A Small Cysteine-Free Protein Acts as a Novel Regulator of Fungal Insect-Pathogenic Lifecycle and Genomic Expression

ABSTRACT Small secreted proteins (SSPs), particularly cysteine-rich proteins secreted during fungal infection, comprise virulence effectors in plant-pathogenic fungi but remain unknown in insect-pathogenic fungi. We report here that only a small cysteine-free protein (CFP) is indispensable for insect pathogenicity of Beauveria bassiana among 10 studied SSPs (99 to 274 amino acids [aa]), including seven hypothetical proteins containing 0 to 12 Cys residues. CFP (120 aa) features an N-terminal signal peptide (residues 1 to 17), a nuclear localization signal motif (residues 24 to 57), and no predictable domain. Its homologs exist exclusively in insect-pathogenic Cordycipitaceae and Clavicipitaceae. Fluorescence-tagged CFP fusion protein was localized in the nucleus but extracellularly undetectable, suggesting an inability for CFP to be secreted out. Disruption of cfp resulted in abolished pathogenicity via normal cuticle infection, attenuated virulence via hemocoel injection, compromised conidiation capacity versus little growth defect, impaired conidial coat, blocked secretion of cuticle-degrading enzymes, impeded proliferation in vivo, disturbed cell cycle, reduced stress tolerance, and 1,818 dysregulated genes (genomic 17.54%). Hundreds of those genes correlated with phenotypic changes observed in the disruption mutant. Intriguingly, nearly 40% of those dysregulated genes encode hypothetical or unknown proteins, and another 13% encode transcription factors and enzymes or proteins collectively involved in genome-wide gene regulation. However, purified CFP showed no DNA-binding activity in an electrophoretic mobility shift assay. These findings unveil that CFP is a novel regulator of fungal insect-pathogenic life cycle and genomic expression and that cysteine richness is dispensable for distinguishing virulence effectors from putative SSPs in B. bassiana. IMPORTANCE Small cysteine-rich proteins secreted during plant-pathogenic fungal infection comprise virulence effectors. Our study confirms that only a cysteine-free protein (CFP) is determinant to insect-pathogenic fungal virulence among 10 small putatively secreted proteins containing 0 to 12 Cys residues. Disruption of cfp abolished insect pathogenicity and caused not only a series of compromised cellular events associated with host infection and disease development but also dysregulation of 1,818 genes, although no DNA-binding activity was detected in purified CFP samples. Nearly 13% of those genes encode transcription factors and enzymes or proteins collectively involved in transcriptional regulation. Altogether, CFP serves as a novel regulator of the fungal insect-pathogenic life cycle and genomic expression. Cysteine richness is dispensable for distinguishing virulence effectors from the fungal SSPs.

determinants (4)(5)(6)(7), which are presumably involved in fungus-insect interactions crucial for fungal infection and lethal action. However, only a few of those have been characterized.
Fungus-insect interactions start from conidial adhesion to insect cuticle, which is impregnated with chitin and proteins and covered with an epicuticular layer composed of lipids and hydrocarbons (5,8,9). Conidial germination leads to formation of germ tubes and invasion of hyphae into the host body through cuticular penetration by secreted cuticle-degrading enzymes, such as subtilisin-like Pr1 family proteases (10), and/or mechanic pressure of appressoria formed at hyphal tips. Such appressoria are often seen in Metarhizium spp. (11,12) but rarely in typical Beauveria bassiana strains, implicating somewhat distinctive mechanisms underlying the host infection of the Beauveria lineage that could have evolved insect pathogenicity from plant pathogens or endophytes approximately 130 million years earlier than the Metarhizium lineage (7). The fungal cuticle infection is initiated by G protein-coupled receptors, which mediate host recognition and the activation of downstream pathways to regulate appressorial formation in Metarhizium spp. (4,12) or developmental and genetic networks in B. bassiana (13), and ultimately leads to hyphal entry into host hemocoel. The mentioned receptors constitute a large family of transmembrane receptors responsible for transduction of external signals into intracellular responses in association with complicated signaling networks (14). Upon entry into hemocoel, hyphae encounter, and hence must overcome, cellular and humoral immune responses of the host (9, 15) while they turn into unicellular thin-wall hyphal bodies (blastospores) to facilitate intrahemocoel proliferation by yeast-like budding until host death from mummification by fungal cells (16)(17)(18). In the dying host, hyphal bodies turn back into hyphae to penetrate the cuticle again for outgrowth on insect cadaver surfaces, where conidia are produced for a new infection cycle or dispersal/survival in host habitats. For an insect-pathogenic fungus, therefore, its ability to grow out of insect cadavers for conidiation can be considered an excellent indicator of its enabling host infection via cuticular penetration. In B. bassiana, dimorphic (hypha-blastospore and vice versa) transition critical for fungal kill action and infection cycle is an asexual development process controlled by the key activator brlA or abaA of the central developmental pathway (CDP) (19) and also orchestrated by other factors involved in transcriptional activation of brlA or abaA (20)(21)(22)(23)(24). The factors required for or associated with asexual development comprise extraordinary virulence factors, because single-gene disruptions may result in abolished or nearly abolished hyphal invasion into host body via cuticular penetration and attenuated virulence via hemocoel injection for cuticle-bypassing infection.
