Clinical isolates of Candida auris with enhanced adherence and biofilm formation due to genomic amplification of ALS4

Candida auris is an emerging multidrug-resistant fungal pathogen and a new global threat to human health. A unique morphological feature of this fungus is its multicellular aggregating phenotype, which has been thought to be associated with defects in cell division. In this study, we report a new aggregating form of two clinical C. auris isolates with increased biofilm forming capacity due to enhanced adherence of adjacent cells and surfaces. Unlike the previously reported aggregating morphology, this new aggregating multicellular form of C. auris can become unicellular after treatment with proteinase K or trypsin. Genomic analysis demonstrated that amplification of the subtelomeric adhesin gene ALS4 is the reason behind the strain’s enhanced adherence and biofilm forming capacities. Many clinical isolates of C. auris have variable copy numbers of ALS4, suggesting that this subtelomeric region exhibits instability. Global transcriptional profiling and quantitative real-time PCR assays indicated that genomic amplification of ALS4 results in a dramatic increase in overall levels of transcription. Compared to the previously characterized nonaggregative/yeast-form and aggregative-form strains of C. auris, this new Als4-mediated aggregative-form strain of C. auris displays several unique characteristics in terms of its biofilm formation, surface colonization, and virulence.


Introduction
Program of China (2021YFC2300400 to GH), and by the National Institutes of Health (NIH) National Institute of General Medical Sciences (NIGMS) award R35GM124594 and the Kamangar family in the form of an endowed chair to CJN. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
subtelomeric regions, are often subject to genetic and epigenetic regulation to generate cellsurface variation in S. cerevisiae [17].
In the current study, we report a new aggregating form of C. auris with strongly enhanced biofilm forming capacity. This multicellular morphology is the result of increased adherence of adjacent cells and is not due to defects in cell division or the failure to release daughter cells. Whole genome sequencing experiments demonstrated that the genomic amplification of the cell wall adhesin gene ALS4 is responsible for this new C. auris aggregating morphology. Further investigation indicated that ALS4 is located at the subtelomeric region of chromosome 5 and exhibits a high level of sequence and copy number variations (CNVs) in clinical C. auris isolates.

Isolation and identification of the nonaggregative/yeast-form and aggregative-form strains of C. auris
We isolated a C. auris nonaggregative-form (yeast-form) strain (SJ01) and an aggregativeform strain (SJ02) simultaneously from a urine sample of a long-term care female patient (aged 81), who had been admitted to the Department of Respiratory Medicine at the Shengjing Hospital of China Medical University (Shenyang, China) and had been diagnosed with pneumonia. The patient received long-term antibiotic treatment therapy for >1 year and had a history of hypertension.
The strains SJ01 and SJ02 were initially identified as C. auris by MALDI-TOF mass spectrometry and were further verified by whole genome sequencing. The two strains belong to C. auris clade III (the South African clade) and are closely related, with only 38 single-nucleotide polymorphisms (SNPs) and 90 short insertions/deletions (INDELs) in total across their genomes (S1 Table and S1 Dataset), indicating their close genetic relationship.

Morphological and virulence analyses of the nonaggregative/yeast-form and aggregative-form strains of C. auris
To compare the biological features of strains SJ01 and SJ02, the strains were plated onto solid YPD rich medium containing the red dye phloxine B and incubated at 30˚C for four days. Strain BJCA001A (Agg-2), a derivative of strain BJCA001, containing previously reported typical aggregative-form cells [18], which displayed a clear defect in cell division [7], served as a reference strain. As shown in Fig 1A, both SJ01 and SJ02 strains formed white and shiny colonies, whereas the typical aggregating isolate BJCA001A formed red and rough colonies on YPD plates containing phloxine B. We next examined the cellular morphologies of strains SJ01 (yeast), SJ02 (Agg-1), and BJCA001A (Agg-2) by differential interference contrast (DIC) microscopy ( Fig 1B) and scanning electron microscopy (SEM) (Fig 1C). Strain SJ01 contained only single yeast-form cells of C. auris, while strains SJ02 and BJCA001A both formed clear multicellular aggregates. Interestingly, the SEM assays indicated that BJCA001A and SJ02 formed obviously distinct aggregating morphologies (Fig 1C). The cells of strain BJCA001A appeared unable to undergo physical separation during the cell division process leading to the formation of "clumps", whereas the cells of strain SJ02 were able to be separated and appeared only to adhere to one another through cell-cell interactions. To further distinguish between these two aggregating morphologies of C. auris, we performed additional SEM assays. As shown in Fig 1C, the mother and daughter cells of strain BJCA001A, but not SJ02, exhibited obvious cell wall division defects.

