Sphingolipid diversity in Candida auris: unraveling interclade and drug resistance fingerprints

Abstract In this study, we explored the sphingolipid (SL) landscape in Candida auris, which plays pivotal roles in fungal biology and drug susceptibility. The composition of SLs exhibited substantial variations at both the SL class and molecular species levels among clade isolates. Utilizing principal component analysis, we successfully differentiated the five clades based on their SL class composition. While phytoceramide (PCer) was uniformly the most abundant SL class in all the isolates, other classes showed significant variations. These variations were not limited to SL class level only as the proportion of different molecular species containing variable number of carbons in fatty acid chains also differed between the isolates. Also a comparative analysis revealed abundance of ceramides and glucosylceramides in fluconazole susceptible isolates. Furthermore, by comparing drug-resistant and susceptible isolates within clade IV, we uncovered significant intraclade differences in key SL classes such as high PCer and low long chain base (LCB) content in resistant strains, underscoring the impact of SL heterogeneity on drug resistance development in C. auris. These findings shed light on the multifaceted interplay between genomic diversity, SLs, and drug resistance in this emerging fungal pathogen.


Introduction
The simultaneous global emergence of distinct clades of Candida auris , along with its resistance to multiple antifungal drugs, has posed a significant challenge for clinicians worldwide (Lockhart et al. 2017 ).Compared to other Candida species, C. auris isolates found in hospitals often exhibit high le v els of resistance to azoles and can display collateral resistance to amphotericin B (AmB) and echinocandins (Chowdhary et al. 2018 ).Azoles target the lanosterol 14-α-demethylase encoded by ERG11 leading to accumulation of toxic sterols in the yeast cell.The commonl y r ecorded r esistance mechanism to fluconazole (FLC) includes transcriptional activation of its target ERG11 or generating mutant variants of the tar get pr otein (White 1997, Morio et al. 2010 ).The increased efflux of FLC facilitated by ov er expr essed export pr oteins fr om the ABC or MFS families is another major determinant that contribute to azole r esistance.Recentl y, segmental c hr omosomal duplications have also been linked to drug resistance in C. auris (Bhattacharya et al. 2019, Kim et al. 2019, Wasi et al. 2019, Rybak et al. 2021, Narayanan et al. 2022 ).The widespread azole resistance observed in C. auris has prompted researchers to investigate alternative mechanisms of drug resistance that may explain this behaviour.Understanding the complex interplay of canonical mechanisms and identifying new strategies to combat drug resistance in C. auris is vital (Chaabane et al. 2019 ).
Lipid molecules, particularly sphingolipids (SLs) and phosphogl ycerides (PGLs) hav e gained attention as molecular determinants influencing the drug susceptibilities of yeast cells (Kohli et al. 2002, Mukhopadhyay et al. 2004 ).While PGLs in yeast cells share compositional similarities with those found in other eukaryotic cells, the SL profiles of fungi are unique.In mammals, SLs consist of glycosphingolipids and gangliosides, whereas yeast cells incor por ate inositol and mannose, along with neutr al SLs, to form complex acidic SLs (Smith and Lester 1974, Del Poeta et al. 2014, Renne and de Kroon 2018 ).The synthesis of SLs initiates in the endoplasmic reticulum through the condensation of serine (an amino acid) and palmitoyl Co-A (fatty acid deri vati ve), which is catalysed by the enzyme serine palmitoyltr ansfer ase (SPT) (Le vine et al. 2000 , Funato andRiezman 2001 ).The product of this condensation reaction is 3-ketodihydrosphingosine, which is reduced to yield dihydrosphingosine (DHS).F rom DHS tw o branches originate in the pathway with different end products .T he neutral br anc h terminates with the formation of glucosylceramides (Glc-Cer) while as the acidic br anc h terminates with the formation of mannosyl diinositolphosphorylceramide [M(IP) 2 C] (Shoma et al. 2023 ).Inter estingl y, the end pr oduct of SL biosynthesis differs among organisms.In Saccharomyces cerevisiae , the final product is M(IP) 2 C, which is generated through three irreversible steps and GlcCer is not synthesized by this yeast (Hanada 2003, Saito et al. 2006, Usmani et al. 2023 ).Lack of GlcCer synthesis has been also reported in Candida glabrata , Candida guilliermondii , and so on (Saito et al. 2006 ).In contr ast, fungi fr om the e volutionary distinct m ucormycotina suc h as Mucor hiemalis , Rhizopus microsporus have GlcCer as the primary complex SL and do not have inositol phosphorylceramides (IPCs; acidic SL) (Aoki et al. 2004 ).Common pathogenic fungi like Cryptococcus neoformans and Candida albicans have both the acidic and neutral pathways active (Oura and Kajiwara 2010, Singh et al. 2017, Garbe et al. 2022 ).The enzymes involved in the biosynthesis of complex SLs in fungi are absent in mammals, making them a potential target for new antifungal ther a pies (Nimric hter and Rodrigues 2011 ).SLs also play crucial r oles in v arious cellular pr ocesses, including signalling, heat str ess response, and serving as structural components (Dickson 2010 ).