Small secreted proteins (SSPs), particularly cysteine-rich proteins secreted during fungal infection, are a class of presumable key virulence effectors in insect mycopathogens (7,9) but have not been explored as well as those involved in pathogen-plant interactions (25)(26)(27). In plant-pathogenic fungi, small cysteine-rich proteins have been characterized as important virulence effectors, such as AvrII, Avr4, Avr4E, and Avr9 to be recognized by resistant proteins in the host plant (28,29) and Ecp1, Ecp2, Ecp4, and Ecp5e as extracellular proteins to invoke a hypersensitive response of host plant to arrest fungal growth (30,31). Revealed in a previous transcriptomic analysis, several hundreds of genes encoding secreted and nonsecreted proteins were induced in the first 48-h infection course of B. bassiana penetrating the cuticle of Plutella xylostella and colonizing the pest hemocoel (32), implicating their involvements in the fungusinsect interaction. Some of those genes have proved determinant to virulence due to reduced or even abolished pathogenicity in the absence of each. The characterized genes encode the vacuolar protein VLP4 (33), cyclophilin B (CypB) (34), and three nonsecreted proteins, including the class II lysyl-tRNA synthetase KRS (35), cytoplasmic Rei1 as a pre-60S subunit export factor (36), and nuclear Ssr4 as a cosubunit of chromatin-remodeling SWI/SNF and RSC complexes (37). Among those, only VLP4 is a small cysteine-rich protein (13 Cys resides in 230 amino acids [aa]). Many more identified SSPs remain unexplored. This study seeks to identify possible virulence effectors from 10 selected SSPs (99 to 274 aa), which share a predicted N-terminal signal peptide (NSP) and are presumably secreted in the fungal infection course. Our goal is to test the hypothesis by clarifying whether, how, and why SSPs are determinant to the fungal virulence denoted by median lethal actions of single-gene disruption mutants against a model insect. Among the 10 SSPs characterized in this study, only a small cysteinefree protein (CFP; previously named PspA-putative secreted protein A) proved determinant to virulence of B. bassiana. As presented below, CFP is involved in a wide array of cellular events and required for the fungal infection cycle and genome stability.

RESULTS
Recognition and sequence features of fungal CFP homologs. The CFP sequence (NCBI accession EJP69086) annotated as hypothetical protein in the B. bassiana genome (5) consists of 120 aa (13.18 kDa), features an NSP (residues 1 to 17) and a nuclear localization signal (NLS) motif (residues 24 to 57) adjacent to the NSP but contains no Cys residue. The CFP sequence also does not contain any known function domain predictable at https://www.ncbi.nlm.nih.gov/Structure/ or http://smart.embl-heidelberg.de/. BLASTp analysis with the query CFP sequence resulted in recognition of one or two homologs (96 to 120 aa) in Beauveria brongniartii and Cordyceps fumosorosea (Cordycipitaceae) and one to three homologs (128 to 145 aa) in four Metarhizium species (Clavicipitaceae) (Fig. 1A). These insect pathogens usually undergo asexual cycles. Unexpectedly, no homolog was found in other entomopathogenic (Ascosphaera apis, Cordyceps militaris, Cordyceps confragosa, Moelleriella libera, and Sporothrix insectorum) and nonentomopathogenic (Aspergillus, Botrytis, Fusarium, Neurospora, and Pyricularia) ascomycetes. The identified CFP homologs (see Fig. S1A in the supplemental material) fall into distinct phylogenetic clades and share a similar NSP (17 aa) and an NLS (29 to 35 aa), and most of them lack any predictable domain. Exceptionally, three of nine Metarhizium homologs share a unique F-BAR_FCHO2 domain. The F-BAR (FES-CIP4 homology and bin/amphiphysin/Rvs) domain exists in proteins involved in membrane dynamics and actin reorganization, while the FCHO2 (FCH domain only 2 protein) domain remains functionally unknown (38). WoLF PSORT analysis (https://wolfpsort.hgc.jp/) revealed a possible involvement of B. bassiana CFP in extracellular activity and membranes and/or lumens of mitochondria, Golgi apparatus, peroxisomes, and endoplasmic reticulum ( Fig. S1B) but failed to clue a link of CFP to the nucleus as predicted at http://nls-mapper.iab.keio.ac.jp/.
These analyses implicate that CFP is distinct from those cysteine-rich virulence effectors in plant-pathogenic fungi (25,27) and that its homologs seem to exist only in some asexual insect pathogens, particularly the Beauveria and Metarhizium lineages, as main sources of fungal pesticides (1).
Subcellular localization and transcriptional profile of CFP in B. bassiana. The predicted NLS implicates a nuclear localization of CFP. This implication was clarified by subcellular localization of a green fluorescence protein (GFP)-tagged CFP fusion protein (CFP::GFP) expressed in the wild-type strain B. bassiana ARSEF 2860 (designated WT). Revealed by laser scanning confocal microscopy (LSCM), CFP::GFP localized exclusively in the nuclei of hyphal cells collected from a 60-h-old Sabouraud dextrose broth plus yeast extract (SDBY) culture and stained with the nucleus-specific dye 49,6-diamidino-2-phenylindole (DAPI) (row 1 in Fig. 1B) and also in the nuclei of aerial conidia produced on SDBY plus agar (SDAY) at an optimal regime of 25°C in a light/dark (L:D) cycle of 12 h:12 h or in full darkness (Fig. 1C). For subcellular responses of CFP to stress cues, hyphae from 48-h-old SDBY cultures were incubated for 12 h in Czapek-Dox broth (CDB) deficient of carbon and nitrogen sources, and those from 58-h-old cultures were triggered for 90 min with H 2 O 2 (4 mM) or NaCl (0.4 M) in one-fourth SDBY (onefourth nutrition strength of SDBY) or exposed to UVB irradiation (1 J/cm 2 ) after being suspended in phosphate-buffered saline (PBS). As a result, CFP::GFP accumulated heavily in the nuclei and weakly in the cytoplasm comprising organelles (rows 2 to 5 in Fig. 1B). The mean ratio (22.7) of nuclear versus cytoplasmic green fluorescence intensities (N/C-GFI) quantified from the hyphae not triggered with any stress cue was 4.6to 6.5-fold of those from the hyphae triggered with the stress cues of carbon/nitrogen starvation, high osmolarity, oxidation, and UVB irradiation (Fig. 1D). Western blotting with an anti-GFP antibody confirmed the presence of CFP::GFP in the nuclear protein extract isolated from the 60-h-old SDBY culture ( Fig. 1E) but revealed no signal for its presence in the precipitant of the culture supernatant (Fig. 1F), suggesting a nonsecreted status of CFP. Moreover, transcription of cfp was increasingly upregulated in the WT culture during 7-day incubation on SDAY at the optimal regime and peaked on day 6 ( Fig. 1G). In contrast, its transcript level in the WT cultures grown in trehalose-peptone broth (TPB) mimicking insect hemolymph (18,33) peaked on day 2 and declined to a minimum on day 4, suggesting a status of its induced expression during early colonization of host hemocoel.
