BrlA and AbaA Govern Virulence-Required Dimorphic Switch, Conidiation, and Pathogenicity in a Fungal Insect Pathogen

Dimorphic insect mycopathogens infect the host by hyphal penetration through the host cuticle and switch from the penetrating hyphae to unicellular blastospores after entry into the host hemocoel, where blastospore propagation by yeast-like budding accelerates host mummification to death. The fungal virulence-required dimorphic switch is confirmed as a process of asexual development directly regulated by BrlA and AbaA, two key activators of the central developmental pathway in an insect mycopathogen. This finding unveils a novel mechanism distinct from the control of the dimorphic switch by multiple signaling systems other than the central developmental pathway in dimorphic plant and human mycopathogens, which switch from the usual yeast growth to filamentous growth required for pathogenicity through host tissue penetration.

by conidial adhesion to the host cuticle, where conidial germination leads to hyphal extension for penetration through the host cuticle by means of the actions of various cuticle-degrading enzymes (2). Upon entry into the host hemocoel, the penetrating hyphae turn into unicellular blastospores (also called hyphal bodies) that are enabled to propagate rapidly by yeast-like budding until the host is mummified to death (3)(4)(5)(6). At or near the time of host death, intrahemocoel hyphal bodies turn back into septate hyphae to penetrate the host cuticle again for outgrowth (7,8). Ultimately, aerial conidia are produced on the surfaces of host cadavers for survival and dispersal for a new infection cycle. The process of dimorphic transition, by which penetrating hyphae turn into hyphal bodies and vice versa, is critical for the virulence and asexual cycle in vivo of a fungal insect pathogen, although the mechanisms involved remain poorly understood. This process has been evidently associated with the functions of many genes characterized in B. bassiana, such as those encoding the histone acetyltransferases Gcn5 and Mst2 (7,9), the histone deacetylases Hos2 and Rpd3 (10,11), the mitogen-activated protein kinase (MAPK) kinase Ste7 (12), the Na ϩ /H ϩ antiporter Nhx1 (13), the blue-light photoreceptor VVD (6), the vacuolar protein VLP4 (14) or ATPase subunit H VmaH (15), the lysyl-tRNA synthetase KRS (16), and the cyclophilin B CypB (17). Interestingly, these studied genes are also responsible for a correlation of dimorphic transition in vitro or in vivo with aerial conidiation and transcriptional activation of key activator genes in the central development pathway, as summarized in Table 1. Thus, the previous studies suggest a closer linkage of the dimorphic switch with the developmental activators in B. bassiana than the signaling factors involved in twocomponent and heterotrimeric G protein signaling systems and the Ras and cyclic AMP (cAMP) signaling and the downstream MAPK signaling cascades, all of which are considered to control the dimorphic switch and pathogenicity in human mycopathogens (18).
The asexual cycles in vitro of filamentous fungal pathogens comprise distinct phases, including vegetative (hyphal) growth, conidiophore development, and conidiation. Hyphal growth starts from conidial germination, forming germ tubes for hyphal extension. After a period of hyphal growth, some hyphal cells differentiate into conidiogenic cells for formation of the conidiophores to support conidial production. This conidiation process is a cellular event precisely timed and genetically programmed in response a The status of dimorphic transition in vivo (SB, severely blocked; NE, not examined) was revealed through microscopic examination of hemolymph samples taken from Galleria mellonella larvae surviving a period after topical application for normal cuticle infection or intrahemocoel injection for cuticle-bypassing infection. The decrease of dimorphic transition in vitro was quantified from submerged liquid cultures of conidia grown in Sabouraud dextrose broth (S), minimal Czapek broth (C), or trehalose-peptone broth (T) mimicking insect hemolymph, as indicated by the abbreviations in parentheses. b Conidiation on the standard medium Sabouraud dextrose agar plus yeast extract (SDAY) under the optimal regime of 25°C in a light/dark cycle of 12:12 suffered severe delay (SD), extremely severe delay (ESD), or no delay (ND). The loss of conidial yield was assessed from the cultures at the end of a 7-to 12-day incubation. c The transcriptional repression of each gene was assessed from the cDNA samples derived from 3-day-old SDAY cultures under the optimal regime. NE, not examined.
to internal and external cues and is genetically controlled by the central developmental pathway consisting of the activators BrlA, AbaA, and WetA, which have been well characterized in Aspergillus and Penicillium and reviewed elsewhere (19,20). In filamentous fungi, BrlA, AbaA, and WetA activate the expression of downstream conidiationspecific genes in a hierarchical manner during the development of conidiophores and the formation and maturation of conidia (21,22). The key activator BrlA is a C 2 H 2 zinc finger transcription factor that governs the initiation of conidiophore development (23,24), followed by sequential activation of AbaA in the middle phase of conidiophore development (23,25,26) and of WetA in the late phase to activate the expression of proteins or enzymes involved in the synthesis of spore wall components (27)(28)(29). Aside from WetA involved in conidial maturation, the velvet protein VosA downstream of the central pathway is required for both the repression of BrlA expression for termination of its control cycle and the biosynthesis of trehalose for conidial maturation (30,31).
