Adaptive responses to environmental stimuli are integral to the survival and virulence of microbial pathogens. The thermally dimorphic human fungal pathogen Histoplasma senses temperature to transition between a mold form in soil and a pathogenic yeast in mammalian hosts. The contributions of chromatin-modifying enzymes to the ability of Histoplasma to appropriately respond to temperature have never been explored. Through chemical inhibition and genetics, we determined that the class I histone deacetylase (HDAC) RPD3 is required for normal Histoplasma yeast morphology at 37 °C. Rpd3 regulated the expression of key morphology-specific genes, including critical virulence factors and transcription factors (TFs), was required for normal DNA-binding activity of yeast-promoting TFs, and influenced histone acetylation levels at the loci of putative pro-filamentation TFs. Furthermore, Rpd3 was required for virulence in a macrophage model of infection. Taken together, Rpd3 is a critical regulatory component that both activates the pathogenesis program and represses the filamentation program to enable thermal dimorphism in Histoplasma. This work uncovers the crucial role that chromatin regulation plays in temperature response of this ubiquitous pathogen.

Ethics statement

All animal work was approved under UCSF Institutional Animal Care and Use Committee protocol # AN181753-02B.

Histoplasma strains and culture conditions

All experiments and strain manipulations were conducted with Histoplasma ohiense G217B unless explicitly indicated. Strains used in this study are listed in S1 Table. For studies involving G217B and different geographical Histoplasma isolates, we used G217B (ATCC 26032), G184AR (ATCC 26027), and G186AR (ATCC 26029) strains generously provided by the laboratory of William Goldman at the University of North Carolina, Chapel Hill and sourced HcH88 (ATCC 32281) from the American Type Culture Collection (ATCC). For TF mutant studies, we used strains generated in Joehnk and colleagues [53] and Assa and colleagues [19] derived from the parental G217Bura5^- WU15 strain that was also provided by the Goldman laboratory. Growth of liquid Histoplasma cultures under host conditions was carried out at 37 °C and 5% CO₂ with shaking in Histoplasma Macrophage Media (HMM) supplemented with Penicillin/Streptomycin (P/S). RT growth conditions were carried out at ambient temperature between 22 and 24 °C with shaking in HMM containing an equimolar amount of GlcNAc (A3286, Sigma Aldrich) instead of glucose as previously described (Gilmore and colleagues [13]). All HDACi treatment experiments were conducted at 37 °C in HMM GlcNAc media. For HDACi growth assays, SAHA (LC laboratories V-8477), Entinostat (MedChem Express HY-12163), Mocetinostat (Selleck Chemicals S1122), Ricolinostat (TargetMol T2489), MC1568 (Cayman Chemical Company 16265), and TMP159 (Selleck Chemicals S8502) were ordered as 10 mM solutions in DMSO or as powders that were resuspended to a concentration of 10 mM in DMSO (Sigma-Aldrich D8418) before being filter sterilized using a 0.22 µM MCE filter (Celltreat 229751) and diluted further in sterile DMSO for growth assays.

Culture conditions for HDACi assays

Experiments where cultures are supplemented with an additive (DMSO or HDACi) followed the general scheme outlined in Fig 1B and were conducted as described previously [19]. Cultures used for growth assays, RNA-seq, and ChIP-seq studies were set up in triplicate biological replicates. When possible, inoculum from a single colony grown at 37 °C was used to start a liquid culture in HMM media. This liquid culture was passaged every 2–3 days in HMM media for a total of 2 passages before being used to inoculate a master culture in HMM + GlcNAc at an OD₆₀₀ of 0.2 that was then aliquoted into replicates with equal volumes of each HDACi concentration being tested or DMSO. Cultures were incubated for 2 days at 37 °C and 5% CO₂ or at RT with shaking. Microscopy samples were harvested and fixed for 30 min with a solution of 4% paraformaldehyde (PFA) every 24 hours for up to 3 days following incubation and stored long-term at 4 °C in PBS. Samples for genomic DNA isolation were harvested through centrifugation. Cell pellets used for isolating RNA, and protein samples were harvested through filtration using SFCA Membrane Nalgene Rapid-Flow Sterile Disposable Bottle Top Filters (Fisher 09-740-22G). 10–25 mL of culture was harvested from each flask and immediately flash frozen in liquid nitrogen before being stored at −80 °C.

Cloning and mutant generation

Mutant and OE strains in this study were generated in Histoplasma ohiense G217B (ATCC 26032). Plasmids used in this study are listed in S2 Table. Constructs were generated in E. coli DH5⍺ (New England Biolabs (NEB) C2988J) and OneShot TOP10 (Fisher Scientific C404003) cells. CRISPR/Cas9 plasmids for targeted KO were generated as previously described [53]. Protospacer sequences flanking the RPD3 and HOS2 loci were selected using a local version of CHOPCHOP [81]. Partial sgRNA cassettes were ordered as gBlocks (S4 Table) from Integrated DNA Technologies (IDT) and pairs of 5′ and 3′ targeting sgRNA cassettes were assembled into pNA16, a pDONR/Zeo entry clone containing an sgRNA insertion site in between the Histoplasma glyceraldehyde-3-phosphate dehydrogenase (GAPDH) promoter and A. nidulans tryptophan synthase terminator (trpC) using Gibson Assembly (NEB E2611S). We used Gateway LR clonase (Thermo Fisher Scientific 11791100) to introduce the fully sgRNA expression cassette assembled in the donor plasmid into pBJ262, a hygromycin (hph) marked Cas9 episomal expression plasmid used for transformation in Histoplasma. Final plasmids were sequence verified for correct assembly and linearized using PacI (NEB R0547L) along with a non-targeting control plasmid before being electroporated into yeast as described previously [53].

Strains were screened on solid HMM media supplemented with 200 µg/mL of Hygromycin B (Life Technologies/Gibco Laboratories 10687-010) to select for positive transformants. Transformants were initially screened via colony PCR (cPCR) using primers (see S3 Table) that fall outside of the sgRNA cut sites. Promising KO isolates were passaged on solid media with selection and subjected to additional cPCR screening using primers internal to one of the sgRNA cut sites. In some cases, single-cell resuspensions of transformants were plated for additional cPCR screening. Transformants displaying a clean KO via flanking or internal cPCR were then grown in liquid HMM and passaged 3–4 times in the absence of hygromycin to facilitate plasmid loss. Following plasmid loss, these strains were then subjected to genomic DNA extraction and whole genome sequencing to validate successful KO (outlined below).

RPD3 overexpression plasmids were generated through amplifying the RPD3 open reading frame (ORF) from genomic DNA using Q5 Polymerase (NEB M0494S) and primers (see S3 Table) overlapping with a pDONR overexpression backbone amplified from pDI71. The primers used mediate insertion of the RPD3 ORF in between the ACT1 promoter and CATB terminator sequences derived from pDI71 via NEB HiFi Assembly cloning (NEB E5520S). To generate an OE plasmid containing the putative catalytic dead variant of RPD3, site-directed mutagenesis (SDM) on the WT entry plasmid (pNA61) was carried out using the Agilent Quickchange Lightning SDM Kit (Agilent Technologies 210518). SDM primers used were informed by the residue mapping outlined in S4A Fig. The resulting RPD3 OE pDONR plasmids were cloned into the hph-marked pSB238 episomal plasmid and validated via whole plasmid sequencing prior to transformation and selection on HMM + hyg solid media. Whole plasmid sequencing for cloning was performed by Plasmidsaurus using Oxford Nanopore Technology with custom analysis and annotation.

Histoplasma imaging

PFA-fixed samples were diluted in PBS and mounted on a glass-bottom 96-well imaging plate with high-performance #1.5 cover glass (Cellvis P96-1.5H-N). Micrograph images were taken using a 40× DIC objective lens on a Zeiss Axiovert 200m microscope.

RNA extraction

Total RNA was extracted from fungal cells as previously described [18,19] with Qiazol (Qiagen 79306). Frozen cell pellets were resuspended in equal volumes of Qiazol and subjugated to bead beating with 0.5 µM zirconia beads (Spectrum Chemical 140-19990-E1) and extraction using chloroform. The resulting aqueous phase was transferred to Epoch SpinTM RNA columns (Epoch Life Sciences 1940-250) to isolate RNA and subjected to sequential washes with 3 M NaOAc (pH = 5.5) and 10 mM TrisCl (pH = 7.5) in 80% ethanol. The columns were then treated with Purelink DNAse (Invitrogen 12185010) and subjected to additional washes with 3 M NaOAc (pH = 5.5) and 10 mM TrisCl (pH = 7.5) in 80% ethanol before being eluted with RNAse-free water. The organic phases of each sample were reserved at −20 °C for total protein extraction.

mRNA isolation

Equal amounts of total RNA (between 5 and 10 ng) across all samples were used for mRNA isolation. These samples were subjected to a second round of DNAse treatment (Turbo DNAse, Fisher Scientific AM1907) and their quality was assessed with an RNA 6000 Nano Bioanalyzer Kit (Agilent Technologies 5067-1511). Three rounds of poly-A selection using oligo-dT Dynabeads (Invitrogen 61005) was used to purify mRNA, and mRNA enrichment was verified with an RNA 6000 Nano Bioanalyzer chip.

