 2 or 3 different karyotypes and during infection the chromosome pattern may change within a few days [41]. Phylogenetically, C. glabrata is more closely related to the model yeast Saccharomyces cerevisiae than to other Candida pathogens [38], as C. glabrata belongs to post-WGD (whole genome duplication) yeasts [37]. Chromosome similarity between 25 clinical Candida isolates was also further evaluated using NJ clustering (this study). Three considered species, namely C. albicans, C. tropicalis, and C. glabrata were characterized by clearly separate three clusters ( Figure 1c). The most accented chromosome polymorphism was observed among the C. albicans group. Nevertheless, due to high genomic diversity and plasticity [15,36,44], it is difficult to discriminate between Candida species based on karyotype profiling only.
As expected [46], DNA content of C. glabrata isolates was found to be the lowest ( Figure 2). DNA content of C. tropicalis isolates was higher than that of C. albicans and DNA content of C. albicans isolates was enormously diverse with a broad range between minimal and maximal values ( Figure 2). When considered relative fluorescent units (arbitrary units), n, 2n, and 4n reference strains were characterized by mean arbitrary units of 0.6, 0.74, and 1.07, respectively, whereas C. glabrata, C. tropicalis, and C. albicans isolates were characterized by mean arbitrary units of 0.46, 0.87, and 0.62, respectively ( Figure 2). Our data confirm DNA content diversity and plasticity of Candida albicans [16,44]. The DNA content/ploidy variation is considered as an adaptive mechanism in human pathogenic fungi [16,44,47]. C. albicans is normally a diploid organism (2n = 16), but a variety Fluorescence microscopy-based analysis of DNA content. Fixed cells (n = 100) were analyzed using an Olympus BX61 fluorescence microscope equipped with a DP72 CCD camera and Olympus CellF software (Olympus, Warsaw, Poland). For DNA visualization, the slides were counterstained with a drop of mounting medium containing 4 ,6 -diamino-2-phenylindole (DAPI) (blue). DNA content of clinical Candida isolates were compared to reference strains, namely haploid (26), diploid (27), and tetraploid (28) strains. DNA content was expressed as arbitrary units (relative fluorescence units from 0 to 4). Representative microphotographs and data distribution (histograms) are shown. C. albicans group (1, 2, and 4-19), C. tropicalis group (20,21), C. glabrata group (3 and 22-25), 302 haploid reference strain (26), SC5314 diploid strain (27), and T15 tetraploid strain (28).
As expected [46], DNA content of C. glabrata isolates was found to be the lowest ( Figure 2). DNA content of C. tropicalis isolates was higher than that of C. albicans and DNA content of C. albicans isolates was enormously diverse with a broad range between minimal and maximal values ( Figure 2). When considered relative fluorescent units (arbitrary units), n, 2n, and 4n reference strains were characterized by mean arbitrary units of 0.6, 0.74, and 1.07, respectively, whereas C. glabrata, C. tropicalis, and C. albicans isolates were characterized by mean arbitrary units of 0.46, 0.87, and 0.62, respectively ( Figure 2). Our data confirm DNA content diversity and plasticity of Candida albicans [16,44]. The DNA content/ploidy variation is considered as an adaptive mechanism in human pathogenic fungi [16,44,47]. C. albicans is normally a diploid organism (2n = 16), but a variety of stresses, namely heat shock, antifungal drug treatment or host-pathogen interactions can stimulate a plethora of aneuploidy events that seems to be well tolerated and may be considered as a selectively advantageous, e.g., may promote antifungal drug resistance [16,44,47]. Indeed, a specific segmental aneuploidy, consisting of an isochromosome composed of the two left arms of chromosome 5 (i5L), was reported to be associated with azole resistance in C. albicans [48]. This was achieved by amplification of two genes involved in fluconazole resistance, namely ERG11 (that encodes lanosterol-14-α-demethylase, the target of fluconazole) and TAC1 (that encodes a transcriptional regulator of ABC-transporter drug efflux pumps Cdr1 and Cdr2 that reduce intracellular azole concentration) [49]. More recently, trisomy of chromosome R and trisomy of chromosome 4 have been also reported to contribute to azole resistance in C. albicans [50,51]. The genomic plasticity is also associated with antifungal drug resistance in C. glabrata as the formation of new chromosomes was established as a virulence mechanism in C. glabrata clinical isolates [43]. Surprisingly, spontaneous changes in ploidy are also widespread in nonpathogenic fungi, namely in the model yeast Saccharomyces cerevisiae [52]. The appearance of diploid cells among haploid yeast cultures evolving for over 100 generations was documented and spontaneous diploidization was observed [52]. This relatively common event was based on both whole genome duplication (endoreduplication) and mating-type switching despite the use of heterothallic strains [52]. It has been suggested that spontaneous diploidization can be advantageous under certain stressful conditions in budding yeast [52]. Chromosomal copy number changes were also observed while analyzing the genome of clinical Saccharomyces cerevisiae strains that highlights the potential importance of large-scale genomic copy variation in yeast adaptation [53].