PLOS PATHOGENS
A new aggregating form of C. auris To degrade extracellular proteins, including adhesins, we next treated the cells of strains SJ01, SJ02, and BJCA001A with proteinase K and trypsin. After treatment with either enzyme, the cell clumps of strain SJ02 separated into single yeast-form cells, while no obvious changes were observed in the aggregating morphologies of strain BJCA001A (Fig 1B).
These results indicate that the two aggregating morphologies of C. auris strains SJ02 and BJCA001A are due to different mechanisms. Specifically, the aggregating form of strain Colony and cellular morphologies of the nonaggregative/yeast-form (SJ01), Agg-1 (SJ02), Agg-2 (BJCA001A), Agg-1Re (SJ02Re), and XM03-1 strains of C. auris. Strains SJ01 and SJ02 were isolated from the same patient. BJCA001A is the aggregating form of strain BJCA001. The nonaggregative/yeast-form strain SJ02Re was spontaneously generated form cells of strain SJ02 grown on YPD medium. Strain SJ02Re carries a single copy of ALS4. Strain XM03-1 is a clinical isolate that is missing ALS4. (A) Colony morphologies. C. auris cells were plated on YPD medium containing 5 μg/mL phloxine B for 4 days at 30˚C. Scale bar = 5 mm. (B) DIC images of C. auris cells treated with ddH 2 O, proteinase K, or trypsin. C. auris cells of a single colony of each strain were washed with 1 x PBS and then treated with ddH 2 O, proteinase K, or trypsin for 1 hour at 37˚C. Scale bars,10 μm. (C) Low and high magnification SEM images of C. auris cells. The culture condition was the same as used in panel A. Scale bar,10 μm. Strains SJ02Re and XM03-1 served as controls. https://doi.org/10.1371/journal.ppat.1011239.g001

PLOS PATHOGENS
A new aggregating form of C. auris SJ02 represents a distinct multicellular morphology resulting from cell surface proteinmediated cell-cell adhesion.
Since cell aggregation has known effects on virulence of C. auris strains, we next evaluated the virulence of strains SJ01 (yeast-form), SJ02 (Agg-1), SJ02RE (yeast-form), BJCA001 (yeast), and BJCA001A (agg-2) using a Galleria mellonella infection model. As shown in Fig 2, compared to the yeast-form cells of SJ01, SJ02RE, and BJCA001, the two aggregating forms of SJ02 and BJCA001A both exhibited decreased virulence in the model. This attenuated virulence observed for these two strains could be due to a reduction in cellular dissemination in the G. mellonella larval bodies, which is a consequence of enhanced aggregation. Survival curves of G. mellonella larvae infected with C. auris nonaggregative/yeast-form (SJ01 and BJCA001) and aggregative-form (SJ02 and BJCA001A) strains. Approximately 1 x 10 6 C. auris cells of each strain were injected into each G. mellonella larva. Ten larvae were injected for each strain. ns, no significant difference; � , significant difference (p<0.05, by log rank test. Strain SJ02 served as the control

PLOS PATHOGENS
A new aggregating form of C. auris

Aggregative-form to nonaggregative/yeast-form transition in C. auris strain SJ02
Interestingly, when grown on solid media for an extended culture period (�5 days), some colonies of the aggregative-form strain SJ02 formed sectored colonies. When observed by microscopy, cells from those sectors were in the single-celled nonaggregating phenotype (i.e., they "returned" to the nonaggregative/yeast-form, strain SJ02Re, Fig 3A), suggesting that the aggregating phenotype of strain SJ02 is able to switch to the nonaggregative-form at a high frequency (~7%). However, no nonaggregative-form to aggregative-form switching was observed in either strain SJ01 or the "returned" nonaggregative-form strain SJ02Re (frequency < 0.02%, Fig 3B and 3C).