RNA-seq library preparation

Poly-A-selected RNA was used to generate libraries with the NEB Next Ultra II Directional RNA Library Prep Kit for Illumina (NEB E7760S). Each sample was dual-indexed with unique primers (UCSF Center For Advanced Technology (CAT)) compatible for multiplexed sequencing. Library quality and size was assessed with the Agilent High Sensitivity DNA Bioanalyzer Kit (Agilent Technologies 5067-4626) to ensure proper fragment size (~300–400 bp) and the absence of contaminating adapters. Final library concentrations were measured using a Qubit HS dsDNA assay kit (Life Technologies Q33230) and equal amounts were pooled into a final library for sequencing. Pooled libraries were used for paired-end sequencing using one NextSeq 2000 lane at the Chan-Zuckerberg Biohub (Fig 2) or single-end sequencing using one NovaSeq X lane at the UCSF CAT (Fig 4).

RNA-seq analysis

Transcript abundances were quantified based on version ucsf_hc.01_1.G217B of the Histoplasma G217B transcriptome (S5 Data of Gilmore and colleagues [18]). Relative abundances (reported as TPM values [82] and estimated counts (est_counts) of each transcript in each sample were estimated by alignment-free comparison of k-mers between the reads and mRNA sequences using KALLISTO version 0.46.2 [83]. Further analysis was restricted to transcripts with estimated counts ≥10 in at least four samples (for the experiment in Fig 2) or at least nine samples (for the experiment in Fig 4). Differentially expressed genes were identified by comparing replicate means for contrasts of interest using LIMMA version 3.46.0 [84,85]. Genes were considered significantly differentially expressed if they were statistically significant (at 5% FDR) with an effect size of at least 2× (absolute log₂ fold change ≥1) for a given contrast.

Protein extraction

Organic phases reserved at −20 °C were washed with 100% ethanol and spun down to pellet DNA and isolate the protein-containing phenol-ethanol supernatant. Proteins were precipitated from this suspension using isopropanol and subjected to washes with 0.3 M guanidinium thiocyanate in 95% ethanol before being resuspended in 100% ethanol. Protein pellets were then spun down and dried at RT before being dissolved in freshly-made urea lysis buffer (9 M Urea, 25 mM Tris-HCl, 1% SDS, 0.7 M β-mercaptoethanol) and boiled for 5 min. Protein concentrations were measured using a Pierce 660 nm Assay Kit (Thermo Scientific 22660) supplemented with the ionic detergent compatibility reagent (Fisher Scientific PI22663).

Western blotting

Total protein extracts were obtained from pellets used to conduct the ChIP-seq and RNA-seq studies outlined in Figs 4 and 5. 10 µg of total protein resuspended in Novex 554 NuPAGE LDS Sample Buffer (Invitrogen NP0007) was boiled for 5 min, loaded on NuPAGE Novex 4%–12% Bis-Tris Gels (Life Technologies Corporation NP0322BOX) and subjected to electrophoresis with MOPS buffer at 140 V. Proteins were subjected to semi-dry transfer on nitrocellulose membranes using the Invitrogen iBlot2 Gel transfer device. Membranes were incubated with Intercept (PBS) Blocking Buffer (LI-COR) for one hour before being incubated in custom primary polyclonal antibodies raised against Ryp1 (ID: 3873), Ryp2 (ID: 387), and Ryp3 (ID: 356) [15–17] or a primary antibody against tubulin (Novus Biologicals DM1A) at 4 °C with shaking overnight. Following incubation, membranes were washed three times with a solution of 0.1% Tween-20 in PBS (PBST) before being incubated with LICOR Infrared Dye-conjugated secondary antibodies (IRDye 800CW goat ⍺-Rabbit IgG for rabbit ⍺-Ryp antibodies and Donkey ⍺-Mouse IgG for mouse ⍺-tubulin antibody) for one hour at RT. Membranes were subjected to three washes in 0.1% PBST before being visualized using the LI-COR Odyssey CLx imaging system.

Detecting secreted proteins in Histoplasma

Secreted proteins were detected as previously described [9]. G217B WT and rpd3Δ cells cultured under standard host conditions were grown to saturation before equal volumes were pelleted through centrifugation. The resulting supernatants were filter sterilized using a 0.22 µM syringe filter and subjected to concentration using Amicon Ultra Centrifugal Filter Units with size cutoff of 3 kDa cutoff (Millipore Sigma UFC900324). For each sample, equal amounts of concentrated supernatants were separated using SDS-PAGE and stained using InstantBlue Coomassie Protein Stain (ISB1L –abcam 119211). Gels were then washed with PBS and imaged using a Canon Canoscan LiDE210.

Genomic DNA isolation and sequencing in Histoplasma

Cell pellets for genomic DNA (gDNA) isolation were harvested through pelleting 1 mL of culture via centrifugation at 12,000 rpm. Cells were washed and resuspended in 1× TE before being subjected to DNA extraction using the Qiagen Gentra Puregene Yeast/Bacteria kit (158567) as previously described [20]. Purified gDNA was treated with RNase A before being quantified and submitted for library preparation using the Illumina DNA prep kit and Whole Genome Sequencing at SeqCenter. Paired-end sequencing was performed using an Illumina NovaSeq X Plus at SeqCenter.

Whole genome sequencing analysis

Reads were aligned to the Histoplasma ohiense G217B genome (GCA_017607445.1) [86] using BWA MEM [87]. We confirmed deletion of target genes and lack large CNV by visual inspection of read coverage.

Chromatin isolation and shearing

Cultures processed for ChIP-seq were crosslinked with formaldehyde (Polysciences 18814-10) added to a final concentration of 1% and kept under culture conditions (shaking at 37 °C with 5% CO₂ or ambient RT) for 20 min to allow fixing to take place. Crosslinking was quenched with glycine added to a final concentration of 125 mM for 5 min at RT. Fixed cells were pelleted at 3,000 g for 10 min at 4 °C and washed twice with TBS (20 mM Tris-HCl (pH 7.5), 150 mM NaCl). Washed cells were harvested through filtration, aliquoted into pellets of equal mass (between 150 and 300 mg/tube) and flash frozen in liquid nitrogen as described above.

Frozen pellets were resuspended in lysis buffer (50 mM Hepes/KOH (pH 7.5), 140 mM NaCl, 1 mM EDTA, 1% Triton X-100, 0.1% sodium deoxycholate, protease/phosphatase inhibitors) at 4 °C before being bead beat with 500 µL of 0.5 mm zirconia beads in 5 × 1 min cycles at RT with 2 min rests on ice in between. The resulting lysate was then spun down at 8,000 rpm for 10 min at 4 °C to isolate the insoluble chromatin fraction. This chromatin pellet was resuspended in 600−900 µL of lysis buffer and sonicated at 4 °C using the Diagenode Bioruptor Pico for 30−35 cycles (30 s on, 30 s off). Sonicated samples were spun down for 5 min at 14,000 rpm at 4 °C to pellet cell debris, and aliquots for input DNA were reserved at −20 °C in TE + 1% SDS. Input samples were also used to assess shearing using a High Sensitivity DNA Bioanalyzer chip from Agilent. Equal lysate volumes were aliquoted for each IP to be performed and brought to a final volume of 500 µL with lysis buffer and 20 µg of each polyclonal ⍺-Ryp antibody (Ryp1: 3873, Ryp2: 387, and Ryp3: 356) or 3 µg of a ChIP-grade monoclonal ⍺-histone-acetyl antibody (H3K9Ac: ab177177, H3K14Ac: ab52946, H4K16Ac: ab109463). Samples were incubated overnight with agitation at 4 °C before being incubated with 50 µL of a 50% slurry of Protein A Dynabeads (Life Technologies Corporation 10002D) pre-washed 2× with TBS and 3× lysis buffer for 2 hours at 4 °C with agitation.

