Integration of Fungus-Specific CandA-C1 into a Trimeric CandA Complex Allowed Splitting of the Gene for the Conserved Receptor Exchange Factor of CullinA E3 Ubiquitin Ligases in Aspergilli

Aspergillus species are important for biotechnological applications, like the production of citric acid or antibacterial agents. Aspergilli can cause food contamination or invasive aspergillosis to immunocompromised humans or animals. Specific treatment is difficult due to limited drug targets and emerging resistances. The CandA complex regulates, as a receptor exchange factor, the activity and substrate variability of the ubiquitin labeling machinery for 26S proteasome-mediated protein degradation. Only Aspergillus species encode at least two proteins that form a CandA complex. This study shows that Aspergillus species had to integrate a third component into the CandA receptor exchange factor complex that is unique to aspergilli and required for vegetative growth, sexual reproduction, and activation of the ubiquitin labeling machinery. These features have interesting implications for the evolution of protein complexes and could make CandA-C1 an interesting candidate for target-specific drug design to control fungal growth without affecting the human ubiquitin-proteasome system.

CandA-N and CandA-C (22). Both genes are separated in most aspergilli by five genes coding for putative proteins, including septation-associated SepK, vacuolar biosynthesis-related Pep5/Vps11, or chitin deacetylase ( Fig. 1A; see also Fig. S1A in the supplemental material). Cand1/A proteins have an armadillo-type fold typical of HEAT repeat proteins (21,23,24). CandA-C and CanA carry N-terminal nuclear localization signal (NLS) sequences (RKRRR) (22). A. fumigatus CanA carries an NTE encoded in exon-1 that corresponds to the deduced protein encoded by A. nidulans AN12234, which we named CandA-C1. This gene is located only 269 bp upstream of candA-C and was not considered before to encode a CandA subunit. CandA-C1, as well as A. fumigatus CanA NTE, have an N-terminal RNase P Rpr2/Rpp21 domain motif and, presumably, disordered C termini (Fig. 1A). This indicates that two A. fumigatus CanA CandA-C1 polypeptide represents the N-terminal part of the CanA protein of A. fumigatus but is encoded by a separate candA-C1 gene (546 bp, 181 amino acids [aa], 19 kDa) in A. nidulans, which is located next and upstream to candA-C (3,254 bp, 1,041 aa, 113.5 kDa), which is separated from candA-N (1,055 bp, 313 aa, 33.6 kDa) by five open reading frames. CandA-C1 has an RNase P Rpr2/Rpp21 motif (pink), and V. dahliae Cand1 and Aspergillus CanA/CandA-C have an NLS (red) and a TATA binding protein (TBP) interaction motif (blue), which is conserved in all eukaryotic CandA C-terminal ends. H. sapiens Cand1, UniProt ID Q86VP6-1; V. dahliae Cand1, MycoCosm ID VDAG_05065T0; A. fumigatus CanA, UniProt ID Q4WMC6, and CanA-N, UniProt ID Q4WMC0; A. nidulans CandA-N, UniProt ID C8VP82; and CandA-C1, AspGD/FungiDB ID AN12234, CandA-C, UniProt ID Q5BAH2). (B) Comparative analysis of CandA-C1 and CandA-C orthologs in different aspergilli. The protein sequence of A. nidulans CandA-C1 (AN12234) was compared to those of different Aspergillus spp. by a BLASTp search in the Joint Genome Institute (JGI) MycoCosm genome portal. Genomic clusters of orthologs of separated candA-C1 (purple) followed by intergenic ORF (iORF; black line) and downstream-located candA-C (yellow) genes are depicted. Three corresponding fused genes with the CandA-C1-like domain marked in purple-yellow stripes followed by an intron (gray) instead of the iORF are depicted in the bottom. Phylogenetic relationships are based on protein identities of the CandA-C1 orthologs in the genus Aspergillus, similar to relations described by de Vries et al. (2). id, protein identity.
CandA Supports Growth and Development in Aspergilli ® proteins correspond to three A. nidulans CandA proteins, with 59% identity between CandA-C1 and CanA-NTE and 79% identity of CandA-C with full-length CanA. The two N-terminal orthologs share 77% protein identity.
