Biocontrol against fungal mycoparasites driven by Bacillus velezensis strains from a mushroom crop microcosm

Agaricus bisporus ; Mycoparasites, Casing material; Biocontrol; Secondary 21 metabolites, Specialised metabolites, Lipopeptides. 22 Abbreviations: NGS, next generation sequencing ; CD, cobweb disease; DBD, dry bubble 23 disease; WBD, wet bubble disease; PLFA, phospholipid fatty acid. 24 Repositories: Sequence data for this study are deposited in NCBI BioProject collection under 25 BioProject ID PRJNA1142391. 26

27 The cultivation of b utton mushroom ( Agaricus bisporus) requires the design of tailor -made 28 substrates that nourish the crop and promote morphology changes from mycelium to 29 basidiome. The agronomic stages of mushroom development are also influenced by the 30 microbiota present in the mushroom crop microcosm, which may have a beneficial impact on 31 mushroom growth, development and quality, or a detrimental impact through reduction of yield 32 or quality as pathogens, competitors or disease vectors in mushroom crops. Bacillus velezensis 33 strains isolated from commercial mushroom casing material have been demonstrated to have 34 antifungal activity against fungal mycoparasites of mushroom crops. In this report we describe 35 the isolation of multiple strains of B. velezensis from mushroom casing material and basidiomes 36 that show antifungal activity towards major mushroom mycoparasites, along with further 37 characterization of their mode of action. Full genomes of B. velezensis CM5, CM19, CM35, 38 EM5 and EM39 were sequenced and annotated , which together with metabolic profiling of 39 specialised metabolites produced by CM5, CM19 and CM35 suggested that antifungal activity 40 of these strains is linked to the production of the lipopeptide fengycin. However, in crop trials, 41 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 2, 2025. ; https://doi.org/10.1101/2025.09.02.673639doi: bioRxiv preprint 3 these strains did not increase mushroom yield or provide significant control of the mushroom 42 pathogen Lecanicillum fungicola. Genomic and analytical tools were designed and used to 43 evaluate B. velezensis persistence in casing when the selected strains were artificially applied. 44 B. velenzensis population levels decrease d significantly after application , potentially 45 contributing to the lack of biocontrol activity observed in crop trials. 46

47 The commercial cultivation of the secondary decomposer white button mushroom (Agaricus 48 bisporus) is based on the transformation of agricultural and farming by-products into selective 49 nutritional substrates by means of a 3-phase composting strategy: i) maturation of raw materials 50 through subsequent mesophilic and thermophilic stages; ii) production of a parasite-free 51 substrate through a pasteurization and stabilization stage; iii) induction of A. bisporus 52 mycelium growth in the selective compost (Wang et al., 2021). When the selective substrate is 53 fully colonized by the mycelium it is placed in environmentally controlled growing units and 54 topped with bio-based casing material to provide the physical, chemical and microbiological 55 requirements to induce basidiome fructification(Dias et al., 2021). 56 Microbial parasites are a major issue for crop loss es in the food industry (Fones et al., 2020). 57 The fungal parasites associated with green mould (Trichoderma spp.), wet bubble (Mycogone 58 perniciosa), dry bubble ( Lecanicillium fungicola ), and c obweb ( Cladobotryum spp. ) are 59 responsible for the majority of reported losses caused by pathogenic or competing fungi within 60 the mushroom industry (Gea et al., 2021). It is worth mentioning the difficulty of breeding for 61 resistance in Agaricus bisporus , which lies in the typical life cycle of this fungus , which 62 hampers the introduction of traits without considerable linkage drag (Sonnenberg et al., 2017). 