Immunoprecipitated DNA was recovered through pulldown on Protein A beads using a magnetic rack. Pellets containing DNA-protein complexes were washed 2× in lysis buffer, 2× with high-salt (500 mM NaCl) lysis buffer, 2× with wash buffer (10 mM Tris/HCl (pH 8.0), 250 mM LiCl, 0.5% NP-40, 0.5% Na-deoxycholate, 1 mM EDTA) and 1× with TE before adding elution buffer (50 mM Tris/HCl (pH 8.0), 10 mM EDTA, 1% SDS) and incubating at 65 °C for 10 min. DNA was eluted using a magnetic rack and the resulting beads were subjected to second round of resuspension and elution with 150 µl TE + 0.65% SDS and pooled with the initial eluate to maximize total yield of DNA. IP and input samples were treated with 1 µL of Proteinase K (20 mg/mL Qiagen 19131) and incubated at 65 °C for approximately 16 hours. IP and Input DNA samples were then treated with 2 µg of RNase A (Qiagen) and incubated at 37 °C for 1 hour. DNA was then purified using the Zymo ChIP DNA Clean and Concentrator Kit (D5205) and eluted in 31 µL of ddH₂O or TE.

ChIP-seq library preparation

Purified IP and input DNA were used for Illumina library preparation using the NEBNext Ultra II DNA Library Prep Kit (NEB E7103S) and Illumina dual-indexed primer pairs sourced from the UCSF CAT. The quality and size distribution of each library was assessed using the Agilent High Sensitivity DNA Bioanalyzer Kit (Agilent Technologies 5067-4626) to ensure proper fragment size and the absence of contaminating adapters. Final library concentrations were measured using a Qubit HS dsDNA assay kit (Life Technologies Q33230) and used to pool equimolar amounts of each sample into final libraries for sequencing. Pooled libraries were used for single-end sequencing (DMSO versus 50 µM Ent samples in Figs 5 and 6) or paired-end sequencing (WT versus rpd3Δ samples in Figs 5 and 6) using one NovaSeq X 10B lane at the UCSF CAT.

Transcription factor ChIP-seq TF analysis

Single or paired-end reads were aligned to the Histoplasma ohiense G217B genome (GCA_017607445.1) [86] using BOWTIE2 version 2.4.2 [88] with default parameters. Peaks were called for each individual replicate with MACS2 version 2.2.7.1 [56] with default parameters except that min-length was set to 100 for paired-end reads as recommended by Gaspar 2018. Aggregate peaks were defined by taking the union of the coordinates of overlapping peaks among replicates for regions where peaks were called for at least two replicates [90]. The aggregate peaks defined for 37 °C samples for a given TF were used for quantification of all samples for that TF. Per peak signal was quantified by counting reads or read pairs in each aggregate peak, log₂ transforming with pseudocount of 1 to handle missing data, and normalizing to the signal in the NPS1 negative control region. This negative control region was defined as the middle 500 bp of NPS1 (I7I48_05472, JAEVHH010000002: 444765..449764 (+)). Differential enrichment was then calculated as the difference of the means of the log₂ signal over replicates from each condition. Peaks were assigned to individual genes if any part of the peak fell in the intergenic region between the stop codon of the upstream gene and before the start codon for that gene, limiting the intergenic size to a maximum of 10 kb as described in [90].

Histone modification ChIP-seq analysis

Single or paired-end reads were aligned to the Histoplasma ohiense G217B genome (GCA_017607445.1) [86] using BOWTIE2 version 2.4.2 [88] with default parameters. Per sample promoter signal was quantified by counting reads or read pairs in each promoter, log₂ transforming with pseudocount of 1 to handle missing data, and normalizing to the signal in the NPS1 negative control region. This negative control region was defined as the middle 500 bp of NPS1 (I7I48_05472, JAEVHH010000002: 444765..449764 (+)). Differential enrichment was then calculated as the difference of the means of the log₂ signal over replicates from each condition.

Phylogenetic analysis and tree generation

HMMSEARCH from HMMER 3.3.2 [91] was used to query the Hist_deacetyl (PF00850) Pfam domain against predicted gene sets for 41 fungal genomes (S18 Data of Gilmore and colleagues [18]) yielding 172 proteins with expectation value ≤ 1e-6. These sequences were aligned with PROBCONS 1.12 [92] and a phylogenetic tree was inferred from the alignment using FASTTREE 2.1.11 [93].

Other software and libraries

We wrote custom scripts and generated plots in R 4.0.4 [94] and Python 3.9.2, using Numpy 1.19.5 [95] and Matplotlib [96]. Jupyter 4.7.1 [97] notebooks and JavaTreeView 1.1.6r4 [97] were used for interactive data exploration.

BMDM differentiation and culture conditions

Bone marrow-derived macrophages (BMDMs) were isolated from the tibias and femurs of 6- to 8-week-old C57BL/6J (Jackson Laboratories stock, No. 000664) mice. Mice were euthanized via CO₂ narcosis and cervical dislocation as approved under UCSF Institutional Animal Care and Use Committee protocol. Cells were differentiated in bone marrow macrophage (BMM) media containing Dulbecco’s Modified Eagle Medium (DMEM High Glucose), 20% Fetal Bovine Serum (FBS), 10% v/v CMG supernatant (the source of CSF-1), 2 mM glutamine, 110 mg/mL sodium pyruvate, 110 mg/mL penicillin and streptomycin. Undifferentiated monocytes were plated in BMM for 7 days at 37 °C with 5% CO₂. Adherent cells were collected and frozen in 40% FBS and 10% DMSO and stored in liquid nitrogen for future use.

Macrophage infections

Macrophage infections with G217B Histoplasma strains were performed as previously described [7–9,102]. Macrophages were seeded in tissue culture-treated dishes in triplicate one day before infection and incubated at 37 °C with 5% CO2. On the day of infection, yeast cells from logarithmic-phase cultures (OD₆₀₀ = 4–7) were collected, resuspended in D-PBS (PBS Ca²⁺, Mg²⁺ free), sonicated for 3 s on setting 2 using a Fisher Scientific Sonic Dismembrator Model 100, and counted at 40× using a hemacytometer. Due to the morphology phenotype of the rpd3Δ mutant, chains of budding yeast cells were counted as one cell. Macrophages were infected at appropriate MOIs and allowed to phagocytose Histoplasma for 2-hours before being washed with D-PBS to remove any extracellular yeast and fresh BMM media was added. For infections lasting longer than 2 days, the remaining cells were supplemented with fresh BMM media.

Lactate dehydrogenase (LDH) release assay

To quantify macrophage lysis, BMDMs were plated (7.5 × 10⁴ cells per well of a 48-well plate) and infected in triplicate as described above. At the indicated time points, the amount of LDH in the supernatant was measured as described previously [99–100]. BMDM lysis was calculated as the percentage of total LDH from supernatant of wells with uninfected macrophages lysed in 1% Triton X-100 at the time of infection. Due to continued replication of BMDMs over the course of the experiment, the total LDH at later time points can be greater than the total LDH from the initial time point, resulting in lysis that is greater than 100%.

Immunofluorescence of rpd3Δ infection

To image macrophages, BMDMs were seeded at 1.5 × 10⁵ cells on 12 mm circular glass coverslips in 24-well tissue culture dishes. Infections were done as previously described with G217B WT and the rpd3Δ mutant Histoplasma. At each time point post-infection, cells were fixed with 1 mL of 4% paraformaldehyde for 20 min at RT. After fixation, coverslips were washed 3 times with D-PBS to remove any residual fixative. Fixed cells were permeabilized and blocked using 0.2% w/v Saponin and 2% FBS in D-PBS for 1 hour at RT. Cells were then incubated with AlexaFluor-647-conjugated anti-F4/80 mouse antibody (UCSF mAB core) at a dilution of 1:500 for 1 hour at 37 °C in a light-blocking incubation chamber. Coverslips were washed 5 times with D-PBS and then briefly rinsed in double-distilled water. Excess liquid was gently removed using edge of a Kimwipe. Finally, coverslips were mounted on glass slides with ~10 µL of Vectashield mounting media with DAPI and 1:100 Calcofluor White (1 mg/mL). Coverslips were sealed to the glass slide using clear nail polish.

Image acquisition and processing

Z-stacks (0.3 mm increments) images were taken on Nikon Eclipse Ti CSU-X1 spinning disk confocal microscope. FIJI [101] was used to generate max-intensity projections for all conditions.

Chemical inhibition of histone deacetylases disrupts Histoplasma thermal dimorphism

To identify HDAC-encoding genes in Histoplasma, we used a Hidden Markov Model of the classical HDAC domain (PF00850) [108] to query a panel of fungal proteomes that included four representative species of Histoplasma. The phylogenetic tree generated by this analysis revealed two distinct classes of Zn²⁺-dependent HDACs in Histoplasma (Fig 1A). We used reference sequences from Saccharomyces cerevisiae to assign Histoplasma orthology corresponding to Class I (Rpd3 and Hos2) and Class II (Hos3 and Hda1) HDACs. Expanding this analysis to include a broader set of environmental and pathogenic fungal species highlighted the shared conservation and minimal expansion of these two HDAC classes across fungi (S1A Fig). While paralogs of the Class I HDAC RPD3 are found in opportunistic fungal pathogens such as C. albicans and C. neoformans [31,34], we did not identify any expansion of HDACs in the panel of thermally dimorphic fungal pathogens included in this analysis (S1A Fig). The Hos1 outgroup found in S. cerevisiae, Candida spp., and C. neoformans [40] was similarly absent in Histoplasma spp. (Fig 1A) in addition to the other fungal genomes included in this analysis (S1A Fig).