PLOS PATHOGENS
A new aggregating form of C. auris

Comparative analysis of biofilm formation in the nonaggregative/yeastform and aggregative-form strains of C. auris
Prior work has shown that the ability to form aggregates influences biofilm development in C. auris [4,7,10]. However, both promoting and inhibitory effects of cell aggregation on biofilm development have been reported [4,7,10]. We next tested the biofilm forming abilities of strains SJ01, SJ02, SJ02Re, BJCA001A, and XM03-1 to develop biofilms on polystyrene plates and silicone squares. Comparatively, strain SJ02 developed robust biofilms, while the typical aggregative-form strain (BJCA001A) developed weak biofilms on both polystyrene and silicone substrates (Fig 4A). The nonaggregative-form stains SJ01, SJ02Re, and XM03-1 exhibited comparable abilities to form biofilms, and were much weaker than biofilms formed by strain SJ02 (Fig 4A). Consistently, SEM microscopy assays performed on silicone squares indicated the same order of biofilm forming abilities of the three strains (SJ02 > SJ01 (SJ02Re or XM03-1) > BJCA001A, Fig 4B). To further evaluate the biofilms, we performed colony forming unit (CFU) assays on the biofilms formed on the bottoms of polystyrene plates, and verified the strong biofilm forming ability of strain SJ02 (Fig 4C). These findings correlate the new aggregating morphology of strain SJ02 with an enhanced capacity for biofilm development.

PLOS PATHOGENS
A new aggregating form of C. auris

Enhanced ability of skin colonization of the new aggregative-form strain of C. auris
Since adhesion and biofilm formation are important features in the colonization and infection of host skin and tissues, we next performed skin colonization/infection experiments with strains SJ01, SJ02, SJ02Re, BJCA001A, and XM03-1 using a newborn mouse skin infection model. C. auris cells of each strain were inoculated on the back skin regions of newborn mice. After 24 hours of infection, the colonized skin regions were gently washed with 1 x PBS and excised for SEM assays. Consistent with the results of biofilm development, the aggregativeform strain SJ02 exhibited the most robust skin colonization ability and formed thick biofilmlike structures (Fig 5). The nonaggregative-form strains SJ01, SJ02Re, and XM03-1 demonstrated relatively weak abilities to colonize skin. Interestingly, strain BJCA001A showed an intermediate ability to colonize skin in this infection model and its colonization led to the formation of numerous small pits on the skin tissue. One possible explanation for this phenomenon is that strain BJCA001A belongs to the South Asian clade (clade I, MTLa) and the other four strains belong to the South African clade (clade III, MTLα). We previously reported that C. auris strains of the South Asian clade secrete more secreted aspartyl proteases than strains of the South African clade at low temperatures [19]. Secreted aspartyl proteases could damage the skin and lead to the formation of small pits.

Genomic analyses reveal that amplification of ALS4 contributes to the aggregating phenotype of strain SJ02
To uncover the underlying mechanism of the formation of the new aggregating phenotype in C. auris, we sequenced the genomic DNA of strains SJ01, SJ02, SJ02Re, BJCA001A, and XM03-1 using next-generation sequencing approaches. The number of pairwise SNPs and associated mutated genes between strains SJ01, SJ02, and SJ02Re were fewer than those between strains isolated from different patients in China of the same genetic clade (S1 Table). Based on the biological roles of the involved genes, we did not find obvious mutations caused by SNPs or INDELs that were associated with the aggregating phenotypes (S1 Table and S1 Dataset).
We next performed copy number variation (CNV) analysis and found substantial amplification of the ALS4 gene (10 copies) in strain SJ02 (Fig 6). For strain SJ02, we found that eight

PLOS PATHOGENS
A new aggregating form of C. auris copies of ALS4 were full-length and just two copies of ALS4 were nearly full-length with~460 bp missing at the 3' end of the ALS4 ORF. The ALS gene family encodes a set of cell wall glycoproteins in several pathogenic Candida species that are often involved in the regulation of adherence to surfaces and biofilm formation [14]. We speculated that the genomic amplification of ALS4 contributed to the aggregating phenotype of C. auris strain SJ02. Consistent with our hypothesis, the nonaggregative-form strain SJ01 had a shortened copy of ALS4, and the "returned" nonaggregative-form strain SJ02Re and typical aggregative-form strain BJCA001A each carried a single copy of ALS4 that was nearly full length (Fig 6A and 6B). We note that since strain SJ01 does not carry a full-length copy of ALS4, it could not be the parental strain of isolate SJ02, which carries ten copies of full-length/nearly full-length ALS4. Interestingly, no sequence signal of ALS4 was detected in strain XM03-1, suggesting that ALS4 has been completely lost in this strain (Fig 6A and 6B).
To further compare the genomic organization of the nonaggregative-form strain SJ01 and aggregative-form stain SJ02, we sequenced the genomes of strains SJ01 and SJ02 using longread sequencing and assembled their genomes. Similar genomic sizes (~12.4 Mb) and chromosome numbers (7) were observed for the two strains and their genomes exhibited overall high levels of synteny (Fig 7). Moreover, we observed that there were many polyA/polyT sequence