The regulatory role of the central developmental pathway in fungal insect pathogens remains poorly understood. In B. bassiana, loss-of-function mutations of wetA and vosA have been shown to reduce conidiation capacity by 98% and 88%, respectively; impair cell walls; and decrease intracellular trehalose accumulation (32). However, the upstream key activators BrlA and AbaA have not been functionally explored due to the lack of conidium production by the deletion mutant of either coding gene and an inability to rescue the deleted gene through conventional Agrobacterium-mediated transformation with conidia (33). Dimorphic fungi, such as phytopathogenic Ustilago maydis and human-pathogenic Candida albicans, usually grow in a yeast form outside the host and require a switch to filamentous growth for tissue penetration during host infection (34)(35)(36). Unlike dimorphic plant and human mycopathogens, filamentous fungal insect pathogens usually grow by hyphal extension outside the host and propagate by yeast-like budding only after entry into the host hemocoel through cuticular penetration. The fungal propagation in the host hemocoel determines the speed of host death from mummification and requires a switch to blastospores from the penetrating hyphae. The opposite direction of the dimorphic switch outside or inside the host suggests a possibly distinct mechanism underlying the dimorphic transition of insect mycopathogens. Since both aerial conidiation and dimorphic transition were closely linked to transcriptional expression of the developmental activator genes (Table 1), we hypothesize that, like aerial conidiation, the submerged dimorphic transition required for fungal virulence could be a process of asexual development to be primarily governed by the central pathway in B. bassiana and other filamentous fungal insect pathogens. This study aims to test the hypothesis. The deletion mutants of brlA and abaA and their complementary strains constructed with the protoplasts from the hyphal cells of the deletion mutants were analyzed in parallel with the parental wild-type strain. Our findings unveil that BrlA and AbaA act as master regulators of not only aerial conidiation but also insect pathogenicity and dimorphic transition in B. bassiana.

RESULTS
Transcriptional profiles and subcellular localization of BrlA and AbaA in B. bassiana. An online search through the B. bassiana genome (37) with the queries of Aspergillus nidulans BrlA and AbaA sequences resulted in identification of orthologous BrlA (EJP63618) and AbaA (EJP70670), which consist of 370 and 901 amino acids with molecular sizes of 41.52 and 100.46 kDa, respectively. The B. bassiana BrlA and AbaA are characterized, respectively, by two C-terminal C 2 H 2 zinc finger (ZnF) motifs and an N-terminal transcriptional enhancer activator (TEA) motif, both of which are shared by the counterparts of 15 other filamentous, yeast, or yeast-like fungi (see Fig. S1 in the supplemental material). The two orthologs are phylogenetically closer to the counterparts of entomopathogenic fungi, including Cordyceps militaris and Metarhizium spp., than of nonentomopathogenic fungi.
Transcriptional profiles of brlA and abaA in the wild-type strain B. bassiana ARSEF 2860 (designated WT here) were monitored during a 7-day incubation of conidia spread on plates of Sabouraud dextrose agar plus yeast extract (SDAY) under the optimal regime of 25°C in a light/dark cycle (L:D) of 12:12 h. Compared to the standard level at the early hyphal extension stage of 24-h incubation, the transcript level of either brlA or abaA was suppressed during hyphal growth (day 2), followed by a sharp elevation to ϳ30-fold on day 5 (Fig. 1A). Notably, the sharp elevation of brlA occurred 24 h earlier (day 3) than that of abaA, which sustained the transcript peak 24 h longer. These data suggest a 24-h-earlier activation of brlA than of abaA in the central developmental pathway of B. bassiana.
Laser scanning confocal microscopic (LSCM) analysis of the green fluorescent protein (GFP)-tagged BrlA and AbaA fusion proteins expressed in the WT strain demonstrated a localization of BrlA (Fig. 1B) or AbaA (Fig. 1C) in both the cytoplasm and nucleus of hyphal cells, which were collected from 3-day-old liquid cultures grown in Sabouraud dextrose broth (SDB; i.e., agar-free SDAY) under the optimal regime and stained with a nucleus-specific dye. The subcellular localization of either fusion protein was consistent irrespective of the hyphal cultures grown under continuous light (L:D 24:0) or dark (L:D 0:24) conditions. The ratios of nuclear to cytoplasmic green fluorescence intensities quantified from the hyphal cells also showed a higher accumulation level of each fusion protein in the nucleus than in the cytoplasm at L:D 0:24, 12:12, or 24:0 (Fig. 1D). The nuclear localization indicated a possibility for either BrlA or AbaA to act as a transcription factor.
BrlA and AbaA are indispensable for conidiation but nonessential for hyphal growth. Either brlA or abaA was deleted from WT via homogeneous recombination of its 5= and 3= coding/flanking fragments separated by the bar marker and rescued by ectopic integration into an identified deletion mutant of the cassette comprising its full-length coding/flanking sequence and the sur marker ( Fig. S2 and Table S1). As a result, fungal colonies initiated by attaching uniform hyphal mass discs (5-mm diameter) to the plates of rich SDAY, a standard medium for cultivation of entomopathogenic fungi, became fluffier in the ΔabaA mutant but showed little morphological change in the ΔbrlA mutant (Fig. S3A). After a 7-day incubation under the optimal regime, colony sizes were not significantly different between ΔbrlA and WT strains (Tukey's honestly significant difference [HSD], P Ͼ 0.05) irrespective of being grown on SDAY, minimal Czapek agar (CZA), and almost all of 38 CZAs amended with different carbon (sugars/polyols) or nitrogen (inorganic/organic) sources (Fig. S3B). In contrast, the ΔabaA mutant grew significantly faster than the control (WT and complementary) strains on most of the minimal media despite moderately decreased or unchanged colony sizes on SDAY and a few CZAs amended with the nitrogen sources NH 4 ϩ , NO 2 Ϫ , tryptophan, and alanine, respectively. These observations indicated a nonessential role for either BrlA or AbaA in vegetative growth of B. bassiana.