SLs and er goster ol, another important lipid component, interact within the microdomain of membrane.While common antifungal drugs target ergosterol synthesis, the precise role of SLs in influencing drug resistance is beginning to emerge (Mazu et al. 2016, Rollin-Pinheir o et al. 2016, P an et al. 2018 ).We have shown earlier that the deletion of erg or SLs biosynthetic genes results in increased susceptibility to w ar ds antifungals in C. albicans .The deletion of SLs or er goster ol also affect membrane localization of major multidrug exporter protein e.g Cdr1p in C. albicans (Pasrija et al. 2008 ) Furthermore, some SLs have been found to attenuate fungal pathogenesis.Inhibition or deletion of enzymes involved in the biosynthetic pathways of specific SLs, such as IPCs and Glc-Cer, have been shown to affect fungal virulence (Zhong et al. 2000, Le v ery et al. 2002, Rittershaus 2006 ).In fact, natural inhibitors of fungal GlcCer synthesis have been identified and possess antifungal properties (Mor et al. 2015 ).
Deletion of IPT1 , a gene involved in the synthesis of M(IP) 2 C, in C. albicans and C. glabrata has been shown to significantl y incr ease susceptibility to azole antifungal drugs (Prasad et al. 2005, Shahi et al. 2022 ).Similarly, the null mutants of FEN1 and FEN12 , which encode enzymes responsible for synthesizing v ery long-c hain fatty acids, exhibit increased susceptibility to AmB in both S. cerevisiae and C. albicans (Sharma et al. 2014 ).Additionall y, upr egulation of SL biosynthesis genes has been observed in FLC resistant C. albicans isolates (Gao et al. 2018 ).Imbalances in the le v els of SLs or er goster ol hav e been found to dir ectl y affect the tr affic king of ABC efflux pump pr oteins, r endering C. albicans highl y susceptible to antifungal drugs .T hese findings highlight the intricate interplay between intracellular drug accumulation, drug efflux mechanisms, and the membrane lipid environment in determining the drug susceptibility phenotype of Candida species (Bagnat et al. 2000, Urbanek et al. 2022 ).Understanding these interactions is crucial for de v eloping str ategies to ov ercome drug r esistance and enhance the effectiveness of antifungal therapies.
Considering the intricate relationship between membrane lipids and drug resistance, and the limited information available regarding these aspects in C. auris , it is crucial to investigate the landscape of SLs and their role in supporting drug resistance.Lar ge-scale mass spectr ometry (MS)-based lipidomic studies have pr ov en v aluable in establishing connections between specific lipid structur es, their le v els, and physiological functions in yeast.Singh et al. ( 2012 ) sho w ed that mitochondrial lipids are associated with cell wall integrity and azole resistance in C. albicans (Singh et al. 2012 ) .By employing the ESI-MS/MS (electr ospr ay ionization tandem MS) a ppr oac h, r ecentl y, Shahi et al. ( 2020 ) pr esented a com-par ativ e lipidome between drug-resistant and susceptible C. auris isolates and pointed to w ar ds a significant remodelling of polar lipids in drug-resistant C. auris .In another study, Kumar et al. ( 2021 ) analysed the molecular SL signatures of drug-resistant clinical isolates of C. auris r ecov er ed fr om Indian hospitals .T he study highlighted the distinct specific molecular species fingerprints of SL classes among the tested isolates, reinforcing their influence on drug resistance (Kumar et al. 2021 ).Together, these studies point to w ar ds the compositional remodelling of SL species structures that could be responsible for drug resistance.Ho w ever, detailed studies, including a larger pool of isolates of known clades of C. auris r ecov er ed fr om differ ent geogr a phical r egions, ar e r equir ed to understand the intricacies of such changes.
The present study explores the sphingolipidomic fingerprint of C. auris clade isolates r ecov er ed fr om differ ent geogr a phical locations .T hese C. auris strains belonging to the South Asian, East Asian, African, South American, and Iranian clades I, II, III, IV, and V, r espectiv el y, wer e included in the present study.The present study could not only highlight the distinct interclade fingerprints of SLs but also r e v eals the div ersity of SL species between drug susceptible and resistant isolates of C. auris.