PLOS PATHOGENS
A new aggregating form of C. auris motifs located adjacent (on both sides) of the ALS4 repeat, possibly promoting the expansion of the ALS4 gene. Overall, the long-read sequencing assays confirmed that the tandem-repeat amplification of ALS4 contributed to the aggregative phenotype of strain SJ02. A comparative

PLOS PATHOGENS
A new aggregating form of C. auris analysis of the genomes of strains SJ01, SJ02, and reference strain B11221 [20] is shown in Fig 7 and S1 Dataset.
The comparative genomic analysis demonstrated that a number of genes including IFF5 and LIP1, adjacent to the ALS4 locus, were missing in strains SJ01, SJ02, SJ02Re, and XM03-1 (possibly due to the instability of the subtelomeric region), but were present in strains BJCA001A and B11221 (S2 Table). IFF5 encodes a GPI-anchored adhesin-like protein, and LIP1 encodes a secreted lipase that is transcriptionally induced during biofilm development. Since these two genes were missing in the genomes of both strains SJ01 and SJ02, the absence of IFF5 and LIP1 could not be associated with the ALS4-mediated adherence and biofilm formation phenotypes observed in strain SJ02.

RNA-Seq and quantitative transcriptional expression analyses
To further confirm the relationship between ALS4 and the aggregative phenotype, we performed transcriptomic analysis by RNA-seq for strains SJ01 and SJ02. As shown in Fig 8A and  S3 Dataset, only 20 genes were found to be differentially expressed between the nonaggregative-form and aggregative-form of strains SJ01 and SJ02 (using a twofold cutoff). Of them, 14 genes were upregulated in cells of strain SJ02 and 6 were upregulated in cells of strain SJ01. As expected, the relative expression level of ALS4 in strain SJ02 was much higher than that of strain SJ01. The average FPKM value of ALS4 in strain SJ02 was about 400 times higher than that of strain SJ01 (Fig 8B). Quantitative transcriptional expression assays verified that ALS4 was highly expressed in strain SJ02, whereas the relative expression levels of ALS4 in cells of SJ02Re and BJCA001A were comparable to that of cells of strain SJ01 (not shown). These results suggest that the genomic amplification of ALS4 in C. auris is a likely reason for the increased aggregation and biofilm formation observed in strain SJ02.
Other than ALS4, the majority of differentially expressed genes only showed a twofold to threefold difference in expression levels between the two strains. The orthologs of 12 other differentially expressed genes have been annotated to be associated with biofilm development in C. albicans (Fig 8B and S2 Dataset; http://www.candidagenome.org/). Eight genes are

PLOS PATHOGENS
A new aggregating form of C. auris involved in metabolism processes, especially mitochondrial metabolism (e.g., CYB2 and MAE1). Interestingly, the carbonic anhydrase-encoding gene NCE103, which is required for the conversion of CO 2 to bicarbonate in fungi, was upregulated in cells of the aggregativeform strain SJ02. Aggregation could hinder CO 2 diffusion in C. auris cells. Given the critical role of CO 2 in the regulation of metabolism and morphological transitions, the increased expression of NCE103 in the aggregated cells of C. auris was not unexpected.
To verify the promoting role of ALS4 in the adherence and biofilm formation in C. auris, we constructed an ectopic expression plasmid pTDH3-ALS4 (Fig 9A). As predicted, ectopic expression of ALS4 under the strong TDH3 promoter in strain SJ01 notably enhanced the formation of cell aggregates and biofilms (Fig 9B and 9C). The enhanced capacity of biofilm formation was confirmed by quantifying C. auris CFUs obtained from the biofilms formed on the

PLOS PATHOGENS
A new aggregating form of C. auris bottoms of 24-well plates (Fig 9D). Consistently, quantitative real-time PCR (Q-RT-PCR) assays demonstrated that the relative expression level of ALS4 in the ALS4-ectopic strain was significantly higher than that of the control strain (Fig 9E). Together, these results confirm the role of Als4 in adherence and biofilm formation in C. auris.