Intriguingly, either brlA or abaA deletion abolished aerial conidiation during 12 days of incubation under the optimal regime on the SDAY plates spread with 100-l aliquots of a hyphal cell suspension for culture initiation. The control strains started aerial conidiation on day 3 by formation of clustered zigzag rachises (conidiophores) and conidia ( Fig. 2A), produced plenty of conidia on day 5 (Fig. 2B), and reached a peak yield of ϳ5 ϫ 10 8 conidia cm Ϫ2 plate culture on day 7 or 8 ( Fig. 2C). In contrast, the hyphae of ΔbrlA and ΔabaA mutants were not normally differentiated during the period of optimal incubation, as revealed by scanning electronic microscopic (SEM) analysis of 5-day-old cultures ( Fig. 2B) or microscopic examination of 8-or 12-day-old cultures ( Fig. 2D). Interestingly, cell clusters like conidiating structures were sporadically present in the 12-day-old ΔabaA cultures. However, such cell clusters did not comprise the same zigzag rachises as appeared in the cultures of the control strains; the clustered cells were unable to be scattered for suspension preparation via supersonic vibration and hence were unlikely conidia (Fig. 2E). In addition, biomass levels quantified from the 5-and 8-day-old cultures of the ΔabaA strain were significantly higher (Tukey's HSD, P Ͻ 0.05) than those from the cultures of the control strains but unaffected in the ΔbrlA strain (Fig. 2F). The completely abolished conidiation in the absence of brlA or abaA highlights an indispensability of either BrlA or AbaA for aerial conidiation of B. bassiana.
BrlA and AbaA are indispensable for dimorphic transition in vitro. The submerged cultures of two deletion mutants and their control strains were initiated with a hyphal cell suspension (1 mg fresh biomass ml Ϫ1 ) in rich SDB (i.e., agar-free SDAY), minimal Czapek broth (CZB), and trehalose-peptone broth (TPB) mimicking insect hemolymph, followed by a 5-day incubation on a shaking bed (150 rpm) at 25°C. For each strain, mean biomass level quantified for three independent cultures at the ends of 3-and 5-day incubations was maximal in SDB ( Fig. 3A), minimal in CZB (Fig. 3B), and intermediate in TPB (Fig. 3C). The ΔabaA mutant produced significantly more biomass than the control strains in all submerged cultures except the 3-day-old SDB culture while the ΔbrlA mutant had biomass levels similar to those of the control strains in all cultures except the 5-day-old SDB culture. As an index of dimorphic transition in vitro, mean (ϮSD) yields at the scale of 10 6 blastospores mg Ϫ1 biomass in the SDB, CZB, and TPB cultures of three control strains increased to 2.08 (Ϯ0.18), 9.64 (Ϯ0.63), and 24.20 (Ϯ3.77) on day 5 from 0.84 (Ϯ0.14), 4.32 (Ϯ0.62), and 17.09 (Ϯ2.46) on day 3, respectively. Apparently, their dimorphic transition rates in the TPB cultures were much higher than those in CZB or SDB. However, blastospore formation was completely abolished in all submerged cultures of both ΔbrlA and ΔabaA mutants although their biomass levels were equal to or higher than those of the control strains. In microscopic examination, no blastospore was found in the culture samples of either deletion mutant, contrasting with the presence of abundant blastospores in the cultures of the control strains (Fig. 3D). These data indicate for the first time an absolute indispensability of BrlA or AbaA for the dimorphic transition in vitro of B. bassiana.
BrlA and AbaA are indispensable for pathogenicity and dimorphic transition in vivo. Since the two deletion mutants produced neither aerial conidia nor submerged blastospores for topical application or intrahemocoel injection in standardized bioassays as described previously (6,8), a blastospore-removed hyphal suspension (10 mg fresh biomass ml Ϫ1 ) of each control strain or deletion mutant was topically applied for normal infection through cuticular penetration by immersing three cohorts of ϳ35 Galleria mellonella larvae in 30-ml aliquots of each suspension for 10 s. The infection resulted in a mean corrected mortality of 87% (Ϯ3.7%) for the control strains against the model insect (Fig. 4A). In contrast, no mortality was attributed to the topical application of the ΔbrlA or ΔabaA hyphae. Moreover, 5 l of a hyphal suspension (2 mg fresh biomass ml Ϫ1 ) was injected into the hemocoel of each larva in each of three cohorts for cuticle-bypassing infection. As a result, the injected larvae were mummified much more rapidly than those that were immersed (Fig. 4B). The injection resulted in an overall mean mortality of 97% (Ϯ3.1%) for the control strains, followed by 93% (Ϯ3.9%) attributed to the ΔbrlA mutant and 75% (Ϯ3.6%) attributed to the ΔabaA mutant. Apparently, either ΔbrlA or ΔabaA strain hyphae lost all ability to infect the larvae through the normal route of cuticular penetration, indicating an indispensability of BrlA or AbaA for the insect pathogenicity of B. bassiana. The injected mutant hyphae caused lower mortality or slower death, hinting at a possibility that certain virulencerelated cellular processes could be retarded in the absence of brlA or abaA. (A to C) Biomass levels and blastospore yields quantified from the 3-and 5-day-old submerged cultures of two deletion mutants and their control strains in SDB, CZB, and TPB, respectively. Each culture was initiated with 1 mg fresh hyphae