Strains, media, and reagents
Candida auris strains used in this study were acquired from the National Culture Collection of Pathogenic Fungi (NCCPF), Postgraduate Institute of Medical Education and Research (PGIMER), Chandigarh (NCCPF 470156, NCCPF 470157 both isolated from blood, and NCCPF 470296, isolated from pus).CBS 10913T, the first clinical isolate of C. auris , was acquired from the Central Bur eau voor Sc himmel Cultur es (CBS), Fungal Biodiv ersity Centr e of the Ro y al Netherlands, Academy of Arts and Sciences (KN AW), the CDC & FDA Antibiotic Resistance Isolate Bank (AR 0383-0386, all isolated from blood and AR 1097 isolated from ear discharge).Strains VPCI 479/P/13 (isolated from blood), and LMDM 1219 (Li et al. 2021 ) were received as a kind gift from Prof. Dominique Sanglard, University of Lausanne and University Hospital Centre, Lausanne, Switzerland.All strains were archived in 25% glycerol at −80 • C and r e viv ed on YPD a gar at 30 • C for experimental purposes.
All solvents and reagents used (unless specified) were LCMS gr ade, purc hased fr om Honeywell NC, USA, and Sigma Aldrich MO, USA.Lipid standards were purchased from Avanti Polar Lipids Inc. AL, USA.

Drug-susceptibility assays
Minim um inhibitory concentr ations (MICs) for FLC and AmB a gainst C. auris str ains wer e determined as described by the Clinical and Laboratory Standards Institute (CLSI) by broth microdilution method with 2-fold serial dilutions in 96-well micr otitr e plates (M27Ed4: Broth Dilution Antifungal Susceptibility, Yeasts; CLSI 2017 ).For FLC, 50% reduction in growth of a particular strain compared to drug-free control (YPD) was considered as the endpoint.Whereas for AmB, 100% growth inhibition was considered as the endpoint.

Lipid extraction
Cultur es wer e gr own in liquid YPD at 30 • C till satur ation.Fr om this, a secondary culture with a starting OD 600 of 0.1 in fresh media (50 ml) w as gro wn up to OD 600 0.8-1 (mid-log phase).Approximately 5 × 10 8 cells in three biological replicates of each strain wer e harv ested by centrifugation at 4000 × g for 5 minutes.Pellets w ere w ashed twice with sterile w ater.Befor e l ysis, C17 Sphingosine and C17 Ceramide (d18:1/17:0), as internal standar ds, w ere added to each pellet and then lysed using glass beads (50 mg, 0.4-0.6 mm) in Fastprep ® (MP Biomedicals , C A, USA).Lipid extraction and base hydr ol ysis was performed using the methods described earlier by Kumar et al. ( 2021 ).Extr acted lipids wer e dried with N 2 flushing and stored at −20 • C until analysed.

Protein estimation
Protein estimation for normalizing lipid data was done using bicinc honinic acid (BCA) Pr otein Assa y kit (G-Biosciences , MO, USA).Fr om cell l ysate of eac h r eplicate, an aliquot of 25 μl was added to the working solution (200 μl) in 96-well plates, and absorbance was read at 590 nm.Serial dilutions of bovine serum albumin (G-Biosciences, MO, USA) were used for standard calibration curve .T he amount of protein (mg ml −1 ) was calculated from the slope of the standard calibration curve.

Liquid chroma togr aphy MS
Extr acted lipids wer e r esuspended in or ganic buffer (methanol containing 1 mM ammonium formate and 0.2% formic acid).A two-buffer mobile system, aqueous and organic was used.Aqueous buffer contained 2 mM ammonium formate and 0.2% formic acid.Fr om eac h sample, 5 μl was injected by the Autosampler, and mobile buffer was pumped at a flow rate of 0.3 ml min −1 to the HPLC fitted with the C8 column (Waters, MA, USA).SL species were detected by m ultiple r eaction monitoring (MRM) methods using QTRAP ® 4500 (SCIEX, USA) mass spectrometer.The MRM scans used were described earlier by Kumar et al. ( 2021 ).

Da ta anal ysis and sta tistical anal ysis
Mass spectrometric chromatograms were processed using MultiQuant TM software (SCIEX).Quantification of different lipid classes and species was done using the internal standard normalization method.The data was further normalized to per mg protein, and the amount of each lipid species was calculated as % of the total SL per mg pr otein.Thr ee biological r eplicates of each sample were used for all analyses.To determine statistical significance between the data sets, Student's t -test was used and a P -value of < .05 was considered significant.PCA was performed and plotted using OriginPr o ® softwar e.Data bars were plotted using Gr a phP ad Prism 8.