To investigate a putative role for HDACs in Histoplasma thermal dimorphism, we assessed the morphology of yeast cells following challenge with small-molecule inhibitors of classical (i.e., Zn²⁺ dependent) HDACs. Because functional redundancy of classic HDACs has been observed in related fungal species [34,41,42], we initially tested the effect of Suberoylanilide Hydroxamic Acid (SAHA), a pan-HDAC inhibitor capable of targeting all classic HDACs within a cell [43]. Yeast cells maintained at 37 °C were passaged in the presence of DMSO (control) or an increasing dose of SAHA for 2 days and subsequently imaged to qualitatively assess any alteration in yeast morphology (Fig 1B). Interestingly, inhibiting all classic HDAC activity with SAHA disrupted normal yeast growth at 37 °C and resulted in improper filamentation in a dose-dependent manner (Fig 1C). To determine if this phenotype could be attributed to a narrower subset of HDACs in the cell, we assessed the effects of class-specific HDAC inhibitors. The Class I-specific inhibitors Entinostat and Mocetinostat had a similar effect to SAHA at comparable doses, while Class II inhibitors failed to elicit any effect on cell morphology at 37 °C (Fig 1D and 1E). The effect of Entinostat was not dependent on the presence of N-Acetyl glucosamine (GlcNAc) in the media (Fig 1F), which has been shown to stimulate filamentation in Histoplasma [13]. Entinostat also hastened the onset of filamentation in cells transitioned to growth at RT (Fig 1G) in a similar manner to what has been observed with cAMP-driven filamentation [19].

Because temperature-mediated morphogenesis is tightly conserved across Histoplasma species [18], we queried the effect of Entinostat in strains that are distinct from our North American laboratory strain, G217B. Entinostat disrupted yeast morphology and induced filamentation at 37 °C in Histoplasma strains isolated from Central America and Africa (Fig 1H). This shared effect suggests a conserved requirement for Class I HDAC activity in temperature-mediated morphogenesis of Histoplasma at 37 °C.

Chemical inhibition of Class I histone deacetylases disrupts temperature-dependent transcription in Histoplasma

Prior transcriptomics studies have identified and characterized unique gene expression signatures that underlie each morphological state in Histoplasma [11–13,44]. Given the morphology phenotypes we observed following Class I HDAC inhibition, we next determined if inhibition of Class I HDACs disrupted the normal gene expression program at 37 °C. We used transcriptional profiling to assess the gene expression signatures of cells undergoing a chemical (HDACi) or physiological (ambient temperature) cue driving filamentation. We grew cultures for 1–2 days at 37 °C with either 10 or 50 µM of the Class I HDACi Entinostat (Ent). In parallel, we passaged yeast cells with vehicle (DMSO) at 37 °C and RT to serve as comparison points for both morphological states (yeast and early hyphae, respectively) as shown in Fig 2A. In contrast to untreated cells that grew primarily as chains of elongated cells shortly after the shift to RT, cells treated with 50 µM of Entinostat formed filaments robustly at RT and grew as a mix of hyphae and elongated cells by day 2 (Fig 2A). We performed RNA-seq and subjected our data to Principal Component Analysis (PCA) to generate a comprehensive view of transcript abundance variance across these conditions (Fig 2B). Overall, Entinostat-treated cells appeared to be transcriptionally distinct from both untreated yeast and early hyphae, with a more pronounced difference for the 50 µM treatment (Fig 2B). The second principal component highlighted a shared transcriptional response in early RT-shifted and Entinostat-treated cells that may represent a common set of gene expression changes associated with the onset of filamentation (Fig 2B).

We performed differential gene expression analysis of control yeast cells at 37 °C versus cells undergoing both filamentation cues to identify the pathways and genes driving the global signatures observed by PCA. While slight temporal and dose-dependent effects across conditions were observed, the expression profiles were highly correlated between the two doses on the same day and between the two days for the same Entinostat treatment, with a slightly stronger effect being observed at day 2 (S2A–S2C Fig). By day 2, 4,241 transcripts were differentially expressed (> 2-fold change, 5% FDR) across the three conditions: 2,421 following RT shift, 1,243 with 10 µM of Entinostat, and 2087 with 50 µM Entinostat (Fig 2C).

We used UpSet plots [45] to compare differentially expressed gene (DEG) sets defined across the conditions we tested. Unsurprisingly, a vast majority of the 10 µM Entinostat signature was shared with the effect observed at a higher dose (Fig 2D and 2E). While more than half of the genes repressed by Entinostat were also repressed during early RT shift (Fig 2D), the number of shared upregulated genes across these two sets was noticeably smaller (Fig 2E). Conventionally, Class I HDACs are thought to repress transcription [46], so we were interested in identifying the genes encoding transcripts that showed an inappropriate increase in abundance in 50 µM Entinostat as they could be regulatory targets of interest. We expanded our analysis to include DEG sets defined in steady-state hyphae [18] since cells grown in 50 µM of Entinostat morphologically resembled these hyphae more closely than cells undergoing filamentation at RT at these timepoints. These comparisons revealed that a much larger fraction of upregulated genes was shared between Entinostat-treated cells and steady-state hyphae than between RT-shifted cells and steady-state hyphae (S2D and S2E Fig). A majority of the genes repressed by Entinostat were also repressed under conditions that promote filamentation (S2F Fig). Furthermore, the correlation between the signatures of Entinostat and stationary hyphae was stronger than that between Entinostat and acutely RT-shifted cells, both globally and with respect to phase-specific genes (Figs 2F, 2G, and S2D). These comparisons suggest that Entinostat treatment was inducing a program that normally emerges over time as hyphae develop at RT.

The genes whose expression was dysregulated by Class I HDAC inhibition have functions that are critical to the basic biology of Histoplasma yeast or hyphae. A majority (22/40 yeast-specific and 103/165 hyphae-specific) of canonically morphology-responsive genes [14] were significantly dysregulated following Class I HDAC inhibition. This set of genes includes secreted effectors required for virulence in vivo such as CBP1, YPS3, and KNOT1 [5–7,9,10] (Fig 2H). These yeast-enriched effectors are strongly repressed in Entinostat-treated cells in a manner comparable to that of cells undergoing shift to RT (Fig 2H). Additionally, despite Entinostat-treated cells being cultured at 37 °C, we observed induction of the hyphae-associated transcripts MS8 [47,48] and MS95/DDR48 [13,18,49] analogous to the induction of these genes shortly after the shift to ambient temperature (Fig 2H). Interestingly, we also observed the improper expression of TFs that promote filamentation in Histoplasma [13,14] and morphogenesis in related fungi [50–52]. We observed strong induction of STU1, PAC2, NOS1, ESDC, and FBC1 in Entinostat-treated cells that was not observed following early RT-shift (Fig 2H) but is typically observed at later timepoints in steady-state hyphae [18]. To determine if the ability of Entinostat to trigger the hyphal program was dependent on a subset of regulatory factors, we compared the Entinostat response of WT Histoplasma with strains lacking individual TFs. We determined that filamentation induced by Entinostat at 37 °C requires the presence of FBC1, a C2H2 family TF required for filamentation at RT, but not PAC2, a WOPR family TF dispensable for filamentation at RT [19] or VEA1, a Velvet domain TF that reduces the rate of filamentation after shift to RT [53,107] (Fig 2I). Altogether, our transcriptional profiling revealed that Class I HDAC inhibition at 37 °C disrupts the expression of key yeast-associated genes while simultaneously promoting the improper expression of hyphae-associated genes, including regulators that promote filamentation. These results suggest a role for Class I HDACs in mediating temperature-dependent gene regulation by maintaining the regulatory networks that promote the expression of yeast-phase genes at 37 °C while simultaneously repressing regulators that promote filamentation.