Biochemical Features Reflect Genomic Diversity and Plasticity of Candida Cells
We have then analyzed some biochemical features of clinical Candida isolates and we focused on elucidation of the usefulness of FTIR and Raman spectroscopy for Candida clinical isolate identification and determination of the effects of selected traits, such as karyotype polymorphism and DNA content on vibrational spectroscopy-based biochemical profiling. Initially, we considered the ability of Candida cells to accumulate glycogen ( Figure 3).
We found that higher DNA content was correlated with higher glycogen storage as judged using C. albicans reference strains of different ploidy, namely n, 2n, and 4n cells ( Figure 3b). Moreover, isolates of C. glabrata with relatively low DNA content were characterized by the lowest ability to accumulate glycogen ( Figure 3a). In contrast, clinical C. tropicalis isolates with higher DNA content compared to C. glabrata cells ( Figure 2) were found to accumulate the highest levels of glycogen among Candida cells considered ( Figure 3a). Additionally, C. albicans samples, namely isolates 2, 10, and 18, with much higher DNA content compared to other C. albicans samples were characterized by much higher ability to accumulate glycogen ( Figure 3a). As the DNA content may reflect biochemical/metabolic features in Candida isolates, we have then considered more sophisticated biochemical profiling using both FTIR and Raman spectroscopy ( Figure 4).
We have then analyzed some biochemical features of clinical Candida isolates and we focused on elucidation of the usefulness of FTIR and Raman spectroscopy for Candida clinical isolate identification and determination of the effects of selected traits, such as karyotype polymorphism and DNA content on vibrational spectroscopy-based biochemical profiling. Initially, we considered the ability of Candida cells to accumulate glycogen ( Figure 3).  We found that higher DNA content was correlated with higher glycogen storage as judged using C. albicans reference strains of different ploidy, namely n, 2n, and 4n cells ( Figure 3b). Moreover, isolates of C. glabrata with relatively low DNA content were characterized by the lowest ability to accumulate glycogen ( Figure 3a). In contrast, clinical C. tropicalis isolates with higher DNA content compared to C. glabrata cells ( Figure 2) were found to accumulate the highest levels of glycogen among Candida cells considered ( Figure 3a). Additionally, C. albicans samples, namely isolates 2, 10, and 18, with much higher DNA content compared to other C. albicans samples were characterized by much higher ability to accumulate glycogen ( Figure 3a). As the DNA content may reflect biochemical/metabolic features in Candida isolates, we have then considered more sophisticated biochemical profiling using both FTIR and Raman spectroscopy ( Figure 4). In general, FTIR spectroscopy and Raman spectroscopy are used for analytical chemistry applications. More recently, vibrational spectroscopy has been used to characterize biological materials, especially in the field of biomedicine for the rapid differentiation, classification, identification and large-scale screening at subspecies level of clinically relevant microorganisms [54][55][56][57]. These reagentless and nondestructive techniques are based on the absorption (FTIR) or scattering (Raman) of light directed onto a sample and provide a highly specific spectroscopic fingerprints of microorganisms by which they can be identified [54][55][56][57], and also enable for a detailed structural analysis to identify certain intracellular macromolecules [58]. However, data on vibrational spectroscopic identification of clinical Candida isolates in parallel with karyotype profiling, DNA content analysis and routine diagnostic phenotypic identification are not available.