All the clade isolates show susceptibility to SL biosynthesis inhibitors
Our collection of clinical C. auris isolates fr om v arious clades comprises three isolates from Clade I (NCCPF 470156, NCCPF 470157, and VPCI 479/P/13), two isolates from Clade II (NCCPF 470296 and CBS 10913T), two isolates from Clade III (AR 0383 and AR 0384), thr ee isolates fr om Clade IV (AR 0385, AR 0386, and LMDM 1219), and one isolate from Clade V (AR 1097).Before conducting lipidomic analysis, we evaluated the susceptibility of all these isolates to SL biosynthesis inhibitors, specifically myriocin (which targets SPT) and aureobasidin A (which targets IPC synthase).Remarkably, all 11 isolates from different clades exhibited susceptibility to myriocin and aureobasidin A, with low MIC values ranging from 0.01 to 0.25 μg ml −1 (Table 1 ).This stands in contrast to C. glabrata and C. albicans r efer ence str ains, whic h demonstr ated higher MIC values with MIC of ∼4 μg ml −1 against myriocin and ∼0.5 μg ml −1 against Aureobasidin A. (Rollin-Pinheiro et al. 2021, Kumar et al. 2021 ).Consequentl y, it can be inferr ed that ther e is no interclade heterogeneity in the susceptibility of C. auris isolates to SL biosynthesis inhibitors.
Notably, the collection of 11 C. auris strains encompassed a variety of susceptible and resistance patterns to FLC and AmB.For instance, within clade I, all three isolates, namely NCCPF470156, NCCPF470157, and VPCI 479/P/13, wer e r esistant to FLC (Table 1 ), while both isolates within clade II, NCCPF 470296, and CBS10913T were susceptible to FLC.Within clade III, two isolates (AR 0383 and AR 0384) demonstr ated r esistance to FLC.In clade IV, two out of three isolates (AR 0385 and AR 0386) displayed high MIC values against FLC ( > 512 μg ml −1 ).LMDM 1219 in clade IV was susceptible to FLC but resistant to AmB (MIC 4 μg ml −1 ).All other isolates were susceptible to AmB with MIC's ranging from 0.5 to 1 μg ml −1 .The sole isolate of clade V was FLC resistant.Importantl y, r egardless of their resistance profiles, all drug-resistant isolates remained susceptible to SLs inhibitors thus underscoring the importance of SL metabolism in drug resistance.It has been already shown that by targeting the SL biosynthesis pathwa y using inhibitors , the MIC v alues of FLC decr eased by manifold (Rollin-Pinheiro et al. 2021 ).There are other studies where it has been demonstrated that creating blocks at specific steps in the SL biosynthesis alters the susceptibility of fungal cells against the common antifungal drugs such as FLC (Oura and Kajiwara 2010 ).
Our analysis confirmed that both the acidic and neutral br anc hes of the SL biosynthesis pathway are active across all C. auris clades .T his observation aligns with C. albicans and separates C. auris from other fungi as discussed abo ve .
Our anal ysis r e v ealed distinct c har acteristics in differ ent clades.Clade I stood out due to its notably higher levels of three sphingoid bases (DHS 2.84%, PHS 16.71%, and SPH 0.3%) and Glc-Cer (17.02%) when compared to the other three clades (Fig. 1 A).Conv ersel y, clade II exhibited the highest Cer content at 0.6%.Clade III, on the other hand, displayed the highest PCer content ( ∼80%) but had lo w er le v els of PHS and αOH Cer (0.8% and 0.18%, r espectiv el y) compar ed to the other clades.In contrast, Clade V sho w ed the highest dhCer content at 16.79%, while it was the least abundant in Clade IV isolates at 7.25% compared to all the clades.Notably, Clade V also had higher levels of αOH Cer (2.03%) and IPCs (0.52%).It's worth mentioning that IPC le v els, r epr esenting the acidic SL biosynthesis pathw ay, w er e onl y detected in tr ace  amounts, except for Clade V, which exhibited characteristically high le v els of IPC (Fig. 1 A).
To assess whether these variations in SL class composition effectiv el y distinguish the five clades, we conducted principal component analysis (PCA) on the r espectiv e SL datasets.As depicted in Fig. 1 (B) ( File S1 , Supporting Information ), the PCA plot distinctl y separ ated eac h clade fr om the others .PC A analysis extr acted thr ee principal components, namel y PC1, PC2, and PC3, which together accounted for 99.9% of the variance within the data.Clade assignments were based on PCA loading values, and SL class assignments were based on scores (Fig. 1 B).Clades I, II, and V displayed noticeable differences when considering the scores depicted in the plot.In contrast, Clades III and IV a ppear ed to be mor e closel y positioned on the plot due to their r elativ el y similar SL profiles.A similar close positioning was also observed in Clade II and V in the plot due to similar content of some SL classes.Specifically, PCer and PHS exhibited the highest scores on PC1 and PC2, aligning with the loading values associated with Clades III and I, r espectiv el y.On the other hand, dhCer attained a high scor e on PC3 and corresponded with Clades II and V, effectiv el y distinguishing these two clades from others (Fig. 1 B).