Bioinformatic analysis revealed that a majority of the 12 Aspergillus species carry the same two separated adjacent CandA-C1-and CandA-C-encoding genes as A. nidulans. Only A. fumigatus, Aspergillus clavatus, and Aspergillus aculeatus are annotated as expressing deduced fusion proteins with similar lengths and intron distributions (2) (Fig. 1B). These results suggest that Aspergillus-specific CandA-C1 could be linked to the conserved CandA protein family, mostly as an independent protein but also in some species as an N-terminal domain fusion to CandA-C. The surrounding positioning of homologous genes is conserved, indicating a common ancestor of all Aspergillus spp. which presumably had encountered a specific rearrangement of candA genes during evolution, which has not yet been described in other fungi.
A. fumigatus expresses a single canA gene, and A. nidulans expresses two separate transcripts, candA-C1 and candA-C. The gene expression levels of candA-C1, candA-C, and candA-N in A. nidulans were compared by quantitative real-time PCRs (qRT-PCRs) from the wild type and a strain overexpressing candA-C1::gfp by the nitrate promoter. Fifty-fold overexpression of candA-C1 neither changed the fungal phenotype caused by overexpression of candA-C1::gfp nor altered the expression levels of candA-C and candA-N ( Fig. 2A). This indicates that candA-C1 and candA-C are separate genes, and the 269-bp intergenic-open reading frame (iORF) region between candA-C1 and candA-C ORFs should include a terminator for candA-C1 and a promoter for the candA-C gene. RNA sequencing (RNA-seq) data deposited in FungiDB showed expressed sequences of both ORFs, with fewer candA-C1 fragments per kilobase of exon model per million mapped reads than candA-C, supporting differences in expression between the two genes.
The ORFs of candA-C1, candA-C, and a putative fusion of candA-C1::candA-C were amplified from wild-type genomic DNA (gDNA) and complementary DNA (cDNA), as well as of mutant strain cDNA, to examine the transcripts of the iORF region (Fig. 2B). Specific PCR products for candA-C1 were obtained from wild-type c/gDNA, ΔcandA-C mutant, and ΔiORF mutant strains (Fig. 2C). Except for a faint signal of candA-C amplification in the ΔiORF mutant strain, a significant PCR product of candA-C was only observed from PCRs on wild-type c/gDNA and when candA-C1 was deleted. This corresponds to the significant downregulation of candA-C expression when the iORF was missing in qRT-PCR, supporting the idea that the iORF includes the promoter sequence for candA-C (Fig. 2D). The candA-C1 gene expression was 3-fold increased in a candA-C deletion strain. These observations are consistent with the phenotypes of a strain lacking the candA-C1 start codon, the ΔiORF mutant, and the ΔcandA-C1/iORF mutant strain ( Fig. 2C and S1B and C). A combined candA-C1::candA-C transcript was amplified and might be an antisense transcript, which is supported by FungiDB RNA-seq data. The A. fumigatus CanA-green fluorescent protein (CanA-GFP) (163 kDa) and A. nidulans CandA-C-GFP (142 kDa) exhibited different molecular weights in Western hybridization, and the expression of a fused A. nidulans candA gene resulted in a 170-kDa protein, which correlates with the sum of the single subunits (Fig. 2E). These results demonstrate that candA-C1 and candA-C are separate genes in A. nidulans, with an iORF containing candA-C1 terminator and candA-C promoter sequences, whereas the orthologous A. fumigatus CanA combines both peptides, including an NTE corresponding to CandA-C1.
CandA-C1 interacts with CandA-C and CandA-N. Localization of GFP-tagged CandA proteins from A. nidulans by fluorescence microscopy revealed that GFP-CandA-N and CandA-C-GFP are mainly localized to the cytosol and nuclei. CandA-C1-GFP is localized to the nuclei, nucleoli, cytosol, and, presumably, mitochondria (Fig. 3A). Bimolecular fluorescence complementation (BiFC) microscopy showed that CandA-C1 interacts with CandA-N and CandA-C in the nuclei and sometimes in mitochondria in wild-type and cullin deneddylation-deficient csnE deletion strains ( Fig. 3B and S2A and B).