These data indicate that CFP is a nucleus-specific protein that may partially migrate to the cytoplasm in response to stress cues. The stress-responsive migration is likely associated with the activities of those organelles predicted with WoLF PSORT. Taken together with transcript changes, a nuclear localization of CFP under normal and stressful conditions implicates its involvement in the fungal nuclear events.
CFP is indispensable for fungal pathogenicity and virulence. The disruption and complementation mutants of cfp were created as described in Materials and Methods and identified via PCR and Southern blot analyses with paired primers and amplified probe (see Fig. S2). The Dcfp mutant lost almost all ability to infect Galleria mellonella fourth-instar larvae via normal cuticle infection (NCI) initiated by topical application (immersion) of a 10 7 -conidia/ml suspension and also was compromised severely in virulence via cuticle-bypassing infection (CBI) by intrahemocoel injection of ;500 conidia per larva, as indicated by time-survival trends in either infection mode and median lethal time (LT 50 ) estimated from the resultant time-mortality records ( Fig. 2A). The control (WT and complemented) strains caused 100% mortality within 9 days and a mean LT 50 of 5.72 days via NCI, contrasting to only ;10% mortality caused by Dcfp 13 days post-NCI. The Dcfp LT 50 via CBI was prolonged by 57% versus a mean of 4.26 days for the control strains. In contrast, single-gene disruptions of nine other SSPs resulted in insignificant LT 50 changes (Tukey's honestly significant difference [HSD], P . 0.05) via NCI (Fig. 2B). These SSPs included hydrophobin-like protein (BBA_02999), alkaline foam protein B precursor (BBA_07009), LysM domain-containing protein (BBA_08602), and six hypothetical proteins containing 0 (BBA_09280), 1 (BBA_07293), 4 (BBA_07148, BBA_07621, and BBA_08776) and 12 (BBA_02631) Cys residues. Due to unaffected virulence, the nine mutants were not studied further. An emphasis turned to explore why cfp is required for fungal insect pathogenicity and virulence-related cellular events postinfection.
The role of cfp in NCI was revealed by fungal outgrowths on the CBI-killed larvae. The control strains grew readily out of insect cadavers and formed a heavy layer of hyphal mass on their surfaces 8 days postdeath, contrasting to an inability for the Dcfp mutant to grow out (Fig. 2C). The inability implicated an essentiality of cfp for either intrahemocoel hyphae to penetrate through insect cuticle for outgrowth or hyphal invasion into insect body via NCI. Next, total activities of extracellular (proteolytic, chitinolytic, and lipolytic) enzymes (ECEs) and Pr1 family proteases collectively required for cuticle degradation during host infection (10) were quantified from the supernatants of submerged cultures from 3-day shaking incubation of a 10 6 -conidia/ml suspension in CDB containing bovine serum albumin (BSA) as the sole nitrogen source and enzyme inducer. Despite little change in biomass level, total ECEs and Pr1 activities decreased by 80% and 96%, respectively, in the Dcfp mutant versus that in WT cultures (Fig. 2D), suggesting a main cause for the mutant inability to infect insects and grow out of cadavers through cuticular penetration.
To observe the proliferation of hyphal bodies in host hemocoel, hemolymph samples taken from surviving larvae 6 days post-NCI were examined under a microscope. Budding hyphal bodies were present in the samples of the larvae infected by two control strains but not in those of the mutant-infected larvae (Fig. 2E). This observation implicated impeded development of hyphal bodies in vivo, which was supported by reduced blastospore production in the 3-day-old cultures of a 10 6 -conidia/ml TPB mimicking insect hemolymph. Compared to the WT cultures, the Dcfp cultures showed 87% reduction in blastospore production but only 26% biomass decrease (Fig. 2F), indicating a 61% contribution of cfp to dimorphic transition in vitro. The latter estimate coincided well with a lethal action of the mutant prolonged by 57% through CBI. Revealed by flow cytometry, the mutant blastospores displayed altered size and complexity (Fig. 2G) and disturbed cell cycle with shortened G 1 and G 2 phases and a prolonged S phase (Fig. 2H).
Altogether, CFP is indispensable for insect pathogenicity and virulence of B. bassiana. The indispensability relies upon its essential role in secretion of NCI-required enzymes and also in dimorphic transition to accelerate proliferation in vivo and host death.
CFP is dispensable for hyphal growth but functional in the stress response. To explore the possible effects of cfp disruption on hyphal growth and invasion into host body, fungal colonies were initiated by spotting 1-ml aliquots of a 10 6 -conidia/ml suspension on the plates of rich SDAY, minimal CDB plus agar (CDA) and CDAs amended with different carbon or nitrogen sources. After an 8-day incubation at the optimal regime, little variability in colony size was found among the tested strains on the tested media ( Fig. 3A). Exceptionally, the Dcfp mutant growth was facilitated significantly on SDAY and the nitrogen source NH 4 NO 3 and moderately suppressed only on sodium acetate as the carbon source. These data indicated that cfp was dispensable for hyphal growth on the scant media like oligotrophic insect integument.
The fungal kill action relies upon an ability to overcome stresses generated from host immune responses during cuticular penetration and subsequent hemocoel colonization (19,39). Thus, fungal colonies were initiated as aforementioned on CDA containing chemical stressors. Compared to the control strains, the Dcfp mutant became more sensitive to NaCl, menadione, H 2 O 2 , and Congo red during hyphal growth but showed null response to the noncation osmotic agent sorbitol or the cell wall antagonist calcofluor white ( Fig. 3B and C). The mutant also showed increased sensitivity to 6-h heat shock at 42°C during the 6-day growth recovery at 25°C after exposure of 2-day-old SDAY colonies to 42°C. Its increased sensitivity to oxidative stress coincided well with 27% and 34% reductions in total activities of superoxide dismutases (SODs) and catalases (Fig. 3D), respectively, required for removal of superoxide anions and H 2 O 2 (19).