The distribution of SL classes varies among the C. auris isolates
To delve deeper into the differences in the composition of SL classes among the isolates, we performed a compr ehensiv e analysis that included all 11 isolates r epr esenting v arious clades, r egardless of their drug-susceptibility profiles.Out of the se v en SL classes and three sphingoid bases, PCer emerged as the predominant class in all 11 isolates.Notabl y, isolates fr om clade IV, specifically AR 0385 and AR 0386, exhibited the highest PCer content among all the isolates, with percentages of 86.4% and 86.1%, respectiv el y (Fig. 2 A; File S2 , Supporting Information ).The second most abundant SL class across six isolates from clades II, III, IV, and V was dhCer, including CBS10913T (18.14%) from clade II, AR 0383 (9.3%) and AR 0384 (11.3%) from clade III, two isolates from clade IV, AR 0385 (8.4%) and AR 0386 (8.7%), and clade V isolate AR 1097 (16.7%).In contrast, GlcCer emerged as the second most abundant SL class in LMDM 1219 (27.4%),NCCPF 470156 (21.16%),NCCPF 470157 (20.17%), and NCCPF 470296 (17%).PHS was the predominant sphingoid base in all isolates across various clades.Notably, clade I isolates, including NCCPF 470156, NCCPF 470157, and VPCI 479/P/13, displayed r elativ el y higher PHS le v els, r espectiv el y accounting for 19.7, 17.8, and 12.4% of the SL content when compared to isolates from other clades.Following these, LMDM 1219 (5.77%) and CBS10913T (3.96%) also sho w ed notable PHS lev-els, while AR 0383 and AR 0384 exhibited the lo w est PHS le v els at ∼0.85%.DHS, another sphingoid base, exhibited significant variation among the isolates.It was most abundant in three clade I isolates (ranging from 2.4% to 3.4%), follo w ed b y LMDM 1219 (1.5%) and AR 0384 (1.1%).For all other isolates, DHS le v els wer e less than 1%.αOH Cer content was highest in the clade V isolate (2%), follo w ed b y clade I isolates and LMDM 1219, with le v els r anging from 1.1% to 1.9%.In contrast, all other isolates exhibited αOH Cer le v els of less than 1%, with the lo w est le v els found in str ains AR 0383, AR 0384, AR 0385, and AR 0386, at 0.09%-0.19%.αOH PCer content was least abundant in the clade V isolate (0.17%) and AR 0385 (0.8%), while all other strains had content ranging from 1% to 3%.As observed in the polar heat map (refer to Fig. 2 A), SL classes such as SPH, Cer, and IPC were among the least abundant across all clades.NCCPF 470296 and CBS10913T had the highest Cer content, whereas its levels were notably low in isolates from clade III and two isolates from clade IV (AR 0385 and AR 0386).SPH was enric hed in thr ee clade I isolates and LMDM 1219, with both classes constituting less than 1% of the SL content.IPCs were detected in low amounts, except in the clade V isolate, where they were r emarkabl y high (0.52%) compared to all other isolates.

Molecular species of SL classes display different profiles among clade isolates
To c hec k how this v ariation of SL classes is constituted in eac h str ain, we anal yzed the molecular species of SL classes that contribute to this v ariation.Ev ery molecular species is a unique structure and differ from one another in the backbone, fatty acyl chain length, degr ee of satur ation, type of head gr oup and number of hydr oxyl gr oups pr esent (Usmani et al. 2023 ).Throughout different classes, fatty acyl chains having carbon chain length of 12-30 were detected in our analysis.Ho w ever, not all contributed equally to the ov er all lipid content showing that onl y fe w ar e pr eferr ed over others for acylation.Fatty acids with carbon chain lengths of 16, 18, 24, 26, and 28 sho w ed predominance in different classes and saturated fatty acids were highly preferred over the unsaturated ones ( File S3 , Supporting Information ).Except GlcCer, most of the SL classes had high abundance of molecular species containing 24, 26, and 18C long fatty acids follo w ed b y ones having 28 and 16 carbons (Fig. 2 B).In GlcCer, the 18C long fatty acid containing specie, GlcCer(d19:2/18:0(2OH))] is r emarkabl y very high follo w ed b y 16C containing GlcCer(d19:2/16:0(2OH))]. Also, this class contained species with different backbones like d18:1, d18:2, and d19:2 coming from different pools of Cer and αOH Cer.GlcCer(d19:2/18:0(2OH[R])) turned out to be the major species follo w ed b y GlcCer(d18:2/18:0(2OH[R])) and Glc- Cer(d19:2/16:0(2OH[R])).This tr end r emained acr oss all the isolates .T he upstr eam pr ecursor classes of neutr al br anc h such as dhCer , Cer , and αOH Cer had abundance of 16C , 18C , 24C, and 26C fatty acids.In dhCer, Cer(d18:0/24:0) was major dhCer specie across all isolates follo w ed b y Cer(d18:0/18:0) or Cer(d18:0/26:0).In Clade II and Clade V isolates, Cer(d18:0/26:0) was the second most abundant species while as in other isolates it was Cer(d18:0/18:0) (Fig. 2 B).Unlike the neutral branch, v ery long c hain fatty acids (VLCFA) having 24C and 26C are more pr eferr ed in the acidic pathway compared to 28C and 18C ones.As already mentioned above that PCer is major SL class in all isolates (Fig. 2 A), but the proportion of PCer molecular species is not uniform.In eight isolates of clade I, III, and IV the ratio of 24C PCer to 26C PCer [Cer(t18:0/24:0):Cer(t18:0/26:0)] was high (1.3-4)due to high pr e v alence of 24C fatty acid while as the same ratio in isolates of Clade II and V was (0.6-1).A similar trend was also observed in case of αOH PCer where the ratio of Cer(t18:024:0(2OH[R])):Cer(t18:0/26:0(2OH[R])) was high ( > 15) in clade I, III, and IV compared to clade II and V ( < 12.5) ( File S3 , Supporting Information ).