The CandA-C backpacks CandA-N into the nucleus (23), and its NLS (RKRRR) at amino acid positions 138 to 142 is required for CandA-C and CandA-N nuclear transport. Nuclear localization of CandA-C1 is independent of the CandA-C NLS and also of the presence of CandA-C or CandA-N. Conversely, CandA-N and CandA-C are also nuclear in the absence of CandA-C1 ( Fig. 3C and S2C). Therefore, CandA-C1 and CandA-C travel independently into the nucleus, and only the nuclear transport of CandA-N is dependent on CandA-C.    (Table S1). Peptides of CandA-N and CandA-C showed highest log 2 (x) label-free quantification (LFQ) intensities in CandA-C-GFP and GFP-CandA-N pulldowns. CulA was identified as among the best candidates, whereas CulC or CulD could not be identified, indicating that A. nidulans CandA is specific for CulA-containing CRLs. A. fumigatus CanA-GFP pulled CanA-N, both CulA and CulC, and RbxA ( Fig. 4A and B). Western hybridization of the CandA-C1-GFP elution fraction using anti-Nedd8 antibody revealed signals of interacting neddylated cullins, as well as free Nedd8, whereas cullins were not identified in LC-MS data analysis of CandA-C1-GFP (Fig. 4C). Western hybridization and MS LFQ intensities corroborate the idea that CandA-C1 is part of a trimeric CandA complex and might exhibit additional cellular functions in A. nidulans independently of CandA-N or CandA-C.
A. nidulans CandA is required for SCF activation. The CulA neddylation status in A. nidulans candA deletion strains was investigated with in vitro deneddylation assays to analyze whether all three subunits have an impact on cullin neddylation. Deneddylated CulA is visualized as lower and neddylated CulA as higher migrating signals in Western hybridization experiments (17,25). Larger amounts of deneddylated CulA were visible in the ΔcandA-C1, ΔcandA-C, ΔcandA-N, and ΔcandA-N/A-C mutant strains than in the wild type, with the most pronounced 2-fold-higher effect in the absence of CandA-N. Fewer ubiquitinated proteins were observed in all candA deletion strains and are presumably a direct consequence of increased inactive CulA. Double-and tripledeletion strains of candA with deneddylase-deficient csnE showed increased neddylated CulA levels, like those of a csnE single-deletion strain. The total neddylated cullins were similar in the ΔcandA mutant strains ( Fig. 5A and B). Therefore, all three CandA subunits contribute to the accurate ratio of neddylated relative to deneddylated CulA within the fungal cell.
The effect of CandA-C1 on SCF activity was investigated using the candA-C1::gfp overexpression strain in wild-type and ΔcandA-N and ΔcandA-C mutant backgrounds. The amount of deneddylated CulA was reduced to wild-type levels when candA-C1::gfp was overexpressed in the ΔcandA-C or ΔcandA-N/A-C mutant, but nothing changed in the ΔcandA-N mutant strain. The overexpression of candA-C1 alone did not have any effect on deneddylated CulA. Elevated candA-C1::gfp expression increased the total neddylated cullin levels of candA-N/A-C single-and double-deletion strains (Fig. 5B). CandA-C1 overexpression can therefore only rescue defects in the CulA neddylation cycle caused by the absence of CandA-C, but not of CandA-N, and it increases the total neddylated cullin pool. These data further suggest that after the initial deneddylation of CRLs by CSN and subsequent CandA-mediated CRL disassembly, CandA has an additional novel function. It is also required to initiate a new cycle of neddylation and activation of CRLs with another substrate receptor. Therefore, CRL activity is not only dependent on a functional CSN but also on the interplay of CandA-N, CandA-C1, and CandA-C. experiments. An excerpt of the heatmap, generated with Perseus (version 1.6.0.7), depicts label freequantification (LFQ) intensities of three biological replicates of GFP control, CandA-N, CandA-C, and four replicates of CandA-C1. Log 2 (x) LFQ intensities range from 0 to 15, not considered to be identified (black); 15 to 18, low intensity (blue); and 18 to 26/32, low to high LFQ intensities (gradient from red to orange to yellow). Colored bars indicate cellular localization, based on KEGG and UniProt databases (green, nucleus; light yellow, cytosol). The molecular pathway of putative interaction partners is labeled as follows: berry, ubiquitin-proteasome system (UPS); dark blue, nuclear transport; purple, stress response; and dark gray, transport and signaling. (B) Heatmap of identified proteins from A. fumigatus CanA-GFP pulldown compared to overexpression GFP control strain. (C) Western hybridization of pulldown elution samples probed with anti-GFP antibody, which shows free GFP (1), GFP-CandA-N (2), CandA-C-GFP (3), CandA-C1-GFP (4), and CanA-GFP (7). The elution sample of CandA-C1-GFP was reprobed with anti-Nedd8 antibody, which highlights neddylated cullins (5) and free Nedd8 (6). Single-and double-deletion A. nidulans candA-N and candA-C mutant strains showed colony diameters similar to those of the wild type but produced fewer conidia. The hyphae and surrounding media were colored dark red-brown, indicating an altered secondary metabolism (Fig. 6A to D and S3A). Analysis of secondary metabolites from asexual development by LC-MS revealed that both strains produce cichorine (VII) that was hardly detectable in the wild type or ΔcandA-C1 mutant. Peak VIII, present in the ΔcandA-C and ΔcandA-N mutants, corresponds to an unknown metabolite with mass of m/z 210.0761 [MϩH] ϩ and deduced molecular formula C 10 H 11 NO 4 . Austinol (I), dehydroaustinol (II), asperthecin (III), emericellin (IV), and shamixanthone/epishamixanthone (V and VI) were increased in the ΔcandA-C1 mutant strain in comparison to the wild type, which does not produce detectable asperthecin ( Fig. 6D and Data Set S1).

CandA Supports Growth and Development in Aspergilli
The ΔcandA-C1 mutant grew 6 times slower but produced 1.5 to 2 times more conidia than the candA-N/A-C single-and double-deletion strains, which was still 30 times less than the amount of spores produced by the wild type ( Fig. 6A to C). Conidial formation was increased, but the colony radius decreased when candA-C1 was overexpressed in the ΔcandA-C or ΔcandA-N/A-C mutant but not in the ΔcandA-N mutant strain, suggesting that candA-C1 expression can rescue conidiation defects caused by the loss of candA-C (Fig. S3B to D). CandA-C1 therefore has a distinct additional cellular function and promotes vegetative growth and conidial formation. In the wild type, most CulA is deneddylated (ϳ96 kDa, }}), which is different from a ΔcsnE strain, which is defective in cullin deneddylation and has most CulA bound to Nedd8 (ϳ106 kDa, }). In the ΔcandA and ΔcandA-C1 deletion strains, deneddylated CulA accumulates. Double and triple deletions of csnE, candA-N, and/or candA-C show an accumulation of neddylated CulA as observed for the ΔcsnE mutant. Signals were quantified with pixel density measurements using BIO1D software (Peqlab) for total 12 replicates (three biological replicates each with four technical replicates). Tubulin (Tub) served as a loading control. (B) Western hybridization probed with anti-CulA (gray), anti-Nedd8 (red), and anti-ubiquitin (␣-Ub) (green) antibodies. Ponceau served as loading control. The ΔcandA mutant strains were compared to candA-N and candA-C deletion strains overexpressing candA-C1. The pixel density ratio was determined with the BIO1D software (Peqlab), quantified against Ponceau, and normalized to wild-type signals. CulA and Ub used three biological with three technical replicates each, and Nedd8 used four biological with three technical replicates each; error bars represent the standard error of the mean. ‫,ء‬ P Յ 0.05; ‫,ءء‬ P Յ 0.01; ‫,ءءء‬ P Յ 0.001; ‫,ءءءء‬ P Յ 0.0001.  The A. fumigatus canA and canA Δexon1 deletion mutant strains were delayed in germination and showed a significant growth retardation and adjourned development, suggesting that CanA promotes spore germination and colony growth. Single deletions of ΔcanA-N or of the canA domain (bp 838 to 4078) corresponding to A. nidulans candA-C, as well as the double-deletion strain, showed a delay in conidial formation that was visible as a white halo surrounding the colony. CandA/CanA supports growth at different temperatures, and the full A. fumigatus CanA complex is required for spore germination at 30°C (Fig. 6E to G and S3F).