ml Ϫ1 SDB, CZB, or TPB mimicking insect hemolymph. Error bars, SD of the mean from three independent cultures. Different lowercase letters marked on the bars of each group denote a significant difference (Tukey's HSD, P Ͻ 0.05). (D) Microscopic images (bar, 50 m) for the status of blastospore production in the 3-day-old SDB, CZB, and TPB cultures of all tested strains. Note that blastospore production occurred in the control strains but was completely abolished in the ΔbrlA and ΔabaA strains. To gain an insight into the abolished cuticle infection, we assessed total activities of extracellular (proteolytic, chitinolytic, and lipolytic) enzymes and Pr1 proteases secreted from the 60-h-old liquid cultures grown in CZB containing 0.3% bovine serum albumin (BSA) as the sole nitrogen source for enzyme induction and initiated with fresh hyphal cells at the fixed rate of 1 mg ml Ϫ1 . Such enzymes are likely involved in cuticle degradation essential for host infection (2). As illustrated in Fig. 4C, the activities of extracellular enzymes and Pr1 proteases quantified from the supernatants of the cultures were reduced by 75% and 78% in the ΔbrlA mutant and 96% and 99% in the ΔabaA mutant, respectively, in comparison to the activities in WT. However, the mean biomass level in the CZB-BSA cultures was unaffected in the ΔbrlA mutant and even enhanced by 47% in the ΔabaA mutant relative to WT. All of these changes were restored by targeted gene complementation. The results imply that the abolition of the cuticle infection could be due to blocked secretion and/or synthesis of cuticledegrading enzymes in the absence of brlA or abaA.
To explore a possible cause for the delayed lethal action of the injected mutant hyphae, we examined the status of dimorphic transition in vivo through microscopic examination of the hemolymph samples taken from the larvae surviving after topical application or intrahemocoel injection as reported previously (4, 6). On day 8 after topical application, the control strains formed abundant hyphal bodies in host hemolymph whereas either the ΔbrlA or ΔabaA mutant formed no hyphal body in the host hemolymph comprising many more intact host hemocytes (upper panels in Fig. 4D). On day 3 after injection, the hyphae of the control strains also formed abundant hyphal bodies in the host hemolymph, but discrete hyphal bodies were rarely found in the host hemolymph injected with the mutant hyphae, which seemingly grew in vivo by direct extension (lower panels in Fig. 4D). Intriguingly, the injected ΔabaA strain hyphae became thicker in the host hemolymph and formed many short branches nearly perpendicular to the stem, implying slower extension of the branched ΔabaA strain hyphae than of the unbranched ΔbrlA strain hyphae in the host hemocoel. In addition, the control strains formed a heavy layer of outgrowth on the surfaces of all cadavers 5 days after death from the injection (Fig. 4E). In contrast, the fungal outgrowth was thin for the ΔbrlA mutant and very sparse for the ΔabaA mutant, leaving some or most of the cadaver surfaces baldly exposed. These observations demonstrated an inability of the two deletion mutants to form hyphal bodies in vivo and hence an indispensability of either BrlA or AbaA for dimorphic transition in the host hemocoel. The slower host death from the injected mutant hyphae was due to mummification by hyphal extension or branching rather than yeast-like budding, which accelerates fungal proliferation in the host hemocoel and hence host mummification to death (4,6).
Regulatory roles of BrlA and AbaA in global gene expression. The roles of BrlA and AbaA acting as transcription factors in the central pathway of B. bassiana were examined by constructing and analyzing the transcriptomes of ΔbrlA, ΔabaA, and WT strains based on the cDNAs derived from two 84-h-old SDAY cultures (replicates), which were grown under the optimal regime. Conidiation was rapidly developing in the WT cultures prepared. The replicated transcriptomes showed a high correlation (R 2 ϭ 0.993 for WT, 0.971 for ΔbrlA strain, and 0.986 for ΔabaA strain), indicating a desired repeatability. The resultant transcriptomes comprised almost all annotated genes (10,363) and hundreds of genes not annotated in the fungal genome (37).
Compared to the WT strain, the ΔbrlA mutant had 707 and 806 genes significantly upregulated (log 2 ratio, 1.00 to 6.62) and downregulated (log 2 ratio, Ϫ8.92 to Ϫ1.00), respectively ( Fig. 5B; see also Table S2). Significantly upregulated (log 2 ratio, 1.00 to 7.56) and downregulated (log 2 ratio, Ϫ13.80 to Ϫ1.00) genes increased to 1,513 and 1,356 in the ΔabaA strain, respectively ( Fig. 5C; also Table S3). These differentially expressed genes (DEGs) took up 14.6% of the whole genome in the ΔbrlA strain while the proportion surprisingly increased to 27.7% in the ΔabaA strain. Many more genes differentially expressed in the ΔabaA than in the ΔbrlA mutant could be likely due to the increasing role of AbaA at the rapidly developing stage of conidiation, since brlA was transcriptionally activated earlier than abaA (Fig. 1A) for initiation of conidiation ( Fig. 3A to C). Intriguingly, the top 10% of the repressed genes (log 2 ratio, ՅϪ3.2) in the ΔbrlA (150) and ΔabaA (350) strains contained the same proportion (40%) encoding hypothetical proteins.