The RPD3 gene is required for yeast-phase growth at 37 °C

Through our phylogenetic analysis shown in Fig 1A, we identified two Class I HDACs encoded in Histoplasma that we herein named RPD3 and HOS2. To determine a genetic basis for the phenotypes observed following Class I HDAC inhibition, we used a CRISPR/Cas9-based system [53] optimized for use in Histoplasma to generate deletion mutants for these HDACs in G217B. We transformed WT yeast with plasmids expressing Cas9 alongside a pair of sgRNAs designed to excise a majority of each gene (Fig 3A). Following transformation, we screened for successful knock-out (KO) mutants via a colony PCR assay (S3A and S3B Fig) and validated successful KO strains through whole-genome sequencing (S3C and S3D Fig). The morphology of rpd3Δ cells was reminiscent of the effect of Class I HDAC inhibitors in WT cells at 37 °C. At 37 °C, cells lacking RPD3 formed filaments (Fig 3B) and wrinkled colonies (S3E Fig), whereas hos2Δ strains displayed no observable colony or cell morphology phenotype (Figs 3B, and S3E). In addition to having aberrant morphology at 37 °C, rpd3Δ cells also displayed a slight growth defect (S3F Fig) and enhanced susceptibility to cell wall stress caused by Congo Red and CFW (S3G Fig).

We generated a construct expressing RPD3 under the ACT1 promoter rather than the RPD3 promoter, for technical reasons. The construct was transformed in parallel into WT and rpd3Δ strains. Re-introduction of RPD3 fully rescued the ability of rpd3Δ strains to grow as yeast at 37 °C (Fig 3D and 3E). Additionally, overexpression (OE) of RPD3 under the control of the ACT1 promoter in the WT background did not appear to have any adverse effect on morphology (Fig 3D). To assess if complementation of the rpd3Δ phenotype by RPD3 required its HDAC activity, we used protein alignments to map the conserved histidine residues required for catalytic activity [54] (S4A Fig) in Histoplasma Rpd3 and generated a plasmid expressing a catalytically dead (CD) Rpd3 mutant. Transformation of this plasmid into rpd3Δ failed to rescue the morphology phenotype (S4B Fig) suggesting the HDAC activity of Rpd3 is required for normal yeast growth. However, sequence-based methods to validate maintenance of the CD plasmid across multiple transformations indicated that this plasmid was poorly maintained in rpd3Δ in a manner that was not observed in WT strains or for the WT RPD3 plasmid in the rpd3∆ strain. Importantly, these findings reveal that the effect of Entinostat is largely attributable to inhibition of Rpd3 function, and indicate that the activity of Rpd3, but not Hos2, is required to maintain the yeast phase at 37 °C.

RPD3 is required for the pathogenic yeast-phase transcriptional program at 37 °C

To investigate the effect of RPD3 and HOS2 on the transcriptional landscape, we subjected WT and mutant strains to RNA-seq. Cultures of WT, rpd3Δ, and hos2Δ strains were grown at 37 °C and then passaged in parallel at 37 °C and RT for 2 days. In contrast to the elongated cells produced by WT and hos2Δ cells 2 days following growth at RT, rpd3Δ cells formed filaments robustly and grew uniformly as hyphae (Fig 4A). We performed RNA-seq on these samples and through PCA found that nearly half of the variance in transcript levels (PC1, 47.49%) across both conditions was driven by Rpd3 (Fig 4B). Of note, a portion of this component is shared between WT and hos2Δ cells grown at RT and rpd3Δ grown at 37 °C, which suggests a subset of the physiological RT transcriptome is improperly induced in the rpd3Δ strain.

We compared transcript abundance between WT cells at 37 °C and RT as well as WT versus each mutant at each temperature. We defined 5,480 DEGs in this dataset through these comparisons (Fig 4C). At 37 °C we identified 2,618 DEGs in rpd3Δ versus WT and 745 in hos2Δ versus WT, while at RT we defined 3,585 DEGs in rpd3Δ versus WT and 2,239 in hos2Δ versus WT. The robust signal at RT for rpd3Δ is consistent with the strong filamentation phenotype observed under these conditions (Fig 4A). Altogether, these observations encompass changes to over half of the annotated transcriptome of Histoplasma [18].

To focus on the underlying gene expression changes driving the rpd3Δ phenotype, we subjected the DEGs defined in our 37 °C data to comparative analysis. At 37 °C in rpd3∆ cells, the abundance of 1,610 transcripts increased whereas the abundance of 1,008 transcripts decreased. These data are in contrast to hos2Δ yeast at 37 °C where 495 transcripts increased in abundance and 250 decreased in abundance, consistent with the lack of phenotypic changes in the hos2∆ mutant. We generated UpSet plots to compare the DEG sets identified in the mutants at 37 °C alongside time-matched WT cells following shift to RT. We found that the largest intersecting set of DEGs in both directions was shared between rpd3Δ at 37 °C and WT cells undergoing filamentation at RT (Fig 4D and 4E). This finding corresponded with our global observations via PCA and, to an extent, recapitulated our findings using Entinostat to disrupt Rpd3 function. A vast majority (85.8%) of phase-specific genes was dysregulated in rpd3Δ: 144/165 hyphae-specific transcripts accumulate improperly at 37 °C, and 32/40 yeast-specific transcripts were improperly depleted at 37 °C (Fig 4F). We compared the global signatures of rpd3Δ and Entinostat-treated cells at 37 °C and found that they were strongly correlated (r = 0.76) (Fig 4F). A majority of differentially expressed genes were also shared amongst the two conditions (Fig 4H and 4I). We compared the global profiles of our HDAC mutants to steady-state hyphae and found that rpd3Δ but not hos2Δ transcriptionally mirrored stationary hyphae and dysregulated phase-specific genes in a manner that corresponded with our observations with Entinostat (S5A–S5C Fig). Furthermore, we also found that more DEGs were shared between rpd3Δ and steady-state hyphae than between Entinostat-treated cells and steady-state hyphae (S5D and S5E Fig). When we further interrogated these gene sets, we similarly discovered strong depletion of key yeast-associated transcripts such as CBP1, YPS3, LDF1, and KNOT1, and simultaneous accumulation of hyphae-associated transcripts, including MS8, FBC1, STU1, YAP1, and ESDC in rpd3Δ cells at 37 °C (Fig 4G). Of note, this effect appeared to be magnified in rpd3Δ grown at RT and in a few instances, for hos2Δ at RT (Fig 4G). Altogether, we uncovered large-scale disruptions to gene expression shared between rpd3Δ and Entinostat-treated cells at 37 °C that encompassed much of the canonical temperature-regulated gene network. These findings further support a role for Rpd3 in maintaining the normal regulatory landscape in Histoplasma that underlies yeast-phase growth in response to temperature.

Rpd3 regulates the activity of transcription factors that drive thermal dimorphism

Virtually all yeast transcripts (96%) depend on a network of temperature-responsive TFs (Ryp1–4) for their proper expression at 37 °C [15–17]. The Ryp TFs bind multiple sites in the genome and share a number of direct regulatory targets [17]. While a majority of Ryp-mediated gene regulation in yeast appears to be indirect, the set of direct targets shared by Ryp1–3 is enriched for yeast-specific genes [17,55]. We noticed that a considerable number of Ryp-dependent genes are transcriptionally dysregulated following chemical and genetic disruption of Rpd3. We compared our DEG sets to the shared targets of Ryp1–3 previously determined by chromatin immunoprecipitation (ChIP) [17] and determined that this core set of yeast-specific genes was significantly enriched for genes that were dysregulated following a high dose of Entinostat or KO of RPD3 (Fig 5A). 24.2% of shared Ryp1–3 targets were dysregulated in 50 µM of Entinostat (Fisher’s exact test, p = 0.019) while 38.7% were dysregulated in rpd3Δ (Fisher’s exact test, p = 1.66e−06) (Fig 5A).

To determine if disruption in yeast phase-specific gene expression following KO of Rpd3 could be attributed to a change in Ryp protein levels, we performed Western blotting in WT cells following 2-day exposure to DMSO or 50 µM Entinostat and rpd3Δ cells at 37 °C. We detected comparable levels of all three Ryp proteins between the two RPD3 perturbations and their controls with a negligible decrease in Ryp1 and Ryp3 in rpd3Δ (Fig 5B), indicating that changes in the transcriptome were not due to depletion of Ryp TFs.

Rpd3 has been found to influence the binding and activity of stress-responsive TFs that regulate the response to heat shock in S. cerevisiae [27–30]. Because Ryp-dependent genes are dysregulated following Rpd3 disruption, we hypothesized that Rpd3 might affect the ability of the Ryp TFs to associate upstream of their regulatory targets. We performed ChIP-seq to investigate the genome-wide binding patterns of Ryp1–3 in WT cells following 2 days of growth with DMSO or 50 µM Entinostat at 37 °C. Ryp signal in control cells was strong at the CBP1 locus, recapitulating our previous observations (Fig 5C) [17]. Strikingly, this signal was predominantly lost in cells treated with Entinostat (Fig 5C), indicating that Entinostat affected Ryp association with the DNA. We similarly observed a decrease in Ryp ChIP signal upstream of CBP1 in rpd3Δ cells in comparison to WT cells (Fig 5D). We used the MACS2 peak finding algorithm [89] to identify sites of Ryp enrichment in promoters across the genome. Although the Ryp2 peak signal was muted, making the Ryp2 analysis difficult, we identified high-quality peaks for Ryp1 and Ryp3 in both control and treatment conditions. We found that the vast majority of Ryp1 and Ryp3 binding events were lost in cells grown with Entinostat (Fig 5E). We identified 102 Ryp1 and 236 Ryp3 binding events in control cells, whereas only 7 Ryp1 and 46 Ryp3 peaks were identified in Entinostat-treated cells (Fig 5E). Of these peaks, only 2 Ryp1 peaks and 20 Ryp3 peaks were shared between the two conditions (Fig 5E). These findings suggest class I HDAC function is required to maintain the normal genome-wide binding patterns of the Ryp TFs that takes place at 37 °C.