The conservation of the function of candA-C1 from both species was examined. The construction of an A. fumigatus strain with integrated An_candA-C1 sequence into the genomic locus of Af_canA exon1 and an A. nidulans strain with the Af_canA exon1 sequence introduced into the genomic locus of candA-C1 revealed that the two sequences are interchangeable (Fig. 7A). Generation of a CanA-like fusion protein in A. nidulans displayed a CandA-C1-CandA-C-GFP protein migrating at higher molecular weight than CandA-C-GFP (Fig. 7B). A comparison of A. nidulans with A. fumigatus candA/canA mutants revealed that CandA-C1 supports vegetative growth. The CandA orthologs have conserved functions in conidiation, although A. nidulans has a trimeric complex and A. fumigatus a dimeric complex.
CandA is required for sexual development, and CandA-C1 coordinates secondary metabolite genes other than those encoding CandA-C and CandA-N. The sexual life cycle of A. nidulans serves for overwintering of ascospores (26,27). Hülle cells protect and nurse the early nests that develop to primordia and maturate in 7 days to fruiting bodies (cleistothecia) (28). Each cleistothecium contains several asci, with each harboring eight ascospores. Strains missing candA-N and candA-C are blocked in sexual development at the stadium of early nest production (22). The colony of the ΔcandA-C1 mutant strain was covered with yellowish Hülle cells forming nests after 7 days of development and produced a volcano-like phenotype by growing vertically with a hole in the middle of the colony. Nests of the ΔcandA-C1 mutant contained only primordia after 7 days, and sexual fruiting bodies with a moderately soft and fragile surface were present after 14 days, indicating a delayed sexual development. These cleistothecia did not contain any ascospores but did contain a complex network of ascogenous hyphae (Fig. 8A to F). Large amounts of CandA-C1-GFP were not sufficient to rescue cleistoth- ecial formation in the candA-C and candA-N deletion strains (Fig. S3E). These data show that CandA-C and CandA-N are required for nest formation and primordial development, whereas CandA-C1 has a later function. CandA-C1 is required for stable cleistothecial wall formation and ascospore development.
Analysis of secondary metabolite production after 7 days of sexual development showed that ΔcandA-C and ΔcandA-N mutants produce cichorine and metabolite VIII, similar to what is observed in asexual development. Asperthecin, the red dye of cleistothecia, as well as emericellin and shamixanthone/epishamixanthone, were de- CandA Supports Growth and Development in Aspergilli ® tected by LC-MS in the ΔcandA-C1 mutant strain (Fig. 8G and Data Set S1), indicating that CandA-C1 coordinates secondary metabolite genes other than those for CandA-C and CandA-N.
In summary, we demonstrate that a likely genomic rearrangement of a single fungal candA gene during evolution required an additional component which had to be integrated and resulted in changes in the subunit compositions of CandA in aspergilli. A. fumigatus and A. nidulans represent two different groups with different solutions, dimeric, which includes an N-terminal extension, versus a trimeric CandA complex including similar genetic information. The NTE found in A. fumigatus CanA corresponds to the separated single CandA-C1 subunit of A. nidulans. This trimeric CandA complex is required for CRL activity, supports asexual and sexual development, and thereby has influence on the secondary metabolism.