Revealed by gene ontology (GO) analysis, 562 ΔbrlA DEGs (255 up-and 307 downregulated) were enriched to five GO terms ( Fig. 5D; see also Table S4). Three of the GO terms were involved in molecular function (oxidoreductase activity, acting on paired donors, with incorporation or reduction of molecular oxygen; flavin adenine dinucleotide binding; and oxidoreductase activity), and two were involved in biological processes (oxidation-reduction process and transmembrane transport), but no term fell into the cellular component. Since many genes may function in multiple cellular processes and events, 7,343 ΔabaA DEGs (4,367 up-and 2,976 downregulated) were significantly enriched into 29 GO terms covering a much wider array of molecular functions (16) and biological processes (12) as well as cellular components (1), as shown in Fig. 5E and Table S5. These analyses hint at a more critical role of AbaA than of BrlA in transcriptional regulation of downstream genes for rapid development of the fungal conidiation activated earlier by BrlA. In addition, 62 and 343 DEGs of ΔbrlA and ΔabaA strains (Tables S6 and S7) were enriched into 8 and 12 KEGG pathways, respectively, at the level of P Ͻ 0.05. However, the corrected P values (0.27 to 0.52) indicated an insignificance for all KEGG enrichments except that for the degradation of aromatic compounds in the ΔabaA strain (corrected P ϭ 0.039).
BrlA and AbaA regulate expression of multiple genes essential for cuticle infection but nonessential for the dimorphic switch. The transcriptomes were further analyzed for an insight into why the hyphae of either the ΔbrlA or ΔabaA mutant lost all ability to infect the insect through cuticle penetration. Nine and 15 genes likely involved in host adhesion and cuticle degradation were sharply repressed in ΔbrlA and ΔabaA strains (Fig. 5F), respectively. Among those, six genes were simultaneously repressed in both deletion mutants, including the class I hydrophobin gene (BBA_03015), critical for cell hydrophobicity and pathogenicity (38); the filamentous hemagglutinin/adhesin gene (BBA_03909), required for host adhesion and infection (39); and three genes encoding thermostable alkaline protease (peptidase S8/S53, subtilisin/kexin/sedolisin; BBA_08901) acting as a virulence factor (40), a subtilase-like protein (BBA_01003), and a chitinase-like protein (BBA_09605) involved in cuticle degradation. Three other genes repressed in the ΔbrlA strain encode two proteases (BBA_07321 and BBA_04644) and one chitinase-like protein (BBA_06317). Nine other genes repressed in the ΔabaA strain encode another hydrophobin-like protein (BBA_06599), one adhesin (BBA_02379), three proteases or subtilase-like proteins (BBA_09270, BBA_05303, and BBA_00443), and four chitinases or chitinase-like proteins (BBA_06317, BBA_02381, BBA_06297, and BBA_09307). The activities of such proteins or enzymes are considered to be very important for the success of fungal infection through cuticle penetration (2). Transcriptional repressions of these putative cuticle-degrading enzyme genes in ΔbrlA and ΔabaA strains correlated well with markedly reduced activities of extracellular enzymes and Pr1 proteases, suggesting essential roles of both BrlA and AbaA in their transcription regulation and hence in the normal infection of B. bassiana through cuticular penetration.
However, almost all of the homologous genes that are considered to control the dimorphic switch in plant and human mycopathogens (18) were not up-or downregulated in the ΔbrlA or ΔabaA mutant at the significant levels of log 2 ratios of Ն1 or ՅϪ1 and of adjusted P of Ͻ0.05 (Table 2). Instead, two other central pathway genes and downstream vosA were more consistently repressed than those putatively acting in the dimorphic switch of the ΔbrlA or ΔabaA strain, such as drkA, cdc42, pakA/B, racA, and ras1/2, which presumably function in two-component, G protein, Ras, and cAMP signaling systems. Neither were almost all of the genes encoding MAPK-cascaded components differentially expressed at significant levels. These data implied that BrlA and AbaA served as master regulators of both aerial conidiation and submerged blastospore production (dimorphic transition) while the signaling systems other than the central pathway had little role in shutting down the dimorphic switch of the ΔbrlA or ΔabaA strain. In other words, it is either BrlA or AbaA that governs the dimorphic switch in vitro and in vivo in B. bassiana.

DISCUSSION
BrlA and AbaA are evidently localized in both the cytoplasm and nucleus and sequentially activated at the transcription level in B. bassiana like their orthologs in the aspergilli (21,26). Our results demonstrate essential roles of both developmental activators in the asexual cycle in vitro and in vivo of a filamentous fungal insect pathogen, as discussed below.