Typical FTIR and Raman spectra of Candida isolate 1 with marked individual vibrations corresponding to functional groups of nucleic acids, phospholipids, carbohydrates, proteins, and lipids are presented in Figure 4. In the FTIR spectrum (Figure 4a), peaks at wavenumbers 879 cm −1 In general, FTIR spectroscopy and Raman spectroscopy are used for analytical chemistry applications. More recently, vibrational spectroscopy has been used to characterize biological materials, especially in the field of biomedicine for the rapid differentiation, classification, identification and large-scale screening at subspecies level of clinically relevant microorganisms [54][55][56][57]. These reagentless and nondestructive techniques are based on the absorption (FTIR) or scattering (Raman) of light directed onto a sample and provide a highly specific spectroscopic fingerprints of microorganisms by which they can be identified [54][55][56][57], and also enable for a detailed structural analysis to identify certain intracellular macromolecules [58]. However, data on vibrational spectroscopic identification of clinical Candida isolates in parallel with karyotype profiling, DNA content analysis and routine diagnostic phenotypic identification are not available.
Typical FTIR and Raman spectra of Candida isolate 1 with marked individual vibrations corresponding to functional groups of nucleic acids, phospholipids, carbohydrates, proteins, and lipids are presented in Figure 4. In the FTIR spectrum (Figure 4a), peaks at wavenumbers 879 cm −1 and 1074 cm −1 are corresponding to C-O, C-O-H, and C-O-C deformation and C-C stretching vibrations of carbohydrates and β(1-3)glucans, nucleic acids and glycogen and PO 2− symmetric stretching vibrations mainly from RNA [58][59][60], respectively. The peak at 1247 cm -1 originates from C-O asymmetric stretching vibrations in phospholipids [59]. Moreover, the vibration at 1290 cm −1 corresponds to amide III [59]. In the FTIR spectrum, peak at 1398 cm −1 originates from C=O of COO − symmetric stretching vibrations in proteins and CH 2 wagging vibrations in lipids and β(1-3)glucans [59]. The peaks at 1540 cm −1 and 1616 cm −1 correspond to amide II and amide I vibrations, respectively [58]. Furthermore, peaks at: 1735 cm −1 , 2914 cm −1 , and 2973 cm −1 originate from CH vibrations in lipids [61]. The last two peaks in the FTIR spectrum (3257 cm −1 and 3397 cm −1 ) correspond to OH vibrations from water and amide A from proteins, respectively [62]. Moreover, in the Raman spectrum (Figure 4b), the vibrations from symmetric benzene/pyrrole in-phase and out-of-phase breathing modes of tryptophan and phenylalanine (904 cm −1 , 981 cm −1 ) are observed. Furthermore, peaks at 1317 cm −1 and 1459 cm −1 correspond to C-H deformation vibrations from proteins [63] and C-H deformation vibrations from lipids [63], respectively. In the Raman spectrum, a vibration at 1587 cm −1 originating from ring stretching vibrations of the deoxyribonucleotide adenosine monophosphate is observed [63]. Amide I vibrations in Raman spectrum are documented at 1648 cm −1 [63]. Moreover, a peak at 2929 cm −1 originates from C-H stretching vibration from lipids is observed [64]. FTIR and Raman spectra of all clinical isolates considered are presented in Supplemental Figure S1 and Supplemental Figure S2. The peak positions and information about vibrations for all samples are denoted in Supplemental Table S1. According to FTIR and Raman spectra, the differences in the signal intensity of functional groups as well as differences in the occurrence of these groups may be noticed (Supplemental Figure S1 and Supplemental Figure S2 and Supplemental Table S1). According to the differences in signal intensities of some selected vibrations of FTIR and Raman spectra, we have performed a comparative analysis between clinical samples belonging to three Candida species ( Figure 5). We have considered 11 vibrations of the FTIR spectrum and seven vibrations of the Raman spectrum ( Figure 5).