Drug-susceptible and drug-resistant isolates of all clades exhibit discernible differences in their molecular species profiles
As pr e viousl y mentioned, our sample collection primarily consisted of a mix of drug-susceptible and drug-resistant isolates acr oss v arious clades.Ho w e v er, an exception was observ ed in clade IV, where we obtained two drug-resistant and one drugsusceptible isolate .T his diversity in susceptibility profiles presented a challenge when directly comparing sphingolipidomic patterns between drug-susceptible and drug-resistant isolates fr om differ ent clades.Ne v ertheless, we pr oceeded with sphingolipidomic analysis on all 11 isolates encompassing various clades and conducted comparisons between three susceptible and eight resistant isolates across the clades .T he primary objective was to pinpoint specific variants within SL classes and species between the available pool of resistant and susceptible isolates.
Upon comparing av er a ge SL class contents betw een the tw o gr oups, man y classes sho w ed noticeable differences betw een them (Fig. 3 ).Ho w e v er, GlcCer and Cer displayed statisticall y significant differences between the two groups.On av er a ge, both Glc-Cer and Cer le v els wer e higher in the drug susceptible group.A close inspection r e v ealed six suc h species form GlcCer and five from Cer contributed to this significant v ariation.The le v els of all these species were distinctly higher in drug susceptible isolates .T he major GlcCer specie GlcCer(d19:2/18:0(2OH[R]) along with GlcCer(d18:1/20:0) and GlcCer(d18:1/18:0(2OH[R])) were ∼2fold higher and other GlcCer species were ∼1.9 fold higher in the susceptible group.Cer, although present in minute quantities sho w ed compar ativ el y mor e v ariation with Cer(d18:1/26:0) being more than 16-fold higher in the susceptible strains compared to the resistant ones.Other Cer species were 3-6-fold higher.The other SL classes that did not show significant differences between the two groups, but still had some molecular species that sho w ed significant variation between the two groups.
To emphasize the statistical significance of these variations, we conducted PCA on the r espectiv e datasets, allowing for the extraction of three components.When plotting the two primary components, PC1 and PC2, which accounted for 99.8% of the variance in the data, a clear differentiation in SL fingerprints between the susceptible and resistant strains emerged (Fig. 4C).The scor es demonstr ated that molecular species suc h as Glc-Cer(d19:2/18:0(2OH[R])) and PHS were abundant and aligned with the loading values of the susceptible str ain, effectiv el y separ ating it from the resistant strains, which were characterized by higher le v els of VLCFA containing dhCer , PCer , and others (Fig. 4 C; File S4, Supporting Information ).Notably, among the abundant 3. Compar ativ e anal ysis of drug-r esistant and -susceptible isolates.Variation in SL classes between drug-r esistant and -susceptible isolates (upper panel) and molecular species that display statistically significant variation between the two groups (lower panel).Bars r epr esent mean ( + SEM) value of given lipid classes or molecular species in drug-susceptible or drug-resistant isolates.Data on Y -axis represents % of total SL per mg protein ( * P < .05) .species, 21 SL species displayed statistically significant differences between the susceptible isolate and all two resistant isolates of clade IV (Fig. 4 ).