In B. bassiana, submerged blastospore production in vitro, particularly in the TPB mimic of insect hemolymph, often correlates with dimorphic transition in vivo, virulence, and activated genes in the central developmental pathway. For instance, transcriptional repression of brlA or abaA by Ն90% (Table 1) resulted in severe or extremely severe defects in blastospore production, blocked or inhibited dimorphic transition in vivo, and greatly attenuated virulence or abolished host infection in the absence of gcn5, rpd3, vvd, or vlp4 (6,8,11,14). In this study, hyphal pathogenicity to insect and blastospore production in vitro and in vivo was completely abolished due to the deletion of brlA or abaA. Previously, normal cuticle infection was also abolished in Δfus3, Δste7, and Δste11 mutants of B. bassiana due to reduced activities of cuticledegrading enzymes largely attributed to their severe growth defects on oligotrophic substrata (12,41,42). However, the three genes in the Fus3 cascade were not differentially expressed in the transcriptomes of ΔbrlA and ΔabaA mutants. Neither was hyphal growth of either mutant retarded in various media. The inability for the hyphae of the ΔbrlA or ΔabaA strain to infect the insect through cuticular penetration could be attributable to the markedly reduced transcripts of multiple genes critical for host adhesion and cuticle degradation. This inference is supported by sharply reduced activities of extracellular enzymes and Pr1 proteases in both deletion mutants. Thus, the present and previous studies indicate an indispensability of either BrlA or AbaA for insect pathogenicity of B. bassiana and also unveil a novel molecular mechanism distinct from the control of the dimorphic switch by the signaling systems other than the central pathway in plant and human mycopathogens (18). Previously, only AbaA was revealed to be involved in the dimorphic switch of Talaromyces (formerly Penicillium) marneffei (43). We consider that the distinct mechanisms underlying the dimorphic switch of these fungal pathogens may stem from distinct host-pathogen interactions and their peculiar lifestyles. The dimorphic switch of an insect mycopathogen occurs only after the success of hyphal infection through cuticular penetration and is required for acceleration of intrahemocoel fungal propagation by yeast-like budding and of host mummification to death. In contrast, dimorphic plant or human mycopathogens require a switch from the usual yeast growth to filamentous growth for initiation of host infection by penetration through host tissues (34,36). Our inference is supported by abolished dimorphic transition in vitro and in vivo and almost all signaling genes not being differentially expressed in either the ΔbrlA or the ΔabaA strain but being considered to control the dimorphic switch in plant and human mycopathogens (18). Thus, both BrlA and AbaA serve as master regulators of the dimorphic switch in B. bassiana. Moreover, the block of the central development pathway by the brlA or abaA deletion also led to abolishment of aerial conidiation but had no negative impact on hyphal growth in various agar or broth media. During a prolonged 12-day period of optimal incubation for conidiation on the standard medium, no sign of hyphal differentiation was observed in the ΔbrlA mutant while the ΔabaA mutant displayed sporadic cell clusters at the end of incubation. However, the clustered cells failed to form short rachises for support of conidial production and hence were not the normal spore balls that comprise many zigzag rachises and conidia and can be readily scattered for a uniform suspension. We speculate that the strange cell clusters could likely result from abnormal hyphal differentiation and hence are similar to aberrant and nonfunctional phialides observed in Penicillium digitatum ΔabaA (44). Thus, the abolished conidiation of ΔbrlA and ΔabaA mutants in this study is consistent with previous observations from the same mutants of other filamentous fungi (26,43,44). This reinforces highly conserved roles of both BrlA and AbaA in the regulation of asexual development and life cycle in filamentous fungi.
Finally, our transcriptomic analyses revealed that the deletion of brlA or abaA resulted in similar repression of the downstream gene wetA, which has been shown to regulate conidial maturation and cell wall integrity in the central pathway of B. bassiana (32). In the previous study, knockout mutation of wetA led to 98% reduction in conidial yield and much more attenuated virulence via intrahemocoel injection than via normal cuticle infection while the fungal virulence through either mode of infection was slightly or insignificantly affected when the downstream vosA lost function. This hints at a possible role of WetA instead of VosA in the dimorphic switch of B. bassiana required for yeast-like budding propagation to accelerate host mummification. However, it is impossible for the limited repression level of wetA to completely shut down the dimorphic switch in vitro and in vivo in ΔbrlA and ΔabaA strains. Like aerial conidiation, therefore, submerged dimorphic transition in vitro and in vivo is a process of asexual development that is governed by BrlA and AbaA in the central pathway of B. bassiana.

MATERIALS AND METHODS
Bioinformatic analysis of BrlA and AbaA in B. bassiana. The A. nidulans BrlA (EAA66002) and AbaA (EAA35779) sequences were used as queries to search through the B. bassiana genome (37) under NCBI accession no. NZ_ADAH00000000 via online BLASTp analysis at https://blast.ncbi.nlm.nih.gov/Blast.cgi. The sequences of BrlA and AbaA orthologs found in the fungal genome were aligned for structural comparison with the queries and counterparts of other filamentous, yeast, or yeast-like fungi using the SMART program at http://smart.embl-heidelberg.de/, followed by phylogenetic analysis with a neighborjoining method in MEGA7 software at http://www.megasoftware.net/.
Transcriptional profiling of brlA and abaA. The WT strain was incubated for 7 days under the optimal regime of 25°C and L:D 12:12 on cellophane-overlaid SDAY (4% glucose, 1% peptone, and 1.5% agar plus 1% yeast extract) plates, which were spread with 100-l aliquots of a 10 7 -conidia ml Ϫ1 suspension for culture initiation. From the end of the 24-h incubation onward, total RNAs were separately extracted daily from three plate cultures using an RNAiso Plus reagent kit (TaKaRa, Dalian, China) and reverse transcribed into cDNAs using a PrimeScript reverse transcription (RT) reagent kit (TaKaRa). Each of the cDNA samples (standardized by dilution) was used as a template to assess transcript levels of brlA and abaA via real-time quantitative PCR (qPCR) with paired primers (see Table S1 in the supplemental material) under the action of SYBR Premix Ex Taq (TaKaRa). The fungal 18S rRNA was used as an internal standard. The threshold cycle (2 ϪΔΔCT ) method (45) was used to compute the relative transcript level of brlA or abaA in the WT strain on a given day with respect to the standard level at the end of the 24-h incubation.