For seven vibrations of FTIR spectrum, we were able to obtain statistically significant differences in signal intensities between C. albicans isolates and other Candida isolates. These vibrations were  Figure 5). Protein components of clinical Candida isolates were also characterized. To determine a secondary protein structure, a deconvolution of FTIR amide I region was considered (Supplemental Figure S3). The abundance of α and β structures, as well as the ratio of α/β structures within analyzed peak were calculated (Supplemental Table S2). C. tropicalis group was characterized by higher ratio of α/β structures compared to other Candida groups (Supplemental Table S2). The lipid-carbohydrate ratio was also analyzed in clinical Candida isolates (Supplemental Table S3) that was calculated on the basis of peak area corresponding to lipid and carbohydrate vibrations (Supplementary Table S4). In general, within C. tropicalis and C. glabrata groups, the lipid-carbohydrate ratio was comparable and the variability was rather slightly accented. In contrast, C. albicans group was characterized by a diverse lipid-carbohydrate ratio, e.g., ranging from 0.16 to 0.93 (Supplemental Table S3). spectra, the differences in the signal intensity of functional groups as well as differences in the occurrence of these groups may be noticed (Supplemental Figure S1 and Supplemental Figure S2 and Supplemental Table S1). According to the differences in signal intensities of some selected vibrations of FTIR and Raman spectra, we have performed a comparative analysis between clinical samples belonging to three Candida species ( Figure 5). We have considered 11 vibrations of the FTIR spectrum and seven vibrations of the Raman spectrum ( Figure 5).  More recently, the quantitation of ergosterol content has been established as a novel method for determination of fluconazole susceptibility of C. albicans [65]. Azole stress has been also reported to cause upregulation of genes involved in sterol uptake and biosynthesis in C. glabrata [66]. Fluconazole treatment resulted in increased mRNA levels of ergosterol biosynthetic genes, namely CgERG2, CgERG3, CgERG4, CgERG10, and CgERG11 and sterol influx transporter AUS1 and sterol metabolism regulators SUT1 and UPC2 in C. glabrata [66]. Moreover, stimulation with exogenous source of cholesterol or ergosterol conferred resistance to fluconazole and voriconazole in C. glabrata [66]. As ergosterol abundance may modulate azole antifungal resistance in clinical Candida isolates, we decided then to analyze ergosterol content using both FTIR and Raman spectroscopy ( Figure 6). Fluconazole treatment resulted in increased mRNA levels of ergosterol biosynthetic genes, namely CgERG2, CgERG3, CgERG4, CgERG10, and CgERG11 and sterol influx transporter AUS1 and sterol metabolism regulators SUT1 and UPC2 in C. glabrata [66]. Moreover, stimulation with exogenous source of cholesterol or ergosterol conferred resistance to fluconazole and voriconazole in C. glabrata [66]. As ergosterol abundance may modulate azole antifungal resistance in clinical Candida isolates, we decided then to analyze ergosterol content using both FTIR and Raman spectroscopy ( Figure 6).  (27), and T15 tetraploid strain (28). Bars indicate SD, n = 3, a.u., arbitrary units.
Representative FTIR and Raman spectra of ergosterol are presented in Figure 6a,b, respectively. Using FTIR spectroscopy, similar levels of ergosterol were revealed in all analyzed samples (Figure 6c). However, using Raman spectroscopy, we were able to show differences in the content of ergosterol (Figure 6c). For ergosterol content analysis, we have selected a peak at 1459 cm −1 instead of a peak at 1602 cm −1 [67] to rule out the possibility of some overlapping with protein signals. The most diverse group in term of ergosterol content was C. albicans group, e.g., sample 7 was characterized by eight times lower levels of ergosterol than sample 9 (Figure 6c).