Discussion
The sim ultaneous a ppear ance of C. auris in m ultiple geogr a phical areas as different clades and the observed phenotypic and genetic differences between them with respect to pathogenicity and resistance to different antifungal drugs demands a closer look (Muñoz et al. 2018, Welsh et al. 2019 ).Why C. auris manifests an exceptionall y high fr equency and le v el of r esistance to w ar ds common antifungals and why some of the clades do not show a high percentage of drug resistance are some open questions that remain unanswered.C. auris has evolved in different geographical niches (Chow et al. 2020 ), which can result in dissimilarity in the membr ane c hemical arc hitectur e, and since membr ane lipids ar e also the target of common antifungals wherein, slight imbalances in lipid homeostasis impact the de v elopment of resistance .T he different clades of C. auris are believed to have independent evolutionary history.Clades have unique genomic, transcriptomic, and metabolomic signatures that differentiate the clades.(Chow et al. 2020, Brandt et al. 2023 ).Genomic analysis of the isolates belonging to different clades r e v eals Clade II to be the evolutionary oldest while as clade IV to be the recent one (Chow et al. 2020 ).We hypothesize that the different drug-susceptibility profiles may be attributed to dissimilar membrane lipid pr ofiles particularl y the SLs , as these pla y m ultiple r oles in yeast cells including stress tolerance and drug resistance.
The present work was planned to highlight the total landscape of SLs among different clades and drug susceptible and resistant isolates.For this, by employing the ESI-based LC-MS/MS technique, we compared the sphingolipidomes of five geographical clades consisting of ele v en clinical isolates of C. auris .Our analysis could identify lipid structures belonging to all major lipid classes of SL and show that all five clades possess composition distinct enough to separate them by m ultiv ariate anal ysis.In addition, these variations were reflected at the level of molecular species, whic h differ ed in fatty acid c hain lengths and the sphingoid backbone as well.Among different SLs, PCer levels emerged as the most varying SL class among all the clades, while SPH was the least varying class of SL.Clade I sho w ed maxim um v ariation Similar results with a higher amount of PCer were also found in our pr e vious study with C. auris , and with C. neoformans and C. gattii strains.PCer has been shown to be critical in maintaining an a ppr opriate plasma membr ane envir onment for the pr oper functioning of membrane proteins in C. neoformans (Farnoud et al. 2014, Kumar et al. 2021 ).Considering their abundance levels, it is likely that PCer is also structurally indispensable in C. auris .
A comparison worldwide isolates has also brought attention to clade-specific variance in drug resistance levels and resistancer elated mec hanisms.While most isolates from clade II are usually susceptible to azoles and other antifungals, almost all isolates from clades I, III, and more than half of the isolates from clade IV ar e r esistant to azoles (Loc khart et al. 2017, Chowdhary et al. 2018, Chow et al. 2020 ).We aimed to exploit the diverse resistance profile of C. auris to search for lipid fingerprints associated with resistance that could be unique to the different clades.Ho w ever, the unavailability of drug-susceptible isolates in all the clades and resistant isolates in Clade II was a major dr awbac k.Ne v ertheless, a broad comparison of the susceptible and resistant isolates across the clades provided some insights.For example, there was higher abundance of total Cer and GlcCer in the FLC susceptible isolates compared to the resistant ones .T he same distinction was also observ ed, with r espect to GlcCer, in clade IV, where the intraclade comparison of FLC susceptible and resistant isolates was possible, suggesting the yet unexplored roles of these lipid molecules in determining the susceptibility.Variations in common species within GlcCer between two sets , ma y likely establish GlcCer as another marker of r esistance de v elopment in C. auris .Apart from these common finger prints, man y other molecules sho w ed variation in Clade IV intraclade comparison and in cross clade comparison.Perturbation of these unique lipid species could be associated with FLC r esistance, pr ovided suc h pr ofiling is done on a lar ge scale .T he variation based on carbon chain length in fatty acids in r espectiv e SL classes also hints at the roles of fatty acyl elongases.Pr e vious studies have shown that the deletion of FEN1 / SUR4 can lead to AMB susceptibility.FEN1/12 are used to synthesize VLCFA in yeasts (Sharma et al. 2014 ).In clade IV, the FLC-resistant isolates also sho w ed an accumulation of VLCFA containing SLs and could indicate SLs' role in the resistant trait of C. auris .In addition, FLC r esistant str ains of clade IV had r educed le v els of some Glc-Cer species, which were also observed previously in FLC resistant strains of clade I (Kumar et al. 2021 ).Altogether, clades can be separated based on the high and low abundance of different classes of SLs.Various transcriptomic and metabolomic data in clinical and FLC-adapted C. auris show dysregulation in SL-related genes and alteration in SL intermediates, strengthening the role of SLs in establishing C. auris -resistant traits (Zamith- Miranda et al. 2019, Jenull et al. 2021, Narayanan et al. 2022 ).
At this stage, it is difficult to comment on the physiological rele v ance of compositional variations among SLs .T he level of dissimilarity in the sphingolipidome between the five clades points to w ar ds another le v el of diversity, suggesting separate levels of SLmediated regulation in different clades.It would be interesting to explore whether these variations control clade specific azole resistance profiles .For instance , the high le v el of LCBs like PHS in clade I drug-resistant isolates could be a major determinant.LCBs have been shown to get induced during stress (Vandenbosch et al. 2012 ).The major sphingoid base PHS is known to activate different protein kinases in the cell such as Pkh1/2 and its downstream effectors Ypk1/2, Pkc1, and Sch9 (Liu et al. 2005 ).Pkc1 via MAPK cascade mediates FLC tolerance in C. albicans (LaFayette et al. 2010 ).Similarly in Clade III and IV, FLC-resistant isolates recorded very high PCer le v els in our anal ysis.PCer is synthesized by acylation of PHS by ceramide synthases.Deletion of ceramide synthase LAG1 increased susceptibility of C. albicans to w ar ds FLC (Gao et al. 2018 ).Cer amide-activ ated pr otein phosphatases suc h as Sit4 hav e been shown for modulating m ultidrug r esistance in yeast via upregulation of PDR genes (Miranda et al. 2010 ).All suc h m ultilayer ed regulatory networks and others mediated by SLs could possibly be responsible for different resistant profiles of C. auris clades.
Our analysis is limited to only 11 isolates; hence more in-depth analysis of sphingolipidomes with more isolates will be required to establish a ppar ent dissimilarities among clades.Mor eov er, the r ele v ance of c har acteristics of SLs in differ ent clades coming fr om div erse nic hes and among drug-susceptible and -r esistant isolates will r equir e a mor e extensiv e anal ysis for its r ele v ance and v alidation.Nonetheless, based on drug susceptibility assays using myriocin and aureobasidin A, it can be inferred that C. auris cannot survive without SL synthesis.It also implies that probably no other lipid classes can compensate for the loss of SL structures in myriocin or aureobasidin A-treated cells, which makes the SL biosynthetic pathway a prime target to treat C. auris infections.