63 Therefore, to fight such harmful fungal mycoparasites, selective a gricultural fungicides have 64 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 2, 2025. ; https://doi.org/10.1101/2025.09.02.673639doi: bioRxiv preprint 4 historically been used in mushroom farms, including chlorothalonil, carbendazim, metrafenone 65 or prochloraz-Mn (Kosanović et al., 2013; Commission E., 2019). However, these fungicides 66 are now either prohibited, or are no longer effective due to resistance outbreaks, necessitating 67 the development of alternative approaches for disease control. Together with efficient hygiene, 68 biocontrol is a potential strategy to replace chemical pesticides as part of integrated disease 69 management programmes (Gea et al., 2021). 70 There are multiple mechanisms of biocontrol described depending on the agent employed , 71 including: antagonism though production of lytic enzymes, antimicrobial peptides , 72 antimicrobial volatile organic compounds (VOCs) or other antibiotics ; spatial and nutritional 73 competition; stimulation of mushroom growth through mycohormone production ; and 74 enhanced nutrient uptake to improve the fungal immune system (Zhang and Sun, 2018 ; 75 Carrasco and Preston, 2020). 76 B. velezensis is a bacterial species that includes several relatively well -studied strains with 77 biocontrol activity. These are reported to produce antimicrobial metabolites that are able to 78 control both bacteria and fungi (Rabbee et al. , 2023; Kenfaoui et al. , 2024 ). Examples of 79 antimicrobial metabolites produced by B. velezensis include bacillaene, bacillibactin, bacilysin, 80 difficidin, fengycin, macrol actin H and subtilin (Pandin et al., 2018). Of these, fengycin - a 81 lipopeptide with strong antifungal activity - is of particular interest in attaining control over 82 fungal parasites of mushrooms. Fengycin (also known as plipastatin) is synthesized by a non-83 ribosomal peptide synthetase (NRPS) complex encoded by a large gene cluster known as the 84 fen operon (pps in B. subtilis ). This operon encodes five enzymes, each responsible for 85 recognizing and incorporating specific amino acids during the process of building the final 86 fengycin peptide (Zihalirwa Kulimushi et al., 2017; Desmyttere et al., 2019; Gao et al., 2019). 87 Fengycin is known to disrupt fungal cell membranes by interacting with sterols, leading to 88 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 2, 2025. ; https://doi.org/10.1101/2025.09.02.673639doi: bioRxiv preprint 5 increased membrane permeability and cell death (Mantil et al., 2019). Fengycin, and biocontrol 89 agents which produce fengycin, have been shown to be active against many fungal parasites of 90 mushroom crops including Trichoderma aggressivum (Pandin et al., 2019; Kosanović et al., 91 2021), Cladobotryum mycophilum (Clarke et al., 2022a; Clarke et al., 2024), Lecanicillium 92 fungicola (Clarke et al., 2022) and Mycogone perniciosa (Novikova and Titova, 2023). 93 There are multiple classes of f engycin broadly classified into two main classes. These are 94 categorised using the amino acid present at the sixth position of the protein chain, with fengycin 95 A containing alanine, and fengycin B containing valine (Steller et al., 1999). Further classes of 96 fengycins produced by Bacillus strains have also been described (Yin et al., 2024), including 97 for instance fengycin S and fengycin C produced by Bacillus amyloliquefaciens LSC04 and 98 Bacillus subtilis EA-CB0015, respectively (Lee et al., 2010; Villegas-Escobar et al., 2013). In 99 B. amyloliquefaciens the production of these lipopeptide analogues has been shown to be 100 subject to positive regulation by the quorum sensing system encoded by the comQXPA gene 101 cluster (Yin et al., 2024). The ComQXPA system is not unique to B. amyloliquefaciens or B. 102 subtilis (Kalamara et al., 2018), since it has also been described to regulate protease activity of 103 B. velezensis DMB05 (Na et al., 2023). 104 In this study, we isolated several novel B. velezensis strains from casing materials and 105 basidiomes that displayed antifungal activity to multiple mycoparasitic fungi. We characterise 106 these novel strains using phylogenetics and comparative genomics . We test three novel B. 107 velezensis strains for antimicrobial activity in crop via artificial inoculation on the casing 108 medium. In addition, both the survival rate of supplemented B. velezensis, and the equilibrium 109 of the endemic microbiota in this crop trial are assessed. Finally, we use LC -HRMS analysis 110 to identify and evaluate the secreted specialised metabolites of these B. velezensis strains that 111 correlate with their antifungal activity. 112 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 2, 2025. ; https://doi.org/10.1101/2025.09.02.673639doi: bioRxiv preprint 6