Copy number variation of ALS4 is common across C. auris clinical isolates
As mentioned above, we isolated a C. auris strain (XM03-1, Figs 1 and 6A and S2 Table) from a patient with candidemia that lost ALS4, supporting the instability at this locus. To support the idea that copy number variation of ALS4 is common in clinical isolates, we analyzed the genome sequences of 1156 C. auris strains available in the NCBI database (https://www.ncbi. nlm.nih.gov/). We found that 26 strains carried more than two copies of ALS4 (2.3%, CNV � 2), 81 strains carried 1.5-2.0 copies (7%, 1.5� CNV< 2), and 5 strains carried less than 0.5 copies of ALS4 (0.4%, CNV � 0.5, Fig 6C and S2 Dataset). These findings provide additional support that ALS4 copy number variation is common among C. auris clinical isolates.
We and our collaborators recently published a report on a C. auris outbreak in a general hospital in China [21]. We revisited the phenotypes of the associated C. auris strains from this recent study and found another clinical isolate exhibiting enhanced adherence. As shown in Fig 10A, four strains (A101, A102, A103, and A104) were isolated from urine samples of the same patient at different time points. Similar to strain SJ02, strain A103 showed an aggregating phenotype and formed enhanced biofilms on both the bottoms of 24-well plates and silicone squares (Fig 10B and 10C). Genomic analysis demonstrated that strain A103 carries seven copies of ALS4, while the other three strains contain a single copy of ALS4 (Fig 10D). We further found that there were only three to eight SNP differences among the four isolates, suggesting that these strains originated from the same ancestor (Fig 10E). Taken together, our findings indicate that the ALS4 locus is unstable in C. auris clinical isolates and is able to undergo copy number changes in vivo.

Discussion
Adhesion is important not only for the development of multicellular aggregations but also for pathogen-host interactions. In pathogenic Candida species, the Als proteins are large cell-surface glycoproteins that play critical roles in the processes of adhesion to host and abiotic surfaces, aggregative behaviors, and biofilm development [14,15,22]. In the present study, we investigated variations of the ALS4 gene in C. auris clinical isolates and identified two clinical strains (SJ02 and A103) with a new aggregative phenotype due to enhanced adherence of adjacent cells. This aggregative morphology differs from the previously reported aggregative phenotype [7], which is caused by cell division defects. This new aggregative morphology, is not the result of cell division defects and plays roles in the adherence of C. auris to host and abiotic surfaces, and in virulence in the G. mellonella infection model (Figs 1,2,4,5,9, and 10). Genomic analyses indicated that amplification of ALS4 conferred the strains SJ02 and A103 with an enhanced ability to adhere to surfaces and form biofilms. Unlike C. albicans, which is commensal to the gastrointestinal and genitourinary tracts, C. auris is considered to be a largely skin colonizer [23]. Since Als4 is a cell surface adhesin, the amplification of ALS4 could be important in both the commensal and pathogenic life-styles of C. auris.
There are nine members of the Als family (Als1-9) in C. albicans, and the ALS genes display variability in transcriptional regulation, gene size, copy numbers, and additional coding regions [14]. Only three ALS orthologs (B9J08_004112/ALS4, B9J08_002582, and B9J08_004498) have been found in C. auris, all of which are orthologs to C. albicans ALS4. In

PLOS PATHOGENS
A new aggregating form of C. auris the present study, we focused on the biological function of C. auris B9J08_004112/ALS4, which exhibits a high variation in copy number in C. auris clinical isolates (Fig 6 and S2  Dataset).
We collected five C. auris isolates (SJ01, SJ02, SJ02Re, BJCA001A, and XM03-1) and performed next-generation and long-read genome sequencing experiments. We found that the ALS4 gene exhibits obvious copy number or sequence variability in the five isolates. Strains SJ01 and SJ02 were isolated from the same patient with a long-term urinary tract infection,