We then assessed differential Ryp binding by quantifying and normalizing ChIP signal in MACS2-defined [56] Ryp binding sites and generated contrasts between control and Entinostat-treated cells. We observed a comprehensive decrease in Ryp ChIP signal in peaks upstream of known and putative effector genes [10] in cells undergoing treatment with Entinostat (Fig 5F). We identified Ryp peaks in rpd3Δ strains using MACS2 and similarly found that a majority of Ryp1–3 binding events were lost in rpd3Δ. In total, 617/775 of the Ryp1 peaks, 266/406 Ryp2 peaks, and 429/658 of the Ryp3 peaks identified in WT cells were lost in rpd3Δ (Fig 5G–5I). Notably, we also identified a substantial number of Ryp1–3 targets that were unique to rpd3Δ. We identified 244 Ryp1, 216 Ryp2, and 243 Ryp3 bound genes in rpd3Δ strains that were not present in WT cells (Fig 5G–5I). This suggested that loss of RPD3 may drive changes to chromatin structure that result in significant changes to association sites of the Ryp proteins across the genome. We followed our peak-level analysis with quantification of differential ChIP signal in WT-defined peak loci in rpd3Δ and similarly observed a loss of Ryp signal in the promoters of putative effector genes that are typically enriched in yeast phase (Fig 5J). Combined, these findings point to a role for Rpd3 activity in maintaining the normal regulatory landscape that facilitates Ryp binding and activity at 37 °C.

Rpd3 is required for normal histone acetylation patterns at the loci of known and putative regulators of filamentation

Rpd3 regulates responses to thermal and osmotic stress in S. cerevisiae by targeting histones for deacetylation at the loci of stress-responsive genes [28,29]. This hypoacetylation subsequently decreases genome accessibility to restrict the binding of stress-responsive TFs and blocks the recruitment of transcriptional machinery [28,29]. At the time of our study, no experimental datasets had defined or characterized histone acetylation patterns in Histoplasma, so we based our selection of Rpd3 substrates on findings in other fungi. In S. cerevisiae, Rpd3 is reported to regulate transcription through the removal of acetyl groups deposited on N-terminal lysine residues of histones H3 and H4 [46,54]. We used ChIP-seq to globally query the dynamics of histone H3 acetylation following chemical or genetic inhibition of Rpd3 at 37 °C. We performed pull-downs on acetylated H3K9 (H3K9Ac) and H3K14 (H3K14Ac), two marks that label transcriptionally active loci [46] targeted by Rpd3 for deacetylation and subsequent transcriptional silencing [57]. We hypothesized the loci surrounding genes normally deacetylated (and thereby transcriptionally repressed) by Rpd3 at 37 °C may display signs of hyperacetylation when Rpd3 function is compromised. We generated coverage tracks of normalized histone acetyl-ChIP signal and found that loci encoding TFs with putative roles in filamentation are hyperacetylated following disruption of Rpd3. Notably, these TFs were also found to be transcriptionally induced following either chemical or genetic perturbation of Rpd3 (S6A Fig). One of the transcripts most strongly upregulated following either Rpd3 disruption or in steady-state hyphae is a homolog of esdC in Aspergillus spp., a putative regulator of sexual development [52]. We observed increased H3K9Ac and H3K14Ac signal spanning the ESDC promoter and gene body in Entinostat-treated and rpd3∆ cells (Fig 6A) that was not present in control cells. A homolog of YAP1, a bZIP TF that regulates environmental stress responses in S. cerevisiae [58], displayed similar hyperacetylation of H3K9 and H3K14 across its promoter and gene body regions (Fig 6B) and underwent transcriptional induction following either perturbation of Rpd3 (S6A Fig). A similar result was observed surrounding the start site of I7I48_01662, a hybrid homeobox/C2H2 zinc-finger TF identified as being strongly induced in conditions that drive filamentation [19] (Fig 6C). We also observed an increase in promoter and gene body acetylation at the locus of canonically temperature-responsive genes such as TYR1 (S7A Fig), a gene whose expression is strongly induced following shift to RT.

We also detected changes in acetylation patterns that correspond with two alternative transcript start sites (TSS) in the 5′ region of WET1, a TF whose regulation is critical to Histoplasma morphology [18]. These alternate TSSs yield WET1 transcripts with two distinct leader sequences—a longer leader in yeast that is poorly translated, and a shorter leader in hyphae that is robustly translated [18]. We observed strong H3K9Ac acetylation peaks surrounding the yeast-specific TSS of WET1 in control cells that was lost in Entinostat-treated and rpd3Δ cells (Fig 6D). Simultaneously, we also detected an increase in H3K9 acetylation surrounding the downstream hyphae-specific TSS and spanning the hyphal isoform of the WET1 transcript (Fig 6D). Of note, we also observed instances of Rpd3-independent H3K9 hypoacetylation at the loci of yeast-specific genes (S6B and S6C Fig), suggesting that there may be additional HDAC/HAT complexes that regulate chromatin state to silence yeast-specific genes following the onset of filamentation. Quite strikingly, the changes in H3K9Ac signal observed at these loci following chemical or genetic ablation of RPD3 at 37 °C phenocopied what is observed in WT cells undergoing filamentation in response to temperature shift (Fig 6A–6D).

We quantified differential H3K9Ac and H4K13Ac ChIP signal on ATG-defined promoter regions across the Histoplasma genome to globally assess changes to histone acetylation. Notably, this quantification of differential acetylation signal revealed a strong correlation (r = 0.83) between Entinostat-treated cells and cells undergoing filamentation following shift to ambient temperature (Fig 6E), highlighting the similarity between the effect of chemical inhibition of Rpd3 and the normal consequences of temperature shift. Overall, only a small subset of promoters displayed signs of differential H3K9Ac and H3K14Ac signal following Entinostat treatment, RPD3 disruption, or temperature shift. Similar pulldowns and promoter analysis of acetylated H4K16, a modification that is not considered a primary regulatory target of Rpd3 [60], failed to identify instances of differential promoter acetylation across the entire genome (S12 Table). We indexed our differential promoter signal generated against H3K9Ac and H3K14Ac on putative TF-encoding genes across the Histoplasma genome (Fig 6F). Clustering this set of data highlighted a subset of TFs with hyperacetylated promoters following Entinostat treatment or KO of RPD3 (Fig 6F). This quantification corroborated our manual observations of ChIP-seq coverage in the promoters of ESDC, WET1, and YAP1 (Fig 6G). We additionally observed increased acetylation signal in the promoters of the known and putative pro-filamentation TFs such as STU1, VELC, NOS1, and WET1 (Fig 6G). Notably, these TFs were also among those we identified for being transcriptionally induced following chemical or genetic targeting of Rpd3 (S6A Fig). These findings suggest Rpd3 may function to drive deacetylation at the loci and cis-regulatory regions of pro-filamentation TFs to repress their inappropriate transcription in yeast cells at 37 °C. Upon temperature shift, acetylation is restored.

Rpd3 is required for virulence in a macrophage model of infection

As Rpd3 regulates a majority of defined temperature-specific genes (Fig 4F), we hypothesized that its regulon would encompass virulence genes coupled to yeast-phase growth. We indexed our RNA-seq data following chemical and genetic ablation of Class I HDACs on a comprehensive set of putative effectors across the G217B genome previously defined based on a combination of high-throughput experiments and sequence features [10]. We found that most of the putative effectors encoded in G217B are repressed at 37 °C when Rpd3 function is chemically or genetically compromised (Fig 7A). Similarly, these transcripts show decreased abundance in control cells undergoing filamentation at RT (Fig 7A).