We have then considered principal component analysis (PCA) and hierarchical cluster analysis (HCA) using both FTIR and Raman spectra (Figure 7).
For PCA, we have selected lipid-carbohydrate ratio and α/β structure ratio (Figure 7a,b). According to FTIR spectra (Figure 7a), isolates from C. tropicalis (samples 20 and 21) and C. glabrata (samples 22 to 25, but not sample 3) species were grouped together within their own categories, whereas C. albicans group was found to be diverse with several separated subgroups, e.g., a subgroup that consists of samples 1, 2, 4-7 or a subgroup that consists of samples 14, 17, and 19 that was also grouped with a haploid reference strain (sample 26). Raman spectra-based PCA did not reveal similar clustering (Figure 7b). We have then considered HCA based on FTIR spectra from 500 to 4000 cm −1 and Raman spectra from 500 to 3000 cm −1 (Figure 7c,d). According to FTIR spectra, Candida isolates were grouped into previously assigned species, namely C. albicans, C. tropicalis, and C. glabrata (Table 1, Figure 7c). Similarly to PCA, several subgroups of C. albicans group were documented, e.g., one containing samples from 1 to 2 and from 4 to 7 and second with samples from 8 to 19 without sample 17 that was grouped as its own category (Figure 7c). In general, such clustering also reflected the differences in DNA content among C. albicans group (Figure 2). In contrast, Raman spectra-based HCA did not provide discrimination between Candida species (Figure 7d). ergosterol (Figure 6c). For ergosterol content analysis, we have selected a peak at 1459 cm instead of a peak at 1602 cm −1 [67] to rule out the possibility of some overlapping with protein signals. The most diverse group in term of ergosterol content was C. albicans group, e.g., sample 7 was characterized by eight times lower levels of ergosterol than sample 9 (Figure 6c).
There are several reports on vibrational spectroscopy identification of clinical Candida isolates, but none of them provide a comparison with karyotype profiling and DNA content analysis. The difficulty of differentiating at the strain level, especially when high accumulated doses of an antifungal agent are involved, has been documented while analyzing FTIR spectra of pathogenic C. albicans isolates from HIV-positive patients [28]. Using FTIR spectroscopy, six species (C. albicans, C. glabrata, C. parapsilosis, C. tropicalis, C. krusei, and C. kefyr) from a collection of 57 clinical strains of Candida and isolated from hospitalized patients were identified with a classification rate of 100% for both microcolonies and 24 h cultures [29]. More recently, next generation sequencing (NGS) of ITS and D1/D2 LSU marker regions together with FTIR spectroscopy were applied to identify 256 pathogenic strains belonging to Candida genus [30]. Strains of C. albicans, C. parapsilosis, C. glabrata, and C. tropicalis were identified with high-throughput NGS sequencing of ITS and LSU markers and then with FTIR, and total percentage of correct identification reached 97.4% for C. albicans and 74% for C. parapsilosis while the other two species showed lower identification rates [30]. The authors concluded that the identification success increases with the increasing number of strains actually used in the PLS analysis [30].
A set of 42 Candida strains comprising five species that are frequently encountered in clinical microbiology was also considered to analyze the usefulness of confocal Raman microspectroscopy for the rapid identification of Candida species [32]. Using multivariate statistical analyses, a high prediction accuracy (97 to 100%) was documented [32]. The authors concluded that confocal Raman microspectroscopy offers a rapid, accurate, and easy-to-use alternative for the identification of clinically relevant Candida species [32]. Raman spectroscopy has been also found an accurate and rapid (12-24 h) alternative for the identification of Candida spp. in peritonitis patients [31].
Finally, we have considered a joined clustering analysis of chromosome number, DNA content, the intensities of 11 vibrations of FTIR spectrum and seven vibrations of Raman spectrum, alpha-helix/beta-sheet ratio, and lipid-carbohydrate ratio (Figure 8). Using this approach, we were able to discriminate between Candida species considered ( Figure  8). One exception was sample 3, C. glabrata species that was classified within C. albicans group (Figure 8). However, this result may be due to fluconazole treatment (Table 1, Figure 8).
