Azole resistance is mediated by integration of sterol gene regulation and membrane transporter production by the zinc cluster-containing transcription factor Upc2A in Candida glabrata

The most commonly used antifungal drugs are the azole compounds that interfere with biosynthesis of the fungal-specific sterol: ergosterol. The pathogenic yeast Candida glabrata commonly acquires resistance to azole drugs like fluconazole via mutations in a gene encoding a transcription factor called PDR1. These PDR1 mutations lead to overproduction of drug transporter proteins like the ATP-binding cassette transporter Cdr1. In other Candida species, mutant forms of a transcription factor called Upc2 are associated with azole resistance, owing to the important role of this protein in control of expression of genes encoding enzymes involved in the ergosterol biosynthetic pathway. Recently, the C. glabrata Upc2A factor was demonstrated to be required for normal azole resistance, even in the presence of a hyperactive mutant form of PDR1. Using genome-scale approaches, we define the network of genes bound and regulated by Upc2A. By analogy to a previously described hyperactive UPC2 mutation found in Saccharomyces cerevisiae, we generated a similar form of Upc2A in C. glabrata called G898D Upc2A. Chromatin immunoprecipitation coupled with Next Generation Sequencing (ChIP-seq) demonstrated that wild-type Upc2A binding to target genes was strongly induced by fluconazole while G898D Upc2A bound similarly, irrespective of drug treatment. We also carried out RNA-seq analysis to determine the genes that were direct or indirect targets of Upc2A transcriptional control. In addition to the well-described ERG genes as Upc2A transcriptional targets, we found a large group of genes encoding components of the translational apparatus along with membrane proteins. These Upc2A-regulated membrane protein-encoding genes are often targets of the Pdr1 transcription factor, demonstrating the high degree of overlap between these two regulatory networks. Finally, we provide evidence that Upc2A impacts the Pdr1-Cdr1 system during the anaerobic response and also modulates resistance to caspofungin. These studies provide a new perspective of Upc2A as a master regulator of lipid and membrane protein biosynthesis. Author summary In the pathogenic yeast Candida glabrata, expression of the genes encoding enzymes in the ergosterol biosynthetic pathway is controlled by the transcription factor Upc2A. C. glabrata has a low intrinsic susceptibility to azole therapy and acquires fluconazole resistance at high frequency. These azole resistant mutants typically contain substitution mutations in a gene encoding the transcription factor Pdr1. Pdr1 does not appear to regulate ergosterol genes and instead induces expression of genes encoding drug transport proteins like CDR1. Here we establish that extensive overlap exists between the regulatory networks defined by Upc2A and Pdr1. Genomic approaches are used to describe the hundreds of genes regulated by Upc2A that far exceed the well-described impact of this factor on genes involved in ergosterol biosynthesis. The overlap between Upc2A and Pdr1 is primarily described by co-regulation of genes encoding membrane transporters like CDR1. We provide evidence that Upc2A impacts the transcriptional control of the FKS1 gene, producing a target of a second major class of antifungal drugs, the echinocandins. Our data are consistent with Upc2A playing a role as a master regulator coordinating the synthesis of membrane structural components, both at the level of lipids and proteins, to produce properly functional biological membranes.

139 function of this transcriptional regulator (25). The relevant mutation (upc2-1) (19) is a 140 change of a glycine to an aspartate residue located at position (G888D) in the carboxy-141 terminus of ScUpc2. Alignment of C. glabrata Upc2A and ScUpc2 indicated that G898 142 was the analogous position in Upc2A. This residue was replaced with an aspartate to 143 form the G898D UPC2A form of the gene. The resulting mutant allele was tagged with 144 a 3X hemagglutinin (3X HA) epitope at its amino terminus as we have previously done 145 for the wild-type UPC2A gene and both these forms of UPC2A were integrated into an 146 otherwise wild-type C. glabrata strain. These tagged strains were then grown to mid-log 147 phase along with isogenic wild-type and upc2A cells. Serial dilutions of each culture 148 were placed on rich medium containing the indicated concentrations of fluconazole 149 ( Figure 1A).

150
Introduction of the G898D mutation into UPC2A led to the resulting factor

171
These same strains were then used to compare expression of their protein 172 products by western analysis with appropriate antibodies. All strains were grown in the 173 absence or presence of fluconazole and whole cell protein extracts prepared. These 174 were analyzed by western blotting using the indicated antibodies ( Figure 1C and D).

175
The presence of the G899D UPC2A allele supported normal fluconazole 176 induction of Cdr1 and showed a modest reduction in Pdr1 activation. The levels of the 177 wild-type and G898D forms of Upc2A were not detectably different as shown by blotting 178 with the anti-Upc2A polyclonal antiserum. Over the time course of fluconazole 179 challenge (two hours), no differences in the levels of these two forms of Upc2A were 180 seen. These data argue that the increased activation seen for ERG11 in the presence 181 of the G898D UPC2A gene was due to increased function of Upc2A rather than a 182 change in its expression compared to the wild-type factor.