PLOS PATHOGENS
and strain XM03-1 was isolated simultaneously from a patient with a C. auris bloodstream infection [19]. SJ02Re and BJCA001A were derivatives of strains SJ02 and BJCA001 [18], respectively. Strain SJ01 bears a truncated version of ALS4, whereas strain SJ02 carries ten tandem copies of the full-length gene. Based on comparative genomic and phylogenetic analyses, SJ01 and SJ02 appear to be two isolates that recently diverged from a common ancestor. The "returned" nonaggregative/yeast-form strain SJ02Re carries a slightly shortened version of ALS4, strain BJCA001A contains a nearly full-length version of ALS4, and ALS4 was completely missing in strain XM03-1 (Fig 6A). Copy number variation analysis of the genomic sequences available in the NCBI database verified that the ALS4 gene is highly variable among C. auris clinical isolates (Fig 6 and S2 Dataset). A revisit of the biological and genomic features of another clinical isolate (A103) that was recently published in a C. auris outbreak in China revealed that like SJ02, A103 also exhibits similarly enhanced adherence and biofilm formation abilities. We found that A103 contains seven copies of ALS4 (Fig 10). These results suggest that the variation of the ALS4 locus is common among clinical isolates and that this subtelomeric region may be subject to rapid change in C. auris. In many other fungi, subtelomeric regions tend to exhibit a high rate of rearrangement and recombination, such as, for example, in the FLO genes of S. cerevisiae [17] and the EPA genes of C. glabrata [24]. Another example is the tandem amplification of the telomere-adjacent gene ARR3, which encodes an arsenite efflux transporter in Cryptococcus neoformans [25]. The subtelomeric regions of eukaryotic organisms are often evolutionarily active and contain repetitive DNA sequences. Genomic rearrangements occur frequently in these regions due to high rates of recombination or double-strand breaks, which cause genomic instability. For example, the FLO genes at the subtelomeric region in S. cerevisiae exhibit a high frequency of intergenic recombination [26,27]. The loss and gain of ALS4 in C. auris could be due to a similar mechanism to that observed for S. cerevisiae. Moreover, the tandem repeat sequences of ALS4 in C. auris could also facilitate recombination and rearrangement of the subtelomeric region. Together, microevolution of the subtelomeric regions may represent a general strategy of fungi to adapt to rapidly changing environments.
Given the adherence and hydrophobic characteristics of Als4, it is reasonable that the amplification of the ALS4 gene in C. auris results in enhanced biofilm development and skin colonization. The transcriptional expression level of ALS4 has been further amplified in the strain SJ02, perhaps due to the expansion of copies of ALS4 to a relatively distant region of the telomere. The repressing effect caused by subtelomeric epigenetic factors could be nullified in the ALS4 copies that are telomere distant. Global transcriptional expression analyses demonstrated that only 20 genes including ALS4 showed a twofold or greater change in expression between the nonaggregative/yeast-form strain SJ01 and aggregative-form strain SJ02 (Fig 8  and S2 Dataset). Of these differentially expressed genes, several are annotated to be involved in the regulation of biofilm development in C. albicans. These results suggest that the amplification of ALS4 in the subtelomeric region has a limited overall impact on the global transcriptional expression profiles of C. auris, but dramatically increases the expression of the Als4 adhesin, thus enhancing adherence and biofilm development.
In this study, we demonstrate that there are two distinct types of aggregating phenotypes in C. auris. The typical aggregative phenotype is due to a defect in cell division and in the release of budding daughter cells and has been reported by several other groups in prior studies. The other new aggregative phenotype that we report here is due to increased cell-cell adhesion likely resulting from the expansion of the adhesin gene ALS4. Overall, our findings provide new insights into the understanding of the biology and virulence features of this important emerging human fungal pathogen. However, many open questions remain to be addressed. For example, does the rapid microevolution of the ALS4 locus contribute to the high

PLOS PATHOGENS
A new aggregating form of C. auris transmission capacity of C. auris within healthcare settings? Is the variation of Als4 associated with the persistence of the fungus on skin and environmental surfaces, thus increasing the potential for hospital outbreaks? How did amplification of ALS4 occur? Does the host environment affect genomic stability of C. auris?

Ethics statement
All animal experiments were performed according to the guidelines approved by the Animal Care and Use Committee at Fudan University. The present study was approved by the Committee.

C. auris isolates and culture conditions
Clinical strains SJ01 (nonaggregative/yeast-form) and SJ02 (Agg-1, aggregative-form) were isolated simultaneously from the urine of a hospitalized patient in the Shengjing Hospital of China Medical University (Shenyang, China). Clinical strain XM03-1 was isolated from the first affiliated hospital of Xiamen University [19]. Strain BJCA001A (Agg-2, aggregative-form) was isolated from the mouse spleen following a systemic infection [6]. Strains A101, A102, A103, and A104 were isolated from urine samples of the same patient in a general hospital in China [21]. YPD medium (10 g/L yeast extract, 20 g/L peptone, and 20 g/L glucose), a rich medium, was used for routine growth of C. auris. To distinguish the colony morphology of nonaggregating/yeast-form and aggregating-form cells, phloxine B (5 μg/mL) was added to YPD medium.