Taken together, we have shown for the first time that vibrational spectroscopy-based biochemical profiling reflected genomic diversity (karyotype patterns, DNA content) of 25 clinical Candida isolates (Figure 8). However, using FTIR or Raman spectroscopy as isolated methods for Figure 8. A joined clustering analysis of chromosome number, DNA content, signal intensities of some selected vibrations of FTIR and Raman spectra (11 vibrations of FTIR spectrum, seven vibrations of Raman spectrum), alpha-helix/beta-sheet ratio and lipid-carbohydrate ratio that allows for proper grouping of three Candida species considered. The effect of antifungal treatment is also denoted. A heat map generated from FTIR and Raman spectroscopy data, karyotype profiling, and DNA content data is shown. Hierarchical clustering was created using ClustVis, a web tool for visualizing clustering of multivariate data (BETA) (https://biit.cs.ut.ee/clustvis/). Using this approach, we were able to discriminate between Candida species considered (Figure 8). One exception was sample 3, C. glabrata species that was classified within C. albicans group (Figure 8). However, this result may be due to fluconazole treatment (Table 1, Figure 8).
Taken together, we have shown for the first time that vibrational spectroscopy-based biochemical profiling reflected genomic diversity (karyotype patterns, DNA content) of 25 clinical Candida isolates (Figure 8). However, using FTIR or Raman spectroscopy as isolated methods for Candida species identification may be limited. FTIR-as well as Raman-based clustering analysis (Figure 7) yielded ambiguous results that were not entirely comparable to karyotype profiling-based clustering analysis (Figure 1c). Thus, only joined clustering analysis of chromosome number, DNA content and vibrational spectroscopy-based biochemical profiling may allow for grouping together the clinical Candida isolates from the same species (Figure 8). The usefulness of vibrational spectroscopy methods for characterization and identification of clinical Candida isolates is also summarized in Figure 9. clustering analysis (Figure 1c). Thus, only joined clustering analysis of chromosome number, DNA content and vibrational spectroscopy-based biochemical profiling may allow for grouping together the clinical Candida isolates from the same species (Figure 8). The usefulness of vibrational spectroscopy methods for characterization and identification of clinical Candida isolates is also summarized in Figure 9. We also postulate that Raman spectroscopy can be adapted for rapid and accurate analysis of ergosterol content in clinical Candida isolates and thus may provide information on azole resistance/susceptibility (Figure 9). Vibrational spectroscopy-based data may be included in global spectral databases for identification purposes and may facilitate diagnosis and targeted therapy of candidiasis ( Figure 9). Indeed, limited use of vibrational spectroscopy-based techniques for medical diagnostics seems to be due to the absence of reliable and validated libraries linked to taxonomically sound identification procedure [30]. More recently, it has been postulated that such libraries should include several tens of strains for each relevant species and the panel of strains needs to be composed of well-identified strains, e.g., deriving from diverse sources and collected over an extensive time period [30]. Postulated approach would require a multidisciplinary effort of specialists working in strain isolation and maintenance, molecular taxonomy, vibrational spectroscopy-based techniques, data management and data basing [30].  . The usefulness of vibrational spectroscopy methods for comprehensive biochemical characterization and identification of clinical Candida isolates. FTIR spectra-and Raman spectra-based biochemical profiling of clinical Candida isolates together with karyotype profiling and DNA content analysis allows for accurate identification of Candida species. Raman spectroscopy can be also adapted for rapid and accurate measurements of ergosterol content that may provide information of azole resistance/susceptibility. Established spectral databases can be useful for diagnosis and targeted therapy of candidiasis.