Importance of CFP for conidiation capacity and conidial quality. Conidiation capacity and conidial quality are important for large-scale production of fungal pesticides. The conidial yield measured from the SDAY cultures initiated by spreading 100 ml of a 10 7 -conidia/ml suspension per plate at the optimal regime decreased by 77% and ;50% in the Dcfp mutant versus that in the WT on days 4 and 5 to 8, respectively (Fig. 4A), despite little difference in their biomass levels during the 8-day incubation ( Fig. 4B). Moreover, several indices of conidial quality were compromised in the Dcfp mutant relative to that in the WT, including a 33% decrease in hydrophobicity and 32% increase in median germination time (GT 50 ) at 25°C (Fig. 4C) and 58% and 23% reductions in median lethal time (LT 50 ) and dose (LD 50 ) for tolerance to 45°C wetheat stress and UVB irradiation, respectively (Fig. 4D). Furthermore, carbohydrate epitopes of conidial surfaces with pathogen-associated molecule patterns (PAMPs) to be perceived by host PAMP-recognition receptors (9,25) were probed in fluorescent lectin-binding assays. The fluorescence intensity changes of labeled Dcfp versus WT conidia demonstrated 37%, 106%, 27%, and 12% increases in the contents of a-glucose and a-N-acetylglucosamine (GlcNAc), b-GlcNAc and sialic acid residues, b-galactose, and mannose residues labeled by concanavalin A (ConA), wheat germ agglutinin (WGA), peanut agglutinin (PNA), and Galanthus nivalis lectin (GNL), respectively (Fig. 4E). Previously, the conidial coat of B. bassiana was characterized as a well-defined outermost layer of hydrophobin-borne rodlet bundles (40). Revealed by scanning electronic microscopy (SEM), such a conidial coat was intact for the control strains, contrasting to impairment or disappearance of most rodlet bundles on conidial surfaces of the Dcfp mutant (Fig. 4F), although conidial size and complexity differed insignificantly between the mutant and control strains (Fig. 4G). In the Dcfp mutant, the impaired conidial coat coincided well with reduced hydrophobicity. Both altered carbon epitopes and impaired rodlet bundles on conidial surfaces suggested an important role of CFP in sustaining cell wall integrity and escaping PAMPs from host receptors to minimize host immune responses.
Profound effect of CFP on genomic expression. For in-depth insight into the effects of CFP on pathogenicity and virulence, transcriptomes were constructed based on three 3-day-old cultures (replicates) of the Dcfp and WT strains grown on cellophane-overlaid SDAY plates at the optimal regime. The resultant data set comprised 10,430 genes mapped to the B. bassiana genome (5) or not mapped (Fig. 5A). Identified from the data set were 1,818 differentially expressed genes (DEGs; up/down ratio, 1,006:801) at significant levels (log 2 R # -1.00 or $ 1.00, q , 0.05), including six and five expressed only in Dcfp and WT strains, respectively (see Table S1). These DEGs comprised 17.54% of the whole genome. Notably, 423 DEGs (up/down ratio, 204:219) encode hypothetical proteins, and another 293 (up/down ratio, 120:173) are not annotatable, including 127 new genes likely involved in the CFP-mediated cellular events.  Table S2). Shown with main GO terms in Fig. 5B, the cellular component class comprised 818 DEGs (up/down ratio, 495:323), which were mostly enriched to the terms cellular component, integral component of membrane, membrane, extracellular region, nuclear chromatin, and membrane-bounded vesicle and eisosome. Enriched to molecular function class were 428 DEGs (up/down ratio, 252:166) involved mainly in oxidoreductase activity, catalytic activity, iron/copper ion binding, hydrolase activity, heme binding, monooxygenase activity, sequence-specific DNA binding, transferase activity, pyridoxal phosphate binding, transcription regulatory region DNA binding, and phosphopantetheine binding. Enriched to biological process were 429 DEGs (up/down ratio, 256:173) involved mainly in the processes of oxidation-reduction, transmembrane transport, metabolism, cellular response to drug, emericellamide biosynthesis, fatty acid biosynthesis, biosynthesis, protein catabolism, glutathione metabolism, interaction with host via protein secreted by type II secretion system, DNA replication initiation, cell wall macromolecule catabolism, and amino acid metabolism. This class also included the smaller, but important, terms (3 to 5 DEGs each) methylglyoxal catabolism, phospholipid biosynthesis, mitotic spindle assembly, regulation of nitrogen utilization, vacuolar transport, polyphosphate metabolism, Zn 21 transmembrane transport, lysosomal microautophagy, cellular protein modification, RNA catabolism, and biosyntheses of several secondary metabolites.
Kyoto  Transcriptomic insights into pleiotropic effects of CFP. Both GO and KEGG enrichment analyses revealed several dozens of DEGs involved in the oxidoreductase activity and oxidation-reduction process, coinciding well with reduced SOD and catalase activities and increased sensitivities to menadione and H 2 O 2 . Thirty-two DEGs were involved in the cell cycle pathway and likely associated with impeded proliferation of hyphal bodies in vivo and reduced blastospore production in vitro, revealing an effect of the disturbed cell cycle on the mutant virulence attenuated via CBI. Another 32 DEGs (up/down ratio, 17:15) were involved in cuticle degradation and fungal pathogenicity via NCI (see Table S4). These DEGs encode cutinases, chitinases, subtilase-like proteins, subtilisin-like proteases, glycoside hydrolases, and secreted lipases aside from bassianolide nonribosomal peptide synthetase and beauvericin biosynthetic protein likely involved in insecticidal activity (41-43). q values for all genes identified in the transcriptomes, which were generated from three 3-day-old SDAY cultures (replicates) of Dcfp and WT strains grown at the optimal regime of 25°C and L:D 12 h:12 h. Differentially expressed genes (DEGs) are those significantly downregulated (log 2 R # 21) or upregulated (log 2 R $ 1) at the level of q value of ,0.05. The rest of the genes are insignificantly affected (21 # log 2 R # 1 or q $ 0.05 when log 2 R # 21 or $ 1). (B and C) Counts of DEGs significantly enriched (P , 0.05) for three function classes (main GO terms shown) and KEGG pathways. Table S4, 29 DEGs (up/down ratio, 13:16) encode proteins involved in cell wall integrity, while eight others (up/down ratio, 2:6) encode heat shock proteins (HSPs) involved in responses to heat shock and other stresses (17,(44)(45)(46). Unexpectedly, the hydrophobin genes hyd1 (BBA_03015) and hyd2 (BBA_00530) required for biosynthesis and assembly of hydrophobin rodlet bundles (40) were upregulated. Moreover, 148 DEGs (up/down ratio, 114:34) encode varieties of transporters, permeases, ATPases, and analogs, which are all vital for maintenance of vacuolar/cellular homeostasis and cellular transport of nutrients, metal cations, and xenobiotics and also for multiple stress responses outside and inside host insect, suggesting a role of CFP in the cellular processes. Eleven DEGs were involved in autophagy, mitophagy, phagocytosis, and endocytosis, although none of them encodes an ATG protein known in fungal insect pathogens (47). Thirty-two DEGs (up/ down ratio, 23:9) encode proteins in various signaling pathways, including 10 involved in the mitogen-activated protein kinase (MAPK) pathways, although none of them is a component of the Hog1, Slt2, and Fus3 cascades that mediate asexual development, multiple stress responses, and/or virulence (23,39). Similarly, none of 11 DEGs (up/down ratio, 7:4) presumably involved in fungal development and sporulation is the CDP gene brlA or abaA required for conidiation and dimorphic transition (19), hinting at some other factors responsible for the Dcfp defects in aerial conidiation and submerged blastospore production. Additionally, many more DEGs (up/down ratio, 146:121) were involved in biogenesis, biosynthesis, and carbon/nitrogen metabolisms, implicating pleiotropic effects of CFP on the fungal life cycle.