Whole genome sequencing and analysis
Single colonies of C. auris strains were inoculated into liquid YPD medium and grown at 30˚C for 24 h. Fungal cells were collected, and genomic DNA was extracted using the TIANamp Yeast DNA Kit (TianGen Biotech, Beijing, China) according to the manufacturer's protocol. Whole genome sequencing assays were performed by Berry Genomics Co., Beijing, China. The methods for library construction, genomic sequencing, and SNP and INDEL analyses were based on our previous publication [28]. The clean reads were mapped to the genomic assembly of C. auris strain B11221 (NCBI accession number: GCF_002775015.1) using BWA mem 0.7.17 software with default settings [29]. SAMTools v1.361 [30], Picard Tools v1.56 (http://picard.source-forge.net), and GATK v2.7.2 were used for variation analyses [31,32]. The sorted BAM datasets were used for copy number variation (CNV) analyses of all coding genes with command "samtools depth" across certain gene regions.
Single molecule real-time sequencing was performed by Beijing Novogene Bioinformatics Technology Co., Ltd. (Beijing, China) using the PacBio Biosciences (PacBio) Sequel small-molecule real-time sequencing system (Pacific Biosciences, Menlo Park, CA, USA). The sequence data generated from the Illumina platform were used to proofread the PacBio assembly sequence. The PacBio sequence data were error-corrected, trimmed, assembled, and scaffolded using Canu v1.8 software guided by a genome size of 12.5 Mb [33]. Quast v5.0.2 was used to determine the quality of the sixteen newly assembled genomes [34]. AUGUSTUS v3.2.1 was used to predict proteins [35]. Putative ORFs were searched against the NR database of NCBI (http://www.ftp.ncbi.nlm. nih) and the Candida genome database (http://www.candidagenome.org/).
For the assembly of the ALS4 tandem structure and its adjacent region, PacBio long reads were first mapped to the genomic assembly of C. auris strain B11221 using the minimap2 v2.17-r941 software [36] with default settings. The sequence depth were analyzed with IGV

PLOS PATHOGENS
A new aggregating form of C. auris v2.12.3 [37]. For the ALS4 insert signal analysis, PacBio reads containing the ALS4 repeat fragment were firstly extracted and re-mapped to the B11221 genome using minimap2 with parameter "-secondary = yes and -N 5" allowing for at least five secondary alignments.

Q-RT-PCR and RNAseq experiments
Q-RT-PCR and RNAseq experiments were performed as described previously with slight modifications [28]. Single colonies of C. auris strains were inoculated and grown in liquid YPD medium at 30˚C until they reached exponential growth. Cells were harvested for total RNA extraction. 1 μg of total RNA per sample was used to synthesize cDNA with RevertAid H Minus Reverse Transcriptase (Thermo Scientific, Inc., Beijing, China). Quantification of transcripts for Q-RT-PCR was performed using a Bio-Rad CFX96 real-time PCR detection system with SYBR green. The values were normalized to the C. auris ACT1 gene. Three biological replicates were used. The following primers used were for PCR amplification: C. auris-pACT1-RT-F:5-ATGTCAACATTCAGGCTGTC-3 C. auris-pACT1-RT-R:5-AGTAGTCAGTCAAGTCTCTTCC-3 C. auris B9J08_004112 (ALS4) RT-F:5-TAACTTGGAAGCCGCTGGTG-3 C. auris B9J08_004112 (ALS4) RT-R:5-TACCGTACTGACCATCATAT-3.

Protease treatment assays
The nonaggregative/yeast-form or aggregative-form cells of C. auris were incubated at 30˚C for 24 h. Cells were harvested and washed 3 times with 1 x PBS. Fungal cells were resuspended in 1 mL 1 x PBS and then treated with ddH 2 O, proteinase K (at a working concentration of 50 μg/mL) (TianGen Biotech, Beijing, China), or trypsin (at a working concentration of 0.25% (Gibico, Inc., Beijing) at 37˚C for 1 h.