DISCUSSION
The ubiquitin-proteasome pathway includes the dynamic interplay of the substrate receptor exchange factor CandA and three macromolecular multiprotein complexes, SCF E3 ubiquitin RING ligase, CSN deneddylase, and the 26S proteasome. The three ZOMES complexes CSN, proteasomal lid, and translation eukaryotic initiation factor 3 (eIF3) presumably have a common origin because they share similar subunits in a common architecture, with some variations in subunit compositions (29,30,64). Similarly, the subunit composition of the CSN antagonist Cand1/A is divergent in eukaryotes. The putative ancestor of all Aspergillus spp. might have had one candA gene encoding a single subunit CandA with N-and C-terminal domains corresponding to human Cand1. A. fumigatus is a representative of a group of species with a dimeric complex, which includes the N-terminal (CanA-N) and an extended C-terminal (CanA) part of human Cand1. A. nidulans represents a larger group of aspergilli which even form a trimeric CandA substrate receptor exchange factor complex, where the N-terminal extension of A. fumigatus CanA corresponds to the third subunit CandA-C1 in addition to the split CandA-N and CandA-C subunits, which we described earlier (22) (Fig. 9A). This represents an interesting example of evolutionary protein complex formation based on the splitting of one gene into two and combining the protein products with a polypeptide of an additional open reading frame providing additional functions, which is expressed as additional exon or as separate gene. The candA-C1 gene was presumably a separate gene which encodes a putative RNase P subunit with an Rpr2/Rpp21 motif in the N-terminus that is also found in A. fumigatus CanA. RNase P is a RNA-protein complex (ribozyme) which can cleave RNAs, such as, 5= precursors of tRNAs, and exists in different compositions of proteins and RNA, whereby A. nidulans encodes one nuclear and one mitochondrial RNA and seven associated proteins, including CandA-C1 (Table S2) (29,30). It is currently elusive whether CandA-C1 is still part of a RNA complex. It is specific for Eurotiomycetes in the division of Ascomycota, whereas higher eukaryotes lack an ortholog. The current gene order of the candA genes could be caused by a DNA double-strand break and subsequent rearrangement. The position of the candA-C gene changed to a position downstream of candA-C1. The A. nidulans iORF includes a terminator of candA-C1 and a candA-C promoter. It is elusive whether CandA-C1 was already functionally linked to CandA before the rearrangement or as result of reordering of genes. The consequences of the rearrangement of the A. nidulans candA genes are separate expression of candA-C1, candA-C, and candA-N. In A. fumigatus, the rearranged canA C-terminal sequence hijacked the upstream candA-C1 gene, which resulted in a fused gene encoding a CanA protein with an N-terminal extension and a separate canA-N gene. The fusion of these genes might have facilitated the response to stress, like temperature, oxidative, or heavy-metal stress, and sterolbiosynthesis-inhibiting triazole fungicides (31)(32)(33). The different organization of canA genes in A. fumigatus could be due to selective pressure that maintains the diversity of pathogens to avoid detection by the host immune system or might correlate with the heterothallism and thereby limited recombination by the sexual life cycle (31,34).
BiFC microscopy and pulldown experiments showed that CandA-N/A-C single sub-units interact with CandA-C1. The pulldown data also suggest that A. nidulans has a dimeric complex of CandA-N/A-C. Overexpression candA-C1 in the absence of candA-C balanced the neddylation ratio of CulA and complemented the conidiation defect. These results underline the idea that CandA-C1 interacts with CandA-N. The two proteins together can partially take over functions of the CandA-N/A-C or CandA-N/A-C1/A-C complexes. The stoichiometry of the complexes needs further investigation. The deneddylation assay supports the idea that all three A. nidulans CandA proteins are required for CRL disassembly, are obligatory for CRLs that lack a substrate, and are therefore deneddylated by the CSN. Disassembled and deneddylated cullins can bind other adaptor-receptor complexes for new substrate ubiquitination cycles, allowing the ubiquitination of diverse substrates involved in different cellular pathways, like the A. nidulans carbon catabolite repression, as recently shown (35). The CandA proteins are essential for optimal CRL activity, as was also shown by a mathematical modeling investigation by Liu and coworkers (15,16). This work demonstrates that CandA-C1 nuclear import