Subcellular localization of BrlA and AbaA. Transgenic strains strongly expressing BrlA::GFP and AbaA::GFP fusion proteins in the WT strain, respectively, were created as described previously (6). Each transgenic strain was incubated for full conidiation on SDAY. The resultant conidia were suspended in SDB and incubated on a shaking (150-rpm) bed for 3 days at 25°C in the L:D cycles of 24:0, 12:12, and 0:24. Hyphal cells taken from the cultures of each strain in each L:D cycle were stained with the nucleus-specific dye DAPI (4=,6=-diamidine-2=-phenylindole dihydrochloride; Sigma) and visualized. LSCM images for BrlA::GFP and AbaA::GFP fusion proteins expressed in the stained cells were merged with an image browser to judge subcellular localization of each target protein in response to the L:D cycles. To verify a nuclear localization of each fusion protein, green fluorescence intensities were measured from the cytoplasm and nuclei of at least 10 hyphal cells (one nucleus per cell) using ImageJ software at https://imagej.nih.gov/ij/. Relative accumulation levels of each fusion protein in the nuclei of the hyphal cells incubated in the L:D cycles of 0:24, 12:12, and 24:0 were calculated as the ratios of nuclear to cytoplasmic fluorescence intensities.
Generation of brlA and abaA mutants. The genes brlA and abaA (tag loci: BBA_07544 and BBA_00300, respectively) were deleted from the WT strain as described previously for the deletion of wetA or vosA (32). Briefly, 5= and 3= partial coding/flanking sequences of each gene were amplified from the genomic DNA of the WT strain using paired primers (Table S1) and inserted into linearized p0380-bar at appropriate enzyme sites. The resultant plasmids p0380-5=x-bar -3=x (x ϭ brlA or abaA) were integrated into the WT via Agrobacterium-mediated transformation for targeted gene deletion. For targeted gene complementation, the full-length coding sequence of brlA or abaA with flanking regions was cloned from the WT DNA with paired primers, digested with the appropriate restriction enzyme, and inserted into linearized p0380-sur-exchange. The resultant p0380-sur-x was ectopically integrated into the protoplasts of each deletion mutant via polyethylene glycol-mediated transformation (46). The used protoplasts were released from the hyphal cells collected from the 3-day-old SDB culture of each deletion mutant by suspending 100-mg aliquots of hyphal cells (fresh weight) in 2 ml of 1.2 M sorbitol containing 1% snailase and 1% lysing enzyme (Sigma) for 5-h cell wall lysing at 37°C. The released protoplasts were collected by filtration through lens-cleaning tissues, rinsed repeatedly with 1.2 M sorbitol, and used as recipients for targeted gene complementation. Putative deletion or complementary mutant colonies grown on a selective medium were screened by bar resistance to phosphinothricin (200 g ml Ϫ1 ) or sur resistance to chlorimuron ethyl (10 g ml Ϫ1 ) and sequentially identified through PCR and Southern blot analyses with paired primers and amplified probes (Table S1). The positive brlA and abaA mutants with expected recombinant events verified (Fig. S2) were evaluated in parallel with the WT strain in the following experiments comprising three independent cultures or samples taken from the cultures. Assessments of growth rates on different media. Hyphal mass plugs (5-mm diameter) were bored from the culture of each strain grown on cellophane-overlaid SDAY (CO-SDAY) plates for 3 days at 25°C and attached centrally to the plates of SDAY, CZA (3% sucrose, 0.3% NaNO 3 , 0.1% K 2 HPO 4 , 0.05% KCl, 0.05% MgSO 4 , and 0.001% FeSO 4 plus 1.5% agar), and amended CZAs containing different carbon or nitrogen sources. The amended CZAs were prepared by deleting 3% sucrose, 0.3% NaNO 3 , or both from CZA; replacing the sole carbon source with 3% glucose, trehalose, lactose, fructose, maltose, mannitol, sorbitol, glycerol, ethanol, sodium acetate (NaAc), oleic acid, or olive oil; and replacing the sole nitrogen source with 0.3% NH 4 Cl, NH 4 NO 3 , NaNO 2 , or one of 20 amino acids, respectively. After a 7-day incubation at 25°C and 12:12 h, all colony diameters were measured as indices of radial growth rates using two measurements taken perpendicular to each other across the center.
Assessment of conidiation capacity. Three 100-l aliquots of a hyphal suspension (fresh hyphal mass, 1 mg ml Ϫ1 ) per strain were evenly spread on SDAY plates and incubated for 12 days under the optimal regime of 25°C and L:D 12:12. From the end of the 3-day incubation onward, three plugs (5-mm diameter) were bored daily from each plate culture using a cork borer. Each plug was placed in 1 ml of 0.02% Tween 80 for the release of its conidia by supersonic vibration. Three samples taken from each of the resultant suspensions were used for assessment of conidial concentration with a hemocytometer, followed by converting the concentration to the number of conidia produced per unit area (cm 2 ) of plate culture. During the period of incubation, the conidiation state of fungal mass samples taken from the cultures of each strain was stained with calcofluor white (a dye specific to cell wall) and observed under a microscope or directly examined via SEM as described elsewhere (6).