Ethics Statement
We also postulate that Raman spectroscopy can be adapted for rapid and accurate analysis of ergosterol content in clinical Candida isolates and thus may provide information on azole resistance/susceptibility (Figure 9). Vibrational spectroscopy-based data may be included in global spectral databases for identification purposes and may facilitate diagnosis and targeted therapy of candidiasis ( Figure 9). Indeed, limited use of vibrational spectroscopy-based techniques for medical diagnostics seems to be due to the absence of reliable and validated libraries linked to taxonomically sound identification procedure [30]. More recently, it has been postulated that such libraries should include several tens of strains for each relevant species and the panel of strains needs to be composed of well-identified strains, e.g., deriving from diverse sources and collected over an extensive time period [30]. Postulated approach would require a multidisciplinary effort of specialists working in strain isolation and maintenance, molecular taxonomy, vibrational spectroscopy-based techniques, data management and data basing [30].

Clinical Specimens and Reference Strains
A total of 25 clinical samples used in this study are listed in Table 1. Clinical isolates were originated from human bronchoalveolar lavage, sputum, pharynx, wound, urine, and vagina. Patient 1 and patients 3, 6, 8, and 25 had been treated with voriconazole and fluconazole, respectively (Table 1). The following three Candida albicans strains of known ploidy were used as reference strains: C. albicans 302 (haploid), C. albicans SC5314 (diploid), and C. albicans T15 (FH6, tetraploid but trisomic for chromosomes 2/3 with multiple copies of chromosome 5L) [49]. The reference strains were a generous gift from Prof. Judith Berman (Department of Molecular Microbiology and Biotechnology, Tel Aviv University, Israel).

Preparation of Agarose-Embedded Yeast DNA
Yeast DNA, isolated from cells at a logarithmic phase of growth (3 × 10 8 cells), was obtained using BIORAD CHEF Yeast Genomic DNA Plug Kit (BIORAD, Warsaw, Poland) using a standard protocol [15] according to the manufacturer's instructions, with minor modifications. Briefly, instead of standard lyticase solution, a mix of standard lyticase and zymolyase 100T, 125 µg/mL (US Biological, Salem, MA, USA), and overnight incubation at 37 • C was used for spheroplast preparation and prolonged proteinase K treatment (48 h at 50 • C) was applied for protein digestion.

Pulsed-Field Gel Electrophoresis (PFGE)
Contour clamped homogeneous electric field (CHEF)-PFGE separation of yeast whole chromosomes was performed on a 1% agarose gel in 0.5× TBE according to the manufacturer's instructions using CHEF-DR ® III Pulsed Field Electrophoresis System (BIORAD, Warsaw, Poland) and the following conditions: 60 to 120 s switch, 6 V cm −1 , 120 • angle for 36 h, followed by 120 to 300 s switch, 4.5 V cm -1 , 120 • angle for 12 h. After CHEF-PFGE separation, yeast chromosomes were stained using ethidium bromide. The dendrogram of chromosomal DNA-based similarity was created using Free-Tree software [68] using neighbor-joining (NJ) method with Sokal-Sneath-Anderberg matrix and FigTree tree figure drawing tool (http://tree.bio.ed.ac.uk/software/figtree/) (access on 12 September 2018).

DNA Content Analysis
Yeast cells from log phase cultures were diluted to 10 7 cells/mL and fixed with 70% ethanol at −21 • C for 24 h. After incubation, the cells were washed with PBS and resuspended in 500 µL of spheroplast buffer, spread onto slides and permeabilized with PBS containing 0.1% Triton X-100. The slides were treated with 100 µg/mL RNAse (Sigma-Aldrich) in 2× saline sodium citrate (SSC) buffer in a humidified chamber at 37 • C for 1 h for enhanced results. Next, the slides were washed three times in PBS buffer. For DNA visualization, the slides were counterstained with a drop of mounting medium containing 4 ,6 -diamino-2-phenylindole (DAPI) (Cambio, Cambridge, UK) and then analyzed using an Olympus BX61 fluorescence microscope equipped with a DP72 CCD camera and Olympus CellF software (Olympus, Warsaw, Poland). The CCD capture conditions were as the following: exposure time 150 ms, 100× oil immersion objective. DAPI fluorescent signals were collected using DAPI filters (λ ex = 345 nm, λ em = 455). Fluorescence microscopy was adapted for DNA content analysis. ImageJ software (http://rsbweb.nih.gov/ij/) (access on 29 July 2018). was used to analyze the nuclear DNA content. DNA content was expressed as arbitrary units [a.u.].