183
230 consensus elements as a more limited repertoire of regulated genes was considered.
231 We examine the binding of Upc2A to its DNA target sites in detail below. 241 target promoters were also associated with Upc2A binding ( Figure 2C). The top 4 GO 242 terms enriched in genes bound by both these Zn 2 Cys 6 -containing transcription factors 243 were associated with transmembrane transport or integral membrane components 244 (Supplementary table 2). As our earlier work had shown that both PDR1 and CDR1 245 were targets of Upc2A along with Pdr1 (18), these new data indicate that the overlap 246 between these two transcriptional circuits extends well beyond the initial two genes.

247
Two different classes of Upc2A target promoters are shown in Figure 2D. The 248 ERG11 gene is an example of a locus controlled by Upc2A but not Pdr1. Binding of 249 wild-type Upc2A is represented by the read depth and can be seen to increase in the 250 presence of fluconazole compared to in the absence of the drug. Binding of G898D 251 Upc2A was constant, irrespective of the presence of the drug. Note that when the lack 252 of a change in Upc2A expression is considered (see Figure 1C), these data support the 253 view that the DNA-binding activity of wild-type but not G898D Upc2A is increased by the 254 presence of fluconazole, possibly by an increase in nuclear localization (22).

255
CDR1 represents a Upc2A target gene that is also regulated by Pdr1. The 256 bound regions for Pdr1 and Upc2A extensively overlap in the upstream region of CDR1.
257 Pdr1 DNA-binding was strongly upregulated in  0 cells, likely due in part to the large 258 increase in PDR1 expression in this background compared to wild-type cells (27).
259 Upc2A DNA-binding to CDR1 was regulated in a manner similar to that seen for 260 ERG11.

261
The data above did not take into account a consideration of target gene

299
The final and largest class of genes found to be bound by Upc2A were less than 300 two-fold induced by fluconazole. GO term analysis of this class of genes indicated that 301 the top three enriched categories were involved in translation and the ribosome (60 302 total). The next highest categories were plasma membrane or amino acid 303 transmembrane transport and represented 74 genes. Together, these data strongly 304 suggest that Upc2A impacts a wide range of cellular process as well as its well-305 described control of expression of genes involved in the ergosterol biosynthetic 306 pathway.

307
To examine the range of expression of Upc2A-responsive genes, we compared 308 the transcriptional response of a range of genes from the ERG pathway with loci that we 309 have previously found to be targets of Pdr1 in C. glabrata (27). This comparison is 310 presented in the form of a heat map ( Figure 3B).

311
The majority of ERG genes showed at least two-fold induction by fluconazole in 312 wild-type cells as long as these genes corresponded to steps later in the ergosterol 313 biosynthetic pathway. ERG10, ERG13, HMG1 and ERG20, which all encode early 314 steps in ergosterol biosynthesis, were not influenced by fluconazole challenge in wild-315 type cells although expression of these genes was strongly depressed under these 316 same conditions in the absence of UPC2A. Genes encoding enzymes that function 317 later in ergosterol biosynthesis (like ERG11) were induced by fluconazole at least two-318 fold in wild-type cells but depressed by at least two-fold in a upc2A background.

319
There were two ERG genes that were exceptions to these general trends of

338
We also used RNA-seq to compare the gene expression profile of wild-type cells 339 to an isogenic G898D UPC2A strain. These strains were grown to mid-log phase in the 340 absence of fluconazole and then standard RNA-seq was carried out to determine the 341 effects of this form of Upc2A on the transcriptome (Table 2).

342
The presence of the GOF form of UPC2A caused relatively small changes in 343 gene transcription. There were only 11 genes observed to be elevated at least 1.4-fold.
344 A striking feature shared by these genes was that nine of eleven encoded products that 345 were involved in the biosynthesis of ergosterol. Five of these 9 genes also contained 346 SREs. As we have seen for ERG11 ( Figure 1B

391
Introduction of the mSRE into ERG1, CDR1 or PDR1 promoters led to a 392 reduction in the level of fluconazole-induced -galactosidase activity produced by each 393 respective fusion gene ( Figure 5A). While some degree of fluconazole inducibility was 394 retained in each mSRE-containing promoter, these data indicate that each SRE 395 identified above is required for normal drug induced promoter activation.

396
To examine the effect of the loss of the SRE from the wild-type CDR1 and PDR1 397 genes, the mSRE mutations were introduced into otherwise wild-type versions of these

437
Anaerobic growth led to 20-fold or higher induction of the genes corresponding to 438 products important in sterol uptake such as the ABC transporter Aus1 and other 439 proteins thought to be essential for this process (Dan1, Tir1) ( Figure 6A). Anaerobic 440 growth also repressed expression of genes encoding mitochondrial proteins involved in 441 ATPase production (Atp3, Atp4) and an electron transport chain component (Sdh2).
442 Two different loci encoding enzyme involved in the tricarboxylic acid cycle were slightly 443 reduced (KGD2) or unaffected (ACO1).