Also listed in
Notably, 58 DEGs (up/down ratio, 34:24) encode transcription factors and analogs likely involved in direct transcription mediation. Many more DEGs (up/down ratio, 126:49) also encode proteins/enzymes as direct and/or indirect gene regulators, since they function in posttranslational modifications, RNA structure and function, and DNA replication, repair, binding, and regulatory activity as well as in translation. The dysregulation of so many genes involved in transcriptional mediation highlights a genomewide effect of CFP on the fungal life cycle.
CFP has no DNA-binding activity. To elucidate the genome-wide effect, electrophoretic mobility shift assays (EMSAs) were performed to probe whether purified CFP extract could bind promoter DNA fragments of 12 genes sharply repressed in the Dcfp mutant (see Fig. S3A). Agarose gel electrophoresis showed no shifting sign for any DNA fragment bound by uploaded CFP samples (0.8 to 4.0 mg) and also no shifting sign for any CFP sample bound to the DNA sample in the gels stained with Coomassie brilliant blue (Fig. S3B). The nonshifting signals of gradient CFP samples in binding the tested DNA samples were distinguished from shifting signals of the promoter DNA extract of ole1 (D9-fatty acid desaturase gene) bound by gradient HapX (bZIP transcription factor) samples as a positive control (Fig. S3C) but similar to nonshifting signals of the ole1 DNA reacting with gradient thioredoxin (Trx1) samples as a negative control (Fig. S3D) (48). These data indicate an inability for CFP to bind the promoter regions of potential target genes for their transcription mediation.

DISCUSSION
CFP homologs seem to exist only in certain asexual entomopathogens of Hypocreales. In B. bassiana, unique CFP in a list of SSPs was postulated to function during host infection (32) and proved essential for insect pathogenicity via NCI and virulence via CBI in this study. Both NCI and CBI bioassay data highlight that CFP is not only a super virulence factor but also a core regulator of the insect-pathogenic life cycle. As NSP-bearing SSPs, cysteine-rich VLP4 and cysteine-free CypB were previously characterized as virulence factors due to their marked contributions to the fungal kill action via NCI (33,34). However, nine other cysteine-rich or -deficient SSPs were dispensable for the fungal virulence in this study. The previous and present studies indicate that SSPs presumably involved in fungus-insect interactions do comprise, but are not necessarily, virulence effectors. As a cysteine-free SSP, CFP is very different from VLP4 and those elucidated as virulence factors in plant-pathogenic fungi (28)(29)(30)(31). The fact that CFP::GFP was localized in the nucleus but extracellularly undetectable indicates that NSP-bearing CFP is not a secreted protein. Its pleiotropic effect on B. bassiana is discussed below.
The fungal insect pathogenicity via NCI can be abolished or compromised severely in different fashions. First, impeded hyphal growth on scant media resembling oligotrophic insect surfaces may inhibit or slow hyphal invasion into the host body. As examples, NCI was abolished in the absence of fus3, ste7, or ste11 in the MAPK Fus3 cascade (23,49) and of ssr4 encoding a nuclear cosubunit of chromatin-remodeling SWI/SNF and RSC complexes (37) and greatly compromised in the absence of rei1 encoding a cytoplasmic pre-60S subunit export factor (36). A similar situation occurred when histone H3 was hypo-or hyperacetylated by loss-of-function mutation of histone acetyltransferase Gcn5 or deacetylase Rpd3 (20,21). The mutants of these genes showed severe growth defects on scant media and reduced conidiation capacity due to sharp repression of CDP genes or dysregulation of many genes involved in carbon/ nitrogen metabolism. Second, an incapability of hyphal invasion into the host body is related to no conceivable defect in hyphal growth but tied to abolished aerial conidiation and submerged blastospore production due to inactivation of CDP by brlA or abaA disruption (19). In our Dcfp mutant, abolished NCI was attributable mainly to blocked secretion of cuticle-degrading enzymes (10) rather than to either an accountable growth defect or repressed expression of CDP or Fus3-cascaded genes. Our transcriptomic analysis revealed 32 dysregulated genes encoding NCI-required enzymes/ proteins. Taken together with the incapable growth of the mutant out of insect cadavers, we infer that CFP may serve as an active player in synthesis and secretion of diverse cuticle-degrading enzymes and hence is indispensable for insect pathogenicity of B. bassiana via NCI.