Biofilm assays
Biofilm assays were performed as described previously with slight modifications [40,41]. Silicone squares (Cardiovascular Instruments Corp, PR72034-060N) and 24-well flat-bottom polystyrene plates (BD Falcon) were used for the biofilm assays. Overnight cultures of each strain in YPD were harvested, washed, and inoculated into fresh RPMI1640 liquid medium (OD 600 = 0.5) for biofilm growth. The cultures were incubated at 37˚C for 90 min at 200 rpm with agitation for initial adhesion. The polystyrene plate bottoms or silicone squares were gently washed with 1 x PBS and fresh RPMI1640 medium was added for an additional 24 h of growth at 37˚C. Biofilms formed in the 24-well polystyrene plates were used for CFU assays. The biofilms were washed with 1 x PBS, treated with proteinase K for 1 hour, resuspended, and then diluted and plated on YPD medium. Fungal cells were cultured at 37˚C for two days. Biofilms of each strains on silicone square were stained with calcofluor white and Fluor 405 nm for confocal scanning laser microscopy (CSLM) visualization. Confocal sections were acquired every 0.5 μm up to a total Z-stack thickness of 60 μm in sequential mode with lasers. CSLM top and side views were constructed.

PLOS PATHOGENS
A new aggregating form of C. auris

Scanning electron microscopy (SEM) assays
SEM assays were performed as described previously [40]. Fungal cells in exponential growth phase or on silicone squares were fixed with 2.5% glutaraldehyde and washed three times with 1 x PBS. The samples were then dehydrated with ethanol (with gradually increasing concentrations: 50%, 70%, 85%, 95%, and 100%), dried, and coated with gold.

Galleria mellonella infection model
Larvae of G. mellonella (0.3-0.4 g) were purchased from Tianjin Huiyu biological technology Co. LTD. (Tianjin, China). Cells of C. auris were cultured on YPD medium at 30˚C for 24 hours, collected, and washed twice with 1 x PBS. Approximately 1 x 10 6 fungal cells in 10 μL 1 x PBS were injected into each larva as previously described [18,42]. After injection, the larvae were placed in plastic culture dishes and placed at 30˚C or 37˚C in the dark.

Mouse skin infection virulence assays
All animal experiments were performed according to the guidelines approved by the Animal Care and Use Committee of the School of Life Sciences at Fudan University. Newborn BALB/c mice (aged 3-5 days) were used. Approximately 5 x 10 6 cells of each C. auris strain in 2 μL ddH 2 O were inoculated onto the dorsal back skin of newborn mice. Small sterile filter papers were used to cover the inoculated areas and medical tape was used to affix the filter papers onto the inoculated areas once the water evaporated. After 24 hours of infection, the infected skin tissues were excised and fixed with 2.5% glutaraldehyde for SEM assays.

Plasmid and ALS4 ectopic expression strain construction
To construct the plasmid pTDH3 for gene ectopic expression in C. auris, we first subcloned a fusion PCR product of the RPS1 locus (5' and 3') into the PstI/BamHI site of plasmid pBluescript II KS(+). Genomic DNA of C. auris strain BJCA001 was used as the template. A StuI restrict enzyme site was introduced into the amplified fragment. Primers LT1704 (atatacCTG CAGAGGTCTCTTTAGAAACTCCATC), LT1705 (CTGTCTCTTGGTGAAAGCAAAGGC CTTGGCGAAAACTCTCAAAAC), LT1706 (GTTTTGAGAGTTTTCGCCAAGGCCTTTG CTTTCACCAAGAGAC AG), and LT1707 (atatacGGATCCTCGTTCACAGAACTTGTTTG) were used for PCR. A fragment of the C. auris TDH3 promoter sequence was amplified from genomic DNA of C. auris strain BJCA001 with primers LT1710 (atatacCTGCAGAGGGAAA GAACTTGACATTCC) and LT1711 (atatacGATATCCATGTTGTATAAGGGAAAATAAA) and subsequently subcloned into the PstI/EcoRV site of the plasmid. Then, the caSAT1 cassette (Candida-adapted nourseothricin-resistant gene) was amplified from the plasmid pNIM1 [43] with primers cauSAT1F (atatacGCGGCCGC CGTCAAAACTAGAGAATAAT) and cau-SAT1R (atatacTCTAGAGACCACCTTTGATTGTAAATAG) and subcloned into the XbaI/ NotI site of the plasmid, generating plasmid pTDH3 (Fig 9A).

PLOS PATHOGENS
A new aggregating form of C. auris

Statistical analyses
Two-tailed paired Student's t tests and log rank tests were used for statistical analyses as stated in the figure legends.