is independent of CandA-C. Pulldowns revealed that CandA-C1 interacts with the importin-␣/␤1 homologs KapA (AN2142) and KapB (AN0906), which have nuclear and perinuclear localization, respectively (36,37). A. nidulans KapA transports the master regulator of secondary metabolism VeA in complex with VelB into the nucleus (38)(39)(40). CandA-C1's nucleolar localization might be due to interaction with the nonessential KapJ that was reported to have a nucleocytoplasmic and nucleolar localization (37). CandA localized to mitochon- CandA Supports Growth and Development in Aspergilli ® dria, indicating that it regulates the activity of CRLs that are connected to the outer mitochondrial membrane (41,42). The deletion of candA genes caused fragmented mitochondria (Fig. S3G), similar to what was reported for a strain lacking the proteasome lid subunit SemA (43). The observed mitochondrial dysfunctions might depend on CandA being required for development and secondary metabolism, and on CandA-C1 as essential vegetative growth factor, which might provide additional RNaseassociated functions. A connection between asexual development and secondary metabolism was also shown recently for the transcription factor SclB that activates the central regulatory pathway for conidiation (44). All three CandA subunits are obligatory for the multicellular development of sexual ascospores, supporting the idea that regulation of the CRL cycle is required for complex organisms. This corroborates with the embryonic lethality of csn or cand malfunction in higher eukaryotes (45)(46)(47)(48). A. fumigatus CanA NTE is obligatory for growth, and the CanA complex is required to cope with low-temperature stress. It is known that the major stress resistance factor for spores in A. fumigatus is trehalose (49). Whether CanA mediates trehalose stability through the UPS to improve stress resistance needs further investigation. The spread of fungal pathogens is difficult to control (50). The discovered impact of CanA on growth and the fact that the CandA-C1 protein domain is not conserved in higher eukaryotes could be beneficial for drug design against Aspergillus-derived diseases, like aspergillosis.
CandA-C1 coordinates secondary metabolite genes other than CandA-N and CandA-C. The ΔcandA-C and ΔcandA-N mutant strains produced cichorine and an unknown metabolite. From the mass of m/z 210.0761 [MϩH] ϩ , the molecular formula C 10 H 11 NO 4 was deduced. A literature search indicated that this substance is most probably emerimidine, which is related to cichorine but was not identified in A. nidulans so far. Emerimidine produced by Emericella variecolor CLB38 was shown to have antimicrobial activity against multidrug-resistant microorganisms like Bacillus subtilis and Staphylococcus aureus but also antifungal activity against Candida albicans and A. fumigatus (51). The thin-layer chromatogram of the extracts from asexually and sexually developed ΔcandA-C and ΔcandA-N mutants showed a blue spot at 366 nm at an R f of 0.43 (Fig. S3H), which is in accordance with the literature (51). The similarity between the UV-Vis spectra of cichorine and emerimidine, which fit very well with the literature, indicates that the two have the same core structure (51,52). Cichorine is synthesized from a nonreducing polyketide synthase, CicF (AN6448) (52). This metabolite is a known phytotoxin, and compounds with a similar framework have been connected to antitumor and antimicrobial activities (52)(53)(54). Cichorine was also identified in a ΔlaeB mutant strain that is impaired in sterigmatocystin synthesis (55,56). Previous studies found orsellinic acid and derivatives in candA, csn, veA, and velB deletion strains and connected these compounds to the dark pigment secreted by those strains (22,57,58). These compounds were not identified in the present study under the growth and metabolite extraction conditions used. Austinol and dehydroaustinol from the aus gene cluster were nearly absent in candA-C and candA-N deletion strains, correlating with the reduced amount of conidia (59). Conidia were observed when candA-C1 was missing, and austinol, dehydroaustinol, and emericellin, as well as xanthones from the mdp cluster, were isolated, which are all connected to asexual development and are present in the wild type (60). The metabolite asperthecin is known as the red pigment of cleistothecia (ascospores) and was found after 7 days in a sexually developed ΔcandA-C1 mutant strain, although no mature cleistothecia or ascospores were present, indicating that CandA-C1 supports the asperthecin synthesis but is impaired in the production of cleistothecia with mature ascospores (61).