The Dcfp virulence attenuated via CBI indicates a pivotal role of CFP in postinfection cellular events crucial for virulence. The transcript-declining trend of cfp in the submerged TPB cultures during the 4-day incubation suggests its role in the fungal proliferation in vivo. Upon entry into host hemocoel, dimorphic transition is critical for intrahemocoel proliferation by yeast-like budding to accelerate insect death from mycosis development. In the Dcfp mutant, blocked dimorphic transition was revealed by both the absence of hyphal bodies in vivo and a 61% decrease of blastospore production in the TPB cultures. Moreover, the mutant conidia were coated with increased amounts of carbohydrate epitopes and impaired hydrophobin rodlet bundles (41). Our transcriptomic data comprised dozens of dysregulated genes involved in cell wall integrity. Of those, upregulation of hyd1 and hyd2 suggest other cell wall-related genes that could be associated with an impaired coat and reduced hydrophobicity of the mutant conidia, such as those genes involved in biogenesis and biosynthesis, although their roles in hydrophobin synthesis and assembly remain unclear. For the Dcfp mutant, both altered carbohydrate epitopes and dysregulated genes implicate that impaired surfaces of injected conidia could have more PAMPs exposed to host receptors (9), which could lead to more intensive host immunity responses to be overcome by longer interaction in vivo. This implication hints at a possible role of CFP in hiding PAMPs to minimize host immune responses during NCI and hemocoel colonization. Additionally, bassianolide nonribosomal peptide synthetase and beauvericin biosynthetic protein are known as the fungal virulence factors (41)(42)(43). The drastic repression of their coding genes suggests reduced secretion of such insecticidal metabolites by the Dcfp mutant, thereby helping to explain a delay of its lethal action against the injected larvae.
In the Dcfp mutant, compromised conidiation and blastospore production were not due to repressed CDP genes. Instead, the mutant had several hundreds of dysregulated genes involved in carbon/nitrogen metabolism, transmembrane transport, cellular homeostasis, signal transduction, autophagy, phagocytosis, and endocytosis aside from those involved in cell cycle. Dysregulation of these genes suggests a core role of CFP in diverse cellular processes and events which are also influential on asexual development and stress responses. The core role is likely linked to a nuclear localization of CFP and its predicted activity in association with various organelles. However, nuclear CFP had no DNA-binding activity in EMSAs and hence was not capable of directly regulating gene transcription by itself. Previously, coding genes of Ssr4 and Rei1 were increasingly upregulated during host infection by B. bassiana (32). Like CFP, Ssr4 and Rei1 had no DNA-binding activity but orchestrated transcription of 2,517 and 1,400 genes, which were dysregulated in the respective Dssr4 and Drei1 mutants displaying severe defects in growth, conidiation, pathogenicity, and virulence (36,37). Many of those genes were involved in carbon/nitrogen metabolisms and transport of nutrients, and dozens were involved in transcription, translation, and posttranslational modifications; but none were involved in CDP. The genome-wide regulatory roles of Ssr4 and Rei1 were speculated to rely upon a role of Ssr4 as a cosubunit of the SWI/ SNF and RSC complexes that can bind the binding sites of transcription factors (50,51) and of Rei1 in orchestrating dissociation and recycling of nucleocytoplasmic pre-60S factors (52)(53)(54). In this study, the Dcfp mutant had many more dysregulated genes (233) involved in direct and indirect gene regulation than those seen in the Dssr4 and Drei1 mutants. This offers insight into the dysregulation of another 1,585 genes, nearly half (up/down ratio: 324/392) of which encode hypothetical or unknown proteins. So many of the unknown genes orchestrated by CFP are likely critical for the fungal insect-pathogenic life cycle. However, it is unclear how CFP is tied to dozens of transcription factors required for gene regulation and to many more proteins/enzymes that may mediate gene transcription in direct and indirect fashions. Since CFP::GFP was localized in the nucleus and also associated with predicted organelles, we speculate that, like Ssr4 and Rei1, CFP might act as a component of certain nuclear and/or nucleocytoplasmic protein complexes. This hypothesis opens a door to explore what complexes recruit the component CFP lacking any predictable domain as a clue to clarify its genome-wide regulatory role and why so small of a simple protein has a profound effect on insect pathogenicity and virulence-related cellular events in B. bassiana.
Conclusively, not all SSPs involved in fungus-insect interactions are virulence effectors. Cysteine richness is not a necessary criterion for SSPs to be considered so. This is unveiled by indispensability of CFP and dispensability of nine other cysteine-rich or -deficient SSPs for the fungal virulence and offers novel insight into roles of fungal SSPs in the insect-pathogenic life cycle.

MATERIALS AND METHODS
Recognition and bioinformatic analysis of fungal CFP homologs. The amino acid sequence of CFP found in the previous B. bassiana transcriptome (32) was used as a query to search through the NCBI databases of ascomycetes, including entomopathogenic and nonentomopathogenic fungi, by BLAST analysis (https://blast.ncbi.nlm.nih.gov/Blast.cgi). The located homologs were structurally compared with the query by sequence alignment with DNAman 8.0 at http://www.bio-soft.net/format/ DNAMAN.htm/ and conserved domain analysis at https://www.ncbi.nlm.nih.gov/Structure/ or http:// smart.embl-heidelberg.de/, followed by phylogenetic analysis with a maximum likelihood method in MEGA7 at http://www.megasoftware.net/. NPS and NLS were predicted from each protein sequence at http://www.cbs.dtu.dk/services/SignalP-4.1/ or http://nls-mapper.iab.keio.ac.jp/. Possible extra-and intracellular activities of B. bassiana CFP were predicted with WoLF PSORT (https://wolfpsort.hgc.jp/). Subcellular localization and transcriptional profiling of CFP in B. bassiana. The coding sequence of cfp (BBA_02121) was amplified from the WT cDNA with paired primers (see Fig. S2B in the supplemental material) and inserted into linearized pAN52-GFP-bar (33) for fusion of cfp to the N terminus of gfp (U55763), forming cfp::gfp in the plasmid vectoring PgpdA (promoter) and TtrpC (terminator). The plasmid was transformed into WT via blastospore transformation, resulting in putative transformants screened by bar resistance to phosphinothricin (200 mg/ml). A transgenic strain showing strong green signal was incubated on SDAY (4% glucose, 1% peptone, and 1.5% agar plus 1% yeast extract) at 25°C and L:D 12 h:12 h or 0 h:24 h for conidiation. Collected conidia were suspended in SDBY for a 48-, 58-, or 60-h incubation on a shaking bed (150 rpm) at 25°C. LSCM was applied to determine subcellular localization of CFP::GFP expressed in the hyphae of the 60-h-old culture stained with DAPI (49,69-diamidine-29phenylindole dihydrochloride; Invitrogen) or in the conidia produced at L:D 12 h:12 h or 0 h:24 h but not stained. Subcellular responses of CFP::GFP to stress cues were examined after 48-h-old hyphae were shaken for 12 h in CDB (3% sucrose, 0.3% NaNO 3 , 0.1% K 2 HPO 4 , 0.05% KCl, 0.05% MgSO 4 , and 0.001% FeSO 4 ) lacking the carbon and nitrogen sources or 58-h-old hyphae were triggered for 90 min with 4 mM H 2 O 2 or 0.4 M NaCl in one-fourth SDBY. Alternatively, 58-h-old hyphae were suspended in 0.8-mm-thick PBS, followed by exposure to UVB irradiation of 1 J/cm 2 (weighted wavelength, 312 nm). The hyphae from each treatment were stained with DAPI, followed by LSCM analysis. ImageJ software (https://imagej.nih.gov/ij/) was used to measure green fluorescence intensities in the cytoplasm and nuclei of $30 cells in 10 hyphae from each treatment. Relative accumulation levels of expressed CFP:: GFP in the nuclei of those hyphae were estimated as N/C-GFI ratios. Additionally, supernatants were collected from the 60-h-old cultures of the transgenic and WT strains by filtration and centrifuged for 10 min at 10,000 Âg, followed by isolation of nuclear protein extracts from the cells with a nuclear and cytoplasmic protein extraction kit (Beyotime, Shanghai, China). The precipitants and the protein extracts (20 mg uploaded per lane) were probed with mouse monoclonal anti-GFP antibodies (OriGene China, Beijing, China) for the Western blot signal to determine whether CFP::GFP was secreted out of cells or fused as expected.