Assessment of dimorphic transition in vitro. Hyphal cells collected from the 3-day-old SDB cultures of each strain were rinsed twice with sterile water and resuspended in 50-ml aliquots of fresh SDB, CZB (i.e., agar-free CZA), and TPB, which was a modified CZB containing the sole carbon source of 3% trehalose and the sole nitrogen source of 0.5% peptone and which mimicked insect hemolymph. Possible blastospores in each culture were removed by filtration through lens-cleaning tissues. All of the aliquots in flasks were standardized to a final concentration of fresh hyphal mass of 1 mg ml Ϫ1 and incubated for 5 days with shaking (150 rpm) under the optimal regime. At the end of 3-and 5-day incubations, three 50-l samples were taken from each of three flasks per strain in each broth. The blastospore concentration (count ml Ϫ1 culture) was assessed from each sample using a hemocytometer. The remaining culture of each flask was dried by pumping in a vacuum, followed by estimation of biomass level (mg ml Ϫ1 ). The two quantities were used to compute the absolute blastospore yield (number of blastospores mg Ϫ1 biomass) as an index of dimorphic transition in vitro in each submerged culture.
Assays for hyphal pathogenicity and activities of extracellular enzymes involved in cuticle degradation. Since ΔbrlA and ΔabaA strains produced neither conidia nor blastospores, the 3-day-old SDB cultures of all deletion mutants and control strains prepared as described above were filtered through lens-cleaning tissues for removal of possible blastospores. The collected hyphae of each strain were suspended in 0.02% Tween 80 and standardized to a concentration of fresh hyphae of 2 or 10 mg ml Ϫ1 . For topical application, three cohorts (replicates) of ϳ35 G. mellonella larvae were separately immersed for 10 s in 30-ml aliquots of each hyphal suspension (10 mg ml Ϫ1 ) for normal infection through cuticular penetration. Alternatively, 5 l of each hyphal suspension (2 mg ml Ϫ1 ) was injected into the hemocoel of each larva in each cohort. The same volume of 0.02% Tween 80 was used as a control in each bioassay. All treated cohorts were maintained at 25°C and 12:12 h and monitored every 12 or 24 h for survival/mortality records until the records no longer changed for two consecutive days. The time-mortality trends of the larvae treated with each strain were corrected using the mortality records in the control. On day 8 after immersion or day 3 after injection, hemolymph samples were taken from surviving larvae and examined for the presence/absence of hyphal bodies under a microscope as described elsewhere (4,6,8). Images for fungal outgrowths on the surfaces of cadavers were taken after 5-day maintenance under optimal conditions. Total activities of extracellular enzymes and Pr1 proteases involved in fungal infection through cuticular penetration were quantified from the liquid cultures of each strain as described previously (4,6,8). Briefly, 50-ml aliquots of a hyphal suspension (fresh hyphae, 1 mg ml Ϫ1 ) in CZB containing 0.3% BSA as the sole nitrogen source for induction of enzyme production were incubated on a shaking bed (150 rpm) for 60 h at 25°C. The cultures were centrifuged at 4°C. Hyphal biomass in each culture was quantified after vacuum drying. Total activities of extracellular enzymes and Pr1 proteases in each supernatant were assessed by reading optical densities at 442 and 410 nm (OD 442 and OD 410 ), respectively. One unit of enzyme activity was defined as an enzyme amount required for an OD 442 or OD 410 increase by 0.01 after a 1-h reaction of each extract relative to a blank control. Total activities were expressed as U ml Ϫ1 supernatant.
Analyses of brlA-and abaA-specific transcriptomes. Two 84-h-old cultures (replicates) of ΔbrlA, ΔabaA, and WT strains grown on cellophane-overlaid SDAY plates under the optimal regime were sent to Novogene (Beijing, China) for construction and analysis of transcriptomes. Total RNAs were separately extracted from the two cultures of each strain. The mRNAs were isolated from total RNAs using magnetic oligo(dT) beads and fragmented into segments. The resultant mRNA fragments were used as the templates to synthesize the first-strand cDNAs with random hexamer primers. The second-strand cDNAs were synthesized using the first-strand cDNAs as the templates. Each of the double-strand cDNAs was purified and end repaired, and single adenines were added to the ends of the cDNA molecules. Finally, the cDNA library was constructed by adding proper adaptors to the cDNA for sequencing on an Illumina HiSeq platform.
Clean tags were generated by filtering all raw reads from the cDNA sequencing and then mapped to the B. bassiana genome (37) based on log 2 ratio (ΔbrlA/WT or ΔabaA/WT strain) of ՅϪ1 (downregulated) or Ն1 (upregulated) at the significance of corrected P Ͻ 0.05. All data were normalized as fragments per kilobase of exon per million fragments mapped (FPKM). All identified DEGs were functionally annotated with known or putative gene information in the nonredundant NCBI protein databases and subjected to GO analysis (http://www.geneontology.org/) for recognition of GO terms falling into biological process, molecular function, and cellular component. Further, Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis (http://www.genome.jp/kegg/) was performed to enrich the DEGs into various KEGG pathways at the significance level of P or corrected P of Ͻ0.05.
Data availability. All data generated or analyzed during this study are included in the published paper and associated supplemental files. All transcriptomic data aside from those reported in supplemental files (Tables S2 to S7) of this paper are available at the NCBI's Gene Expression Omnibus under the accession no. GSE132277.