One of the Rpd3-dependent effectors we identified is CBP1, a secreted virulence factor required for macrophage lysis in vitro and pathogenesis in a mouse model of infection [5–9]. CBP1 is one of the most abundant and efficiently translated transcripts produced by Histoplasma yeasts [18], so we speculated that the decrease in Cbp1 transcript abundance observed when Rpd3 is compromised would be sufficient to result in a decrease to Cbp1 protein levels. We used an established SDS-PAGE assay on concentrated culture supernatants [9,10] harvested from WT and rpd3Δ cultures grown to saturation at 37 °C to qualitatively assess Cbp1 protein levels. We observed a marked decrease in the abundance of Cbp1 in rpd3Δ strains in addition to other apparent alternations in the profile of secreted proteins (Fig 7B). We previously found that the extent of macrophage lysis following infection with Histoplasma correlates with the abundance of Cbp1 [8]. To determine if the repression of virulence genes, including Cbp1, in rpd3Δ affects Histoplasma virulence in a macrophage model of infection, we infected murine BMDMs with WT or rpd3Δ Histoplasma and measured macrophage lysis as previously described [7,99]. We found that the rpd3Δ mutant is completely unable to lyse BMDMs and that rpd3Δ-infected BMDMs are virtually indistinguishable from uninfected BMDMs throughout the time-course of infection (Fig 7C). This was in stark contrast to WT-infected BMDMs that were entirely lysed by the end of the infection (Fig 7C).

Histoplasma yeasts attain a high intracellular fungal burden prior to the onset of macrophage lysis [61]. We wanted to verify that the striking lysis defect in rpd3Δ was not confounded by an intracellular growth defect in BMDMs as the rpd3Δ mutant displayed a slight in vitro growth defect (S2F Fig). Due to its altered morphology, we faced technical challenges with plating-based methods of assessing intracellular growth of rpd3Δ so, as an alternative, we subjected BMDMs infected with WT and rpd3Δ to fluorescence imaging to qualitatively assess intracellular growth over the time-course of infection. Remarkably, we observed that rpd3Δ cells replicated within macrophages and reached a high fungal burden without lysing BMDMs (Fig 7D). Interestingly, this recapitulates the phenotypes of strains lacking a functional copy of CBP1, as cbp1Δ yeast are capable of replicating within macrophages but fail to induce host cell lysis [5–9]. In addition, we found that the growth of hyphae that occurs during infections with the rpd3Δ strain is not sufficient to mechanically breach the macrophage plasma membrane (Fig 7D) in contrast to what has been observed in other intracellular fungi that utilize filamentation and polarized growth to induce host cell lysis [62]. These observations indicate that Rpd3 is required for virulence in macrophage infections through its role in promoting the expression of key yeast-associated virulence genes at 37 °C, including Cbp1.

Fig 8. Rpd3 integrates with transcription factor networks to promote Histoplasma yeast-phase growth under host conditions.

Proposed model for Rpd3 function under host conditions. Rpd3 is required for the repression of hyphal-phase-specific (HPS) genes and the expression of yeast-phase-specific (YPS) genes at 37 °C. Our data suggest a model where Rpd3 represses HPS gene expression through the removal of acetyl (Ac) groups on histone H3 (histone), most notably in the promoters of HPS transcription factors (TFs) that may antagonize the Ryp network. We hypothesize the genomic landscape established by Rpd3 is indirectly (dashed arrows) required to promote activation of YPS genes through promoting the DNA-binding activity of the Ryp TFs (Ryp1-3). Thus, Rpd3 is required for Histoplasma yeast-phase morphology and virulence gene expression to promote macrophage lysis at 37 °C.

https://doi.org/10.1371/journal.pbio.3003341.g008

S1 Fig. Comprehensive assessment of classic HDAC phylogeny in fungi.

Phylogenetic tree of classic HDAC-encoding genes in a panel of 41 fungal proteomes encompassing human fungal pathogens as well as environmental, entomopathogenic, and phytopathogenic fungi. S5 Table contains the full list of gene names and their corresponding genomes. Genes found in the G217B laboratory strain used in this study are bolded. The underlying tree file can be found in S2 Data.

https://doi.org/10.1371/journal.pbio.3003341.s001

(S1_Fig.PNG)

S2 Fig. Entinostat induces hyphal gene expression across different conditions at 37 °C and drives filamentation through known morphology TFs.

Scatterplots comparing differential RNA-seq signal in all G217B transcripts between 50 µM of Entinostat on day 1 and day 2 (A), 10 µM of Entinostat on day 1 and day 2 (B), 10 and 50 µM of Entinostat on day 2 (C), and 2-day RT shift and steady-state hyphae (D) (Gilmore and colleagues [18]). Venn diagrams comparing the regulons of 2-day RT (short RT), 50 µM Entinostat, and steady-state hyphae (previously published in Gilmore and colleagues [18]) to identify shared and unique upregulated (E) and downregulated genes (F). Underlying data can be found in S6 Table.

https://doi.org/10.1371/journal.pbio.3003341.s002

(S2_Fig.TIF)

S3 Fig. Validation and primary characterization of HDAC KO mutants.

Colony PCR products run on a 1% Agarose gel to identify successful KO isolates for RPD3 (A) and HOS2 (B). The top half of each gel contains PCR products amplified with a primer pair external to the sgRNA sites that is indicated in Fig 3A. The bottom half uses a primer pair that flanks both sides of the 5′ sgRNA cut site such that negative signal is indicative of clean KO. Coverage tracks displaying sequencing signal at the chromosomal coordinates surrounding RPD3 for control and rpd3Δ strains (C), and at HOS2 for control and hos2Δ strains (D). The underlying data can be found in SRA SRP593199. E. Plate morphology phenotypes of WT, rpd3Δ, and hos2Δ strains. F. OD₆₀₀ growth curve of WT and rpd3Δ strains grown for 5 days at 37 °C. Points represent the average OD₆₀₀ values of samples in triplicate. G. OD₆₀₀ growth curve of rpd3Δ passaged at 37 °C for 4 days in the presence of cell wall stressors. Points represent the average OD₆₀₀ values of samples in triplicate. Underlying data can be found in S16 Table.

https://doi.org/10.1371/journal.pbio.3003341.s003

(S3_Fig.TIF)

S4 Fig. Identifying putative histidine residues required for catalytic activity of Rpd3 in Histoplasma.

A. Alignment of the Rpd3 HDAC domain across fungal species and alongside human HDAC1 (HDAC1/124-198). Histoplasma G217B Rpd3 is indicated by the red asterisk. The red arrows indicate the 150H and 151H residues observed to be critical for catalytic activity in orthologous species. B. Representative micrographs of transformants yielded following transformation of WT and rpd3Δ strains with an empty vector or a plasmid overexpressing a Catalytic Dead (CD) variant of Rpd3.

https://doi.org/10.1371/journal.pbio.3003341.s004

(S4_Fig.TIF)

S5 Fig. The global gene expression signature of rpd3Δ is better correlated with hyphae and Entinostat than hos2Δ.

Scatterplots comparing differential RNA-seq signal in all G217B transcripts between steady-state hyphae and the following: rpd3Δ at 37 °C (A) and hos2Δ at 37 °C (B). Venn diagrams comparing the regulons identified in 2-day 50 µM Entinostat-treated cells, rpd3Δ at 37 °C, and steady-state hyphae (Gilmore and colleagues [18]) to identify shared and unique sets of upregulated (C) and downregulated (D) genes. E. Global scatterplot comparing differential RNA-seq signal as in (A, B) between 37 °C signal for hos2Δ cells and cells grown with 50 µM Entinostat for 2 days. Underlying data can be found in S6 and S7 Tables.

https://doi.org/10.1371/journal.pbio.3003341.s005

(S5_Fig.TIF)

S6 Fig. Correspondence between transcriptional activation and histone acetylation in phase-specific genes.

A. Scatter plot displayed in Fig 4F indexed on TF-encoding genes in the G217B genome. The green stars point out candidate hyphal TFs that are significantly induced under one of the two Rpd3-disrupting conditions. Underlying data can be found in S6 and S7 Tables. Coverage tracks displaying fold enrichment traces for H3K9Ac signal surrounding the loci of yeast-associated genes. Red traces indicate signal in control cells for both conditions in the loci surrounding CBP1 (B) and YPS3 (C). The underlying data for these panels can be found in GSE316185, GSE316355, and GSE316386.

https://doi.org/10.1371/journal.pbio.3003341.s006

(S6_Fig.TIF)

S7 Fig. Rpd3 is required for hypoacetylation of hyphal and conidial genes at 37 °C.

Coverage tracks displaying traces of genomic ChIP/input signal for pulldowns on two different histone H3 acetylation marks, H3K9Ac and H3K14Ac. 2-day 37 °C cultures of WT G217B yeast treated with DMSO or 50 µM Entinostat along with G217B WT and rpd3Δ strains were used for chromatin isolation and subsequent pulldown. Colored traces (H3K9Ac in red and H3K14Ac in green) depict the signal in WT or DMSO control cells while the black traces represent signal with Entinostat or in rpd3Δ surrounding the loci of TYR1 (A), GAD1 (B), and MS95 (C). The underlying data for these panels can be found in GSE316185, GSE316355, and GSE316386.

https://doi.org/10.1371/journal.pbio.3003341.s007

(S7_Fig.TIF)

S5 Table. Histone deacetylase genes.