Glycogen Storage Assay
The ability of yeast cells to accumulate glycogen was evaluated using iodine staining of yeast colonies [69] on the basis that glycogen gives a reddish-brown coloration with iodine. Briefly, 2 µL of Candida cell suspensions at 10 7 cells/mL were inoculated on solid YPD medium and glycogen storage was detected by flooding 3-day colonies with 5 mL of iodine solution (0.2% I 2 in 0.4% KI). The staining reactions of the colonies were recorded 1 min after adding the iodine and glycogen content [a.u.] was calculated using ImageJ software (http://rsbweb.nih.gov/ij/) (access on 12 July 2018). Correlation between glycogen content (a.u.) and DNA content (a.u.) was considered. Correlation analysis of the data was performed using linear correlation (Pearson r) test.

FTIR Spectroscopy
Fourier-transform infrared (FTIR) spectroscopy measurements were performed using the Vertex 70 (Bruker, Poznan, Poland) spectrometer using the attenuated total reflectance (ATR) technique. The range of selected infrared radiation was the average IR (400-4000 cm −1 ). 32 scans with 2 cm -1 spectral resolution were performed. Normalization and baseline correction of obtained spectra were considered. All spectra were analyzed using OPUS software (Bruker, Poznan, Poland).
3.9. Deconvolution of Amide I region (1600-1700 cm −1 ) The secondary protein structure was analyzed by means of curve fitting using MagicPlot 2.1. software (https://magicplot.com/downloads.php) (access on 3 July 2018). First, the secondary derivative spectra were determined based on the ATR-FTIR spectra to determine the initial peak positions for curve fitting, and the peaks were fitted using Gaussian function. The area under the curve was considered 100% and each component was expressed as its percentage after fitting.

Raman Spectroscopy
FT-Raman spectra were recorded using a Nicolet NXR 9650 FT-Raman Spectrometer equipped with an Nd:YAG laser (1064 nm) and a germanium detector. Measurements were performed in the range of 150 to 3700 cm −1 with a laser power of 1.5 W. An unfocused laser beam of a diameter of approximately 100 µm and a spectral resolution of 8 cm −1 was used. Raman spectra were processed by the Omnic/Thermo Scientific software based on 64 scans.

Lipid-Carbohydrate Ratio
To evaluate lipid-carbohydrate ratio, an area of peaks corresponding to lipid and carbohydrate vibrations were calculated [70]. The sum of the lipid as well as carbohydrate peak area were then calculated and the ratio of the sum of lipid and carbohydrate was calculated. To evaluate lipid-carbohydrate ratio, ORIGIN software was used.

Ergosterol Content
To estimate the levels of ergosterol, FTIR and Raman spectra of ergosterol were obtained. Ergosterol (Sigma-Aldrich, Poznan, Poland) was used as a reference standard. The value of intensity of individual peak from FTIR as well as Raman spectra at 1247 cm −1 and 1459 cm −1 , respectively, was considered.

Multivariate Data Analysis
All obtained spectra were subjected to multivariate analysis using principal component analysis (PCA) and hierarchical cluster analysis (HCA) using PAST 3.0. software. HCA was based on Euclidean distance and Ward's algorithms. The PCA and HCA were performed for all FTIR as well as Raman spectral ranges.
Moreover, a joined clustering analysis of chromosome number, DNA content, 11 vibrations of FTIR spectrum, seven vibrations of Raman spectrum, alpha-helix/beta-sheet ratio and lipid-carbohydrate ratio was performed using ClustVis, a web tool for visualizing clustering of multivariate data (BETA) (https://biit.cs.ut.ee/clustvis/) (access on 3 September 2018). [71]. Species clustering as well as antifungal treatment were included. A heat map was generated on the basis of karyotype profiling, DNA content analysis and signal intensities of some selected vibrations of FTIR and Raman spectra,