To examine transcriptional profiles of cfp in different WT cultures, 100-ml suspension aliquots (10 7 conidia/ml) were spread on cellophane-overlaid SDAY plates for a 7-day incubation at the optimal regime or 100-ml aliquots of 10 6 -conidia/ml TPB (CDB amended with 3% trehalose and 0.5% peptone) were incubated for 4 days on the shaking bed. From day 2 onwards, total RNAs were extracted daily from three TPB or SDAY cultures with an RNAiso Plus kit (TaKaRa, Dalian, China) and reversely transcribed into cDNAs with a PrimeScript reverse transcription (RT) reagent kit (TaKaRa). Transcripts of cfp in the cDNA samples were assessed via real-time quantitative PCR (qPCR) with paired primers (Fig. S2B) under the action of SYBR Premix Ex Taq (TaKaRa). The fungal b-actin gene was used as a reference. A threshold cycle (2 2DDCT ) method was used to compute relative transcript levels of cfp in the daily WT cultures with respect to the standard on day 2.
Creation of cfp mutants. The cfp mutants were constructed as described previously for targeted vlp4 disruption and complementation (33). Briefly, the 39 and 59 coding/flanking fragments of cfp were amplified from the WT DNA and inserted into the restriction enzyme sites of EcoRI/BamHI and XbaI/HpaI in the backbone plasmid p0380-bar, yielding p038-39cfp-bar-59cfp for integration into the WT strain for targeted cfp disruption through homologous recombination of the bar-separated 39 and 59 fragments. The full-length coding sequence of cfp with flanking regions was amplified from the WT DNA and ligated to the sites of HindIII/XbaI in p0380-sur-gateway to exchange for the gateway fragment, forming p0380sur-cfp for ectopic integration into an identified Dcfp mutant for targeted cfp complementation. Putative mutants were screened by bar resistance to phosphinothricin (200 mg/ml) or sur resistance to chlorimuron ethyl (10 mg/ml). Expected recombinant events in the colonies were identified via PCR and Southern blot analyses with paired primers and amplified probe (Fig. S2). The positive mutants Dcfp and Dcfp::cfp were evaluated together with parental WT in the following experiments of three independent replicates to generate data for one-factor analysis of variance and Tukey's honestly significant difference (HSD) test for their phenotypic differences.
Bioassays for fungal virulence. The virulence of each strain to G. mellonella larvae was assayed through two infection modes. Briefly, three groups (replicates) of ;35 larvae were immersed for 10 s in 40 ml of a 10 7 -conidia/ml suspension for NCI. For CBI, each larva in each group was injected with 5 ml of a 10 5 -conidia/ml suspension. The control included three groups of larvae immersed or injected with 0.02% Tween 80 used in conidial suspension. All treated groups were kept at 25°C and examined every 12 h for survival records until no more change was observed. LT 50 was estimated as an index of virulence via either infection mode by probit analysis of the resultant time-mortality trend in each group. Aside from Dcfp and control strains, mutants of nine other SSP genes (shown in Fig. 2B) were constructed using the same strategy as for cfp disruption and assayed for their virulence via NCI. These mutants were not studied further due to the lack of virulence change.
Assessments of cellular events associated with virulence. A series of cellular events associated with NCI and proliferation in vivo were examined to explain abolished pathogenicity and attenuated virulence of the Dcfp mutant in the bioassays. First, larvae that died from injection were incubated at 25°C to observe whether intrahemocoel hyphae can penetrate the host cuticle for outgrowth as an indicator of fungal ability to infect the insect through cuticular penetration. Second, total ECEs and Pr1 activities required for NCI (10) were quantified as units per milliliter from the supernatants of 3-day-old cultures that resulted from shaking incubation of 50 ml 10 6 conidia/ml CDB-BSA at 25°C and compared with biomass levels as described previously (17,22). Third, hemolymph samples were taken from surviving larvae 6 days post-NCI and microscopically examined for proliferation status of hyphal bodies. Next, 50-ml aliquots of a 10 6 -conidia/ml TPB mimicking insect hemolymph were incubated for 3 days on the shaking bed at 25°C, followed by measuring blastospore concentration and biomass level to compute dimorphic transition rate (number [no.] of blastospores/mg biomass) in each culture. Flow cytometry was used to assess size and complexity of 2 Â 10 4 blastospores with the readings of forward scatter (FSc) and side scatter (SSc) detectors, followed by fluorescence-activated cell sorter (FACS) analysis to determine G 1 , G 2 , and S phases of the cell cycle in 2 Â 10 4 blastospores stained with the DNA-specific dye propidium iodide.