This work demonstrates an evolutionarily changed CandA complex that comprises the newly identified 20-kDa CandA-C1 subunit in Aspergillus nidulans, which is part of the CanA C-terminal subunit in A. fumigatus. This additional subunit is required in both aspergilli for vegetative growth and asexual conidia formation ( Fig. 9B and C). A. nidulans CandA proteins are required to fullfill the sexual life cycle and all together are a prerequisite for priming CRL assembly, conferring dynamic protein turnover. There-fore, a trimeric CandA complex is as possible as a dimer of CandA-N/A-C or CandA-N/ A-C1, whereby CandA-C1 has a dual function in CRL regulation and putatively as an RNase P subunit. With all this information, Aspergillus CandA-C1 is an promising target to control fungal spread.

MATERIALS AND METHODS
Strains and media. The oligonucleotides and plasmids used for strain design are described in the supplemental material (Text S1). Strains were cultivated in liquid or solid minimal medium (MM) (44). Modified minimal medium (32) was used for A. fumigatus spotting or Western hybridization experiments. Four thousand conidiospores were used for spot tests. A. nidulans plates were incubated at 37°C (if not indicated otherwise) in the light for asexual development, or plates were sealed with Parafilm and incubated in darkness for sexual development. Vegetative mycelium was obtained from liquid MM or modified MM cultures inoculated with 1 ϫ 10 6 to 2 ϫ 10 6 spores/ml at 37°C for 20 h with agitation. Conidia were quantified as described previously (44).
Isolation of fungal genomic DNA and RNA and cDNA synthesis. gDNA was extracted with DNA lysis buffer (200 mM Tris-HCl [pH 8.5], 250 mM NaCl, 25 mM EDTA, 0.5% [wt/vol] SDS) and 8 M potassium acetate solution from ground mycelia obtained from liquid overnight cultures. gDNA was precipitated with isopropanol and the pellets resolved in distilled water (dH 2 O). RNA was isolated according to the RNeasy plant minikit (Qiagen) protocol. cDNA was transcribed from 0.8 g RNA using the QuantiTect reverse transcription kit (Qiagen).
Gene expression measurements. Transcription levels were analyzed by qRT-PCR with the primers from Table S3, using the equipment described previously (32). Expression levels were quantified relative to the housekeeping gene (h2A) with the ΔΔCT method (62).
cDNA amplification assay. PCRs were performed with 2 l cDNA (ϳ2 g) per reaction with the primers listed in Table S3 and Phusion high-fidelity DNA polymerase (Thermo Fisher Scientific), according to the manufacturer's instructions. PCR fragments were analyzed by agarose gel electrophoresis.
Secondary metabolite extraction. Two agar plates were inoculated with 10 6 spores per strain and incubated for 7 days under asexual or sexual development-inducing conditions. Two plugs were punched out per plate with a 50-ml Falcon tube. The plugs were homogenized with a 20-ml syringe and mixed with 8 ml H 2 O (LC-MS grade; Merck) and 8 ml ethyl acetate (LC-MS grade; Roth) at 220 rpm agitation at 20°C overnight. Samples were centrifuged at 2,500 rpm for 10 min at 4°C. The upper phase was collected and evaporated. The remaining metabolites were reconstituted in methanol (LC-MS grade; Fisher Scientific) for LC-MS analysis.
LC-MS analysis of secondary metabolites. A Q Exactive Focus Orbitrap mass spectrometer coupled with an UltiMate 3000 high-performance liquid chromatography (HPLC; Thermo Fisher Scientific) was used to examine the reconstituted metabolites. The HPLC column (Acclaim 120, C 18 , 5 m, 120 Å, 4.6 by 100 mm; Thermo Fisher Scientific) was loaded with 5 l extract per sample and a linear acetonitrile with 0.1% (vol/vol) formic acid in H 2 O with 0.1% (vol/vol) formic acid gradient (from 5% to 95% [vol/vol] acetonitrile with 0.1% formic acid in 20 min, with an additional 10 min with 95% [vol/vol] acetonitrile with 0.1% formic acid) at a flow rate of 0.8 ml/min at 30°C was applied. The measurements were conducted in a mass range of m/z 70 to 1,050 in positive mode. For tandem MS (MS 2 ) spectra, a stepped collision energy of 20, 30, and 40 eV was applied. Data were analyzed with Xcalibur 4.1 (Thermo Fisher Scientific) and FreeStyle 1.4 (Thermo Fisher Scientific).