Figure 1 .
Figure 1.SL class composition differentiates the geographical clades.(A).SL biosynthesis pathway of C. auris and le v els of different SL intermediates (mean + SEM) in each of the five clades as detected by ESI-LCMS/MS.Data on Y -axis represents % of total SL per mg protein).Clades showing significant variations are marked * P < .05,* * P < .01,and * * * P < .00.DHS-dih ydrosphingosine, PHS-ph ytosphingosine, dhCer-dih ydr ocer amide, Cer-ceramide, αOH Cer-alpha hydroxyl ceramides, αOH PCer-alpha hydroxyl phytoceramides , GlcCer-glucosylceramides , and IPC-inositol phosphorylcer amides.Ov als r epr esent the thr ee major sphingoid bases (18-carbon long amino-alcohol), the r ectangles r epr esent the SL classes deriv ed fr om fatty acylation of sphingoid bases.Hexa gons r epr esent the complex SLs having a head gr oup at C1 position of the sphingoid base.(B) 3D PCA biplot showing separation of five clades of C. auris based on the composition of different SL classes.Vectors represent the projection of clades based on loading values on three axes PC1, PC2, and PC3.SL classes (r ed) occupy positions in the plot based on scor es ac hie v ed in PCA.

Figure 2 .
Figure 2. SL distribution in C. auris isolates.(A) Polar heatmap showing SL class composition in the eleven clinical isolates of C. auris .The colour gradient is based on log 10 scale of the lipid proportion of each class in all isolates.(B) Heat map depicting variation of SL molecular species, which were abundant and differed highly between the isolates numbered on top 1-11: NCCPF470156, NCCPF 470157, VPCI 479/P/13, NCCPF 470296, CBS10913T, AR 0383, AR 0384, AR 0385, AR0386, LMDM 1219, and AR 1097, r espectiv el y.Thr ee r eplicates of eac h isolate ar e shown in the heatma p.

Figure 4 .
Figure 4. Clade IV drug-susceptible -resistant strains are differentiated by specific lipid imprints.Bar graphs representing the molecular species that were abundant and showed statistically significant differences ( * P < .05,* * P < .01) between the drug-susceptible and -resistant isolates of clade IV.Eac h bar r epr esents the mean + SEM of three biological replicates.Data on the Y -axis r epr esents % of total SL per mg protein) (A) Molecular species within different classes that were abundant in the drug susceptible isolate LMDM 1219.(B) Molecular species that were abundant in drug-resistant isolates AR 0385 and AR 0386.(C) PCA biplot showing separation of the drug-susceptible isolate from the resistant isolates based on the overall lipid composition of three isolates of clade IV.Vectors (arrows) represent projection of isolates based on loading values obtained on PC1 and PC2.Red dots show position of molecular species in the plot based on PCA scores.

Table 1 .
List of clinical isolates of C. auris and their drug susceptibility profiles.