To explore possible effect of cfp disruption on hyphal growth and invasion into a host body, each strain was grown on SDAY, CDA, and CDAs amended with different carbon or nitrogen sources by spotting 1 ml of a 10 6 -conidia/ml suspension per plate for colony initiation. After an 8-day incubation at the optimal regime, the diameter of each colony was estimated as a growth index, with two measurements taken perpendicular to each other across the center. For hyphal responses to stress cues possibly encountered during host infection and hemocoel colonization, aforementioned 8-day colony growth was initiated on CDA containing NaCl (0.7 M) or sorbitol (1 M) for osmotic stress, menadione (0.03 mM) or H 2 O 2 (2 mM) for oxidative stress, and Congo red (10 mg/ml) or calcofluor white (5 mg/ml) for cell wall-perturbing stress, followed by measuring colony diameters. For response to thermal stress, 2-day-old SDAY colonies initiated at 25°C were exposed to 42°C for 6 h, followed by 6-day growth recovery at 25°C. The sensitivity of each strain to each stress was assessed as percentage of relative growth inhibition ([D c -D s ]/D c Â 100; where D c and D s are diameters of control and stressed colonies, respectively). Additionally, a catalase activity assay kit (Jiancheng Biotech, Nanjing, China) and SOD activity assay kit (Sigma) were used to assay total catalase and SOD activities (U/mg) in the protein extracts isolated from the 3-day-old SDAY cultures according to the manufacturers' guides.
To reveal possible effect of cfp disruption on conidial quality associated with virulence, 100-ml aliquots of a 10 7 -conidia/ml suspension were sprayed on SDAY plates (9-cm diameter) and incubated for 8 days at the optimal regime. From day 4 onwards, conidial yield was measured as the number of conidia per unit area (cm 2 ) of culture from three samples taken from each plate culture with a cork borer (5mm diameter) as described previously (22). During the period, biomass levels were also measured from cellophane-overlaid SDAY cultures initiated with the same method. Conidial quality was assessed as indices of hydrophobicity in an aqueous-organic system, GT 50 (h) for 50% germination at 25°C, LT 50 (min) for tolerance to 45°C wet-heat stress, and LD 50 (J/cm 2 ) for resistance to UVB irradiation (weighted wavelength, 312 nm), as described previously (8,24,37). Next, conidia were labeled with the Alexa Fluor 488labeled lectins ConA, WGA, PNA, and GNL (Vector Laboratories, Burlingame, CA, USA) according to the manufacturer's guide. Fluorescence intensities of 2 Â 10 4 labeled conidia were assessed as indices of carbohydrate epitopes through FACS analysis. Mean size and complexity of 2 Â 10 4 conidia were also assessed with the FSc and SSc readings from flow cytometry. Finally, the integrity of conidial coat was shown with SEM images.
Transcriptomic analysis. Three 3-day-old cultures (replicates) of Dcfp and WT strains grown on cellophane-overlaid SDAY plates were prepared as aforementioned and sent to Lianchuan BioTech Co. (Hangzou, China) for construction and analysis of a CFP-specific transcriptome. Total RNA extraction from each culture, isolation of mRNA from total RNA, fragmentation of mRNA, and syntheses of firstand second-strand cDNAs were as described previously (36). Each double-stranded cDNA was purified and end repaired by adding a single adenine to its end. The final cDNA library was sequenced on an Illumina NovaSeq 6000 platform.
Clean tags generated by filtration of all raw reads from the sequencing were mapped to the B. bassiana genome (5). All data were normalized as fragments per kilobase of exon per million fragments mapped (FPKM). All DEGs were identified at significant levels of both log 2 R of #21 (downregulated) or $1 (upregulated) and q value of ,0.05, annotated with known or putative gene information in the NCBI protein databases, and subjected to GO analysis (http://www.geneontology.org/) for enrichments of GO terms to three function classes (P , 0.05) and KEGG analysis (http://www.genome.jp/kegg/) for pathway enrichment (P , 0.05).
EMSAs for binding activity of CFP to promoter DNA fragments. Due to the profound effect of CFP on genomic expression, EMSAs for protein-DNA interactions (55) were performed to explore whether CFP can bind promoter DNA fragments as described previously (36,37). Briefly, CFP detected in the cell lysates of Escherichia coli by SDS-PAGE analysis was purified via affinity chromatography column and dialysis and standardized to 0.4 mg/ml. The purified protein extract was used as a template to detect its binding activity to the promoter DNA fragments (amplified from the WT DNA with paired primers in Fig. S3A) of 12 selected genes drastically repressed in the Dcfp mutant. For each DNA extract, 4-ml aliquots (100 ng/ml) were mixed with 0.8, 1.6, 2.4, 3.2, and 4.0 mg (2 to 10 ml) of the CFP extract in 10-ml aliquots of binding buffer comprising 25 mM HEPES (pH 7.4), 50 mM KCl, 5 mM MgCl 2 , 0.5 mM EDTA, 1 mM dithiothreitol, and 5% glycerol. Each EMSA included two negative controls, namely, a combination of 4 ml DNA extract with no protein extract and of 4.0 mg CFP extract with no DNA extract. All of the mixtures were reacted for 30 min at ambient temperature. CFP-bound DNA fragments in the reaction were detected through agarose gel electrophoresis. The gels stained with Coomassie brilliant blue were rinsed repeatedly in washing buffer and visualized for signal of the protein bound to DNA. The signals of the gradient CFP samples in binding each DNA extract were compared with those of the ole1 (BBA_07664) DNA extract bound by gradient protein samples (0.5 to 9 mg) of the bZIP transcription factor HapX (EJP64582) as a positive control rather than by the samples of the thioredoxin Trx1 (EJP68440) as a negative control (48).
Data availability. All data generated or analyzed during this study are included in the paper and associated supplemental files. All RNA sequencing (RNA-seq) data from this study are available at the NCBI's Gene Expression Omnibus under the accession no. GSE158758.

SUPPLEMENTAL MATERIAL
Supplemental material is available online only.