Table of histone deacetylase genes in Excel-compatible comma-separated values format. Each row is a gene with columns “Gene Name”: name as given in S1 Fig, “Gene ID”: systematic ID from genome annotation, “Genome”: name of genome annotation, as given in S18 Data of Gilmore and colleagues [18]. The final columns give the genus and species of the source genome.

https://doi.org/10.1371/journal.pbio.3003341.s012

(S5_Table.CSV)

S6 Table. KALLISTO estimated counts and LIMMA-fit values for Entinostat response expression profiles.

Excel-compatible tab-delimited text conforming to JavaTreeView extended CDT format. Each row is a transcript, with the UNIQID column giving the ucsf_hc.01_1.G217B systematic gene name. The NAME column gives short names taken from S1 Data of Voorhies and colleagues [86] with additional names and corrections based on human curation. Description and HcG217B_pred (WUSTL G217B predicted gene accessions), HcG217B_rc (Edwards and colleagues G217B transcript accessions [11]), and HcG217B_acc (UCSF1 gene accessions [86]); and Ryp1_ChIP, Ryp2_ChIP, Ryp3_ChIP, and Ryp4_ChIP (indicating genes with promoters bound by Ryp TFs in [17]) are taken from S1 Data of [86]. The next 7 columns give LIMMA BH-adjusted p-values for differential expression in each of the 7 contrasts, and the corresponding LIMMA-fit log2 ratios are given in the final 7 columns. The next 21 columns give KALLISTO estimated counts for each transcript in each sample. GWEIGHT is a place-holder column for JavaTreeView compatibility. The estimated counts in this file are sufficient to recapitulate the LIMMA analysis.

https://doi.org/10.1371/journal.pbio.3003341.s013

(S6_Table.TXT)

S7 Table. KALLISTO estimated counts and LIMMA-fit values for HDAC mutant expression profiles.

Excel-compatible tab-delimited text conforming to JavaTreeView extended CDT format. Each row is a transcript, with the UNIQID column giving the ucsf_hc.01_1.G217B systematic gene name. The NAME column gives short names taken from S1 Data of [86] with additional names and corrections based on human curation. Description and HcG217B_pred (WUSTL G217B predicted gene accessions), HcG217B_rc (Edwards and colleagues G217B transcript accessions [11]), and HcG217B_acc (UCSF1 gene accessions [86]); and Ryp1_ChIP, Ryp2_ChIP, Ryp3_ChIP, and Ryp4_ChIP (indicating genes with promoters bound by Ryp TFs in [17]) are taken from S1 Data of [86]. The next 5 columns give LIMMA BH-adjusted p-values for differential expression in each of the 5 contrasts, and the corresponding LIMMA-fit log2 ratios are given in the final 5 columns. The next 18 columns give KALLISTO estimated counts for each transcript in each sample. GWEIGHT is a place-holder column for JavaTreeView compatibility. The estimated counts in this file are sufficient to recapitulate the LIMMA analysis.

https://doi.org/10.1371/journal.pbio.3003341.s014

(S7_Table.TXT)

S8 Table. Ryp1 ChIP enrichment in response to Entinostat.

Excel-compatible tab-delimited text conforming to JavaTreeView extended CDT format. Each row represents a unique gene to peak association as given in the UNIQID column. The NAME column gives short names taken from S1 Data of [86] with additional names and corrections based on human curation. Description and HcG217B_pred (WUSTL G217B predicted gene accessions), HcG217B_rc (Edwards and colleagues G217B transcript accessions [11]), HcG217B_acc (UCSF1 gene accessions [86]), and HcG217B (Gilmore and colleagues transcript accessions [18]); and Ryp1_ChIP, Ryp2_ChIP, Ryp3_ChIP, and Ryp4_ChIP (indicating genes with promoters bound by Ryp TFs in [17]) are taken from S1 Data of [86]. Following a GWEIGHT column, included for JavaTreeView compatibility, is ChIP enrichment for each sample followed by mean enrichment over all samples (A) and differential enrichment (DMSO/Ent50), with all enrichment values on a log2 scale.

https://doi.org/10.1371/journal.pbio.3003341.s015

(S8_Table.TXT)

S9 Table. Ryp3 ChIP enrichment in response to Entinostat.

Excel-compatible tab-delimited text conforming to JavaTreeView extended CDT format. Each row represents a unique gene to peak association as given in the UNIQID column. The NAME column gives short names taken from S1 Data of [86] with additional names and corrections based on human curation. Description and HcG217B_pred (WUSTL G217B predicted gene accessions), HcG217B_rc (Edwards and colleagues G217B transcript accessions [11]), HcG217B_acc (UCSF1 gene accessions [86]), and HcG217B (Gilmore and colleagues 2015 transcript accessions [18]); and Ryp1_ChIP, Ryp2_ChIP, Ryp3_ChIP, and Ryp4_ChIP (indicating genes with promoters bound by Ryp TFs in [17]) are taken from S1 Data of [86]. Following a GWEIGHT column, included for JavaTreeView compatibility, is ChIP enrichment for each sample followed by mean enrichment over all samples (A) and differential enrichment (DMSO/Ent50), with all enrichment values on a log2 scale.

https://doi.org/10.1371/journal.pbio.3003341.s016

(S9_Table.TXT)

S10 Table. Ryp1, Ryp2, and Ryp3 ChIP enrichment in HDAC mutants.

Excel-compatible tab-delimited text conforming to JavaTreeView extended CDT format. Each row represents a gene associated with at least one Ryp as given in the UNIQID column. The NAME column gives short names taken from S1 Data of [86] with additional names and corrections based on human curation. This is followed by Ryp-specific peak coordinates; where more than one peak exists for a given gene, the peak with the best MACS2 score is given. Description and HcG217B_pred (WUSTL G217B predicted gene accessions), HcG217B_rc (Edwards and colleagues G217B transcript accessions [11]), HcG217B_acc (UCSF1 gene accessions [86]), and HcG217B (Gilmore and colleagues transcript accessions [18]); and Ryp1_ChIP, Ryp2_ChIP, Ryp3_ChIP, and Ryp4_ChIP (indicating genes with promoters bound by Ryp TFs in [17]) are taken from S1 Data of [86]. Following a GWEIGHT column, included for JavaTreeView compatibility, is log2 differential ChIP enrichment between WT and rpd3 samples for each Ryp.

https://doi.org/10.1371/journal.pbio.3003341.s017

(S10_Table.TXT)

S11 Table. Enrichment of Histone 3 modifications in response to Entinostat or rpd3 mutation.

Excel-compatible tab-delimited text conforming to JavaTreeView extended CDT format. Each row represents a protein-coding gene as given in the UNIQID column. The NAME column gives short names taken from S1 Data of [86] with additional names and corrections based on human curation. Description and HcG217B_pred (WUSTL G217B predicted gene accessions), HcG217B_rc (Edwards and colleagues G217B transcript accessions [11]), HcG217B_acc (UCSF1 gene accessions [86]), and HcG217B (Gilmore and colleagues transcript accessions [18]); and Ryp1_ChIP, Ryp2_ChIP, Ryp3_ChIP, and Ryp4_ChIP (indicating genes with promoters bound by Ryp TFs in [17]) are taken from S1 Data of [86]. Following a GWEIGHT column, included for JavaTreeView compatibility, is log2 differential ChIP enrichment for H3K9 of H3K14 acetylation for the given contrasts.

https://doi.org/10.1371/journal.pbio.3003341.s018

(S11_Table.TXT)

S12 Table. Enrichment of Histone 4 Lysine 16 modifications in response to Entinostat.

Excel-compatible tab-delimited text conforming to JavaTreeView extended CDT format. Each row represents a protein-coding gene as given in the UNIQID column. The NAME column gives short names taken from S1 Data of [86] with additional names and corrections based on human curation. Description and HcG217B_pred (WUSTL G217B predicted gene accessions), HcG217B_rc (Edwards and colleagues G217B transcript accessions [11]), HcG217B_acc (UCSF1 gene accessions [86]), and HcG217B (Gilmore and colleagues transcript accessions [18]); and Ryp1_ChIP, Ryp2_ChIP, Ryp3_ChIP, and Ryp4_ChIP (indicating genes with promoters bound by Ryp TFs in [17]) are taken from S1 Data of [86]. Following a GWEIGHT column, included for JavaTreeView compatibility, is log2 differential ChIP enrichment for H4K16 acetylation.

https://doi.org/10.1371/journal.pbio.3003341.s019

(S12_Table.TXT)