444
Having confirmed that the expected anaerobic gene regulation was seen under 445 our growth conditions, we next tested expression of Cdr1, Pdr1 and Erg11 using the 446 western blot assay described previously ( Figure 6B). Both Cdr1 and Pdr1 were strongly 447 induced, from 3-to 5-fold, while Erg11 was also induced albeit roughly 2-fold.

448
Since all these Upc2A target genes were induced in this assay, we wanted to 449 determine if the SREs associated with CDR1 and PDR1 were required for this 450 anaerobic activation. We used the strains described early in which either wild-type 451 versions of CDR1 and PDR1 were present or these same genes containing mutations in 452 their respective SREs were used. These isogenic strains were grown in the presence 453 or absence of oxygen and levels of Cdr1, Pdr1 and Erg11 (as a control for anaerobic 454 conditions) measured by western blotting.

455
Anaerobic induction of both Cdr1 and Pdr1 was diminished in the absence of the 456 SRE motifs in their promoters ( Figure 6C). As seen before for fluconazole challenge, 457 the SREs in the CDR1 and PDR1 promoters are required for normal induction. Erg11 458 was not affected as its SRE was unaltered in this experiment.

459
To test if the expression of CDR1 and PDR1 was involved in normal anaerobic 460 growth, we carried out a competitive growth assay. Isogenic wild-type and double 478 tested for resistance to caspofungin, caffeine or high pH using a serial dilution assay.
479 Caffeine and high pH are cell wall stresses and reflect general cell wall dysfunction (35).
Loss of UPC2A caused hypersensitivity to all these agents ( Figure 7A) while both 481 epitope-tagged alleles behaved like the wild-type strain. The finding of a caspofungin 482 susceptible phenotype prompted us to examine expression of the three FKS genes in C.
483 glabrata to determine if any of these showed a response to the G898D allele of UPC2A.
484 None of these genes were altered in the presence of this gain-of-function form of 485 UPC2A while both ERG1 and AUS1 were elevated ( Figure 7B), confirming the 486 functionality of this hypermorphic form of Upc2A.

487
To explain the observed caspofungin hypersensitivity of the upc2A strain, levels

498
To determine if the SRE was required for normal expression of FKS1, a lacZ 499 translational fusion gene was prepared in which the FKS1 regulatory region determined 500 expression of -galactosidase. Both the wild-type and mSRE-containing FKS1 501 promoters were used and introduced on a low-copy-number plasmid into wild-type C.
502 glabrata cells. FKS1-dependent -galactosidase activities were then determined in the 503 presence or absence of caspofungin induction.

504
Loss of the SRE from the FKS1 promoter caused a significant reduction in FKS1-505 dependent expression of lacZ in the absence of caspofungin ( Figure 7E). These data 506 provide evidence that Upc2A-mediated gene activation is required for normal 507 expression of FKS1 and wild-type caspofungin resistance.

523
The large number of Upc2A target genes illustrates the breadth of processes that 524 are transcriptionally influenced by this factor. Clearly, the ERG genes are an important 525 set of genetic targets but these are a small fraction of the whole. Upc2A appears to be 526 coordinating a broad group of genes including a large number of plasma membrane-527 localized proteins (See supplementary table 1 and 3). Coupled with its control of 528 ergosterol in this membrane, Upc2A appears to be a central determinant of the 529 composition of this membrane compartment in C. glabrata. The regulation of plasma 530 membrane constituents is of obvious importance in modulating the ability of substances 531 to cross this barrier between the external and internal environments.

532
Comparing the members of the target gene sets defined by Upc2A and Pdr1 533 suggests a hierarchical relationship between these two transcription factors. Here we 534 establish that a binding site for Upc2A lies upstream of PDR1 and is required for normal 535 activation of PDR1 expression ( Figure 6C) as well as many of the other genes 536 controlled directly by Pdr1. We suggest Upc2A provides overarching control of both the 537 Pdr1 regulon but also a variety of other genes that are not under Pdr1 control, serving to 538 link these different classes of genes through this common transcriptional regulation.

539
While the full range of Upc2A target genes illustrate the global importance of this 540 transcription factor, the ERG genes are especially sensitive to the level of activity of this 541 regulator. The G898D UPC2A allele has a surprisingly limited effect on gene 542 expression as this allele was seen to trigger significant transcriptional changes almost 543 exclusively in genes associated with ergosterol biosynthesis ( 556 The conserved location of these hypermorphic alleles suggests the possibility that a 557 common function is being disrupted in both organisms.

567
The finding of the interaction between Upc2A and the FKS1 gene provides an 568 interesting connection between azole resistance, well-known to be impacted by Upc2A 569 (41), and echinocandin resistance. These two antifungal drugs have been considered 570 to be defined by genetically separable pathways but here we provide evidence that 571 Upc2A may provide a link between them. Intervention in Upc2A-mediated 572 transcriptional activation may be able to cause reductions in resistance to both azole 573 drugs and the echinocandins.

574
Finally, our data also illuminate the complexity and interrelationship of expression