113 Isolation of the cultivable bacterial microbiome 114 Cultivable bacteria were obtained from three casing materials used by the mushroom industry 115 in commercial crops (commercially available casing materials cquired from local providers : 116 black peat, based on peat moss (Euroveen B.V., BVB Substrates, Grubbenvorst, Limburg, The 117 Netherlands); blonde peat, based on Spaghnum peat moss (Valimex KF, Valimex SL, Valencia, 118 Spain); and a mixture of both casing materials (50 % each) (Euroveen B.V. and Valimex KF), 119 the physical and chemical characteristics of which have been previously reported (Carrasco et 120 al., 2019). The methodology described by Aslani et al. (Aslani et al., 2018) was implemented 121 to isolate bacteria from basidiomes of Agaricus bisporus. 122 Microbiome c ultivation: i) for isolation of casing -colonising bacteria, 10 g of casing was 123 suspended in 100 mL of 0.1% sterile bacteriological peptone broth (Thermo Scientific 124 #LP0037B); ii) for isolation of basidiome-colonising bacteria, fresh and healthy closed 125 mushroom caps were externally disinfected by submerging them in 1% NaClO solution for 30 126 s and rinsing with sterile distilled water. 10 g of internal mushroom cap tissue was crushed and 127 homogenized with sterile pestle and mortar and suspended in 100 mL of 0.1% sterile 128 bacteriological peptone broth. Samples, either i) or ii) were added to Erlenmeyer flasks and 129 incubated in an orbital shaker-incubator (Ovan 1000001087, Badalona, Spain) at 120 rpm and 130 25 ºC for 10 min. The supernatant was transferred to a microtiter plate where successive 10 -131 fold dilutions (0.1% sterile bacteriological peptone broth ) were disp ensed. 100 μL of each 132 solution was transferred and homogeneously spread on growth media in 90 mm Petri dishes. 133 Media recipes employed in this work are described in Table S1. 134 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 2, 2025. ; https://doi.org/10.1101/2025.09.02.673639doi: bioRxiv preprint 7 Isolation and preservation of individual colonies: Exploring maximum diversity, one hundred 135 and forty-one individual colonies (Table S2) were re-isolated from plates based on colonial 136 morphology and transferred to plates containing LB Agar (LBA) medium. The re -isolated 137 colonies were incubated at 28 °C for 24-48 h. Single colonies were then added again to LB and 138 incubated in Erlenmeyer flasks between 25 and 28 °C in an orbital shaker at 150 rpm between 139 5 and 18 hours. The colony multiplication ratio was determined by spectrophotometer (Thermo 140 Fisher Scientific - Genesys 10-S) of the optical density in the culture broth at a wavelength of 141 600 nm (OD600). The fermentation was stopped when the absorbance at OD600 reached a 142 value of 1. The resulting cultures were re-suspended in 20% glycerol solution in sterile 2 mL 143 Eppendorf tubes. Finally, the tubes were stored at -80ºC (Panasonic MDF-U5386S-PE) until 144 further use. 145 Genomic extraction and sequencing 146 DNA extraction and Sanger sequencing of 16S rDNA: Single colonies were transferred to 10 147 mL LB at and incubated at 28 ºC on a platform shaker at 220 rpm for 24 h. The resulting 148 cultures were pelleted by centrifugation at 13,000 rpm for 1 min, and washed twice using 149 MilliQ water before aliquoting into 1.5 mL Eppendorf tubes. DNA was extracted from pellets 150 using the NucleoSpin Plant II, Mini kit for DNA from plants (MACHEREY -NAGEL GmbH 151 & Co. KG). The 27F and 1492R primers (Stackebrandt and Goodfellow, 1991) were used for 152 PCR amplification of 16S rDNA, which was performed using Phusion High -Fidelity PCR 153 Master Mix (Thermo Fisher, UK) , with an initial denaturation of 98 °C for 30 s, followed by 154 30 cycles of 7 s at 98ºC, 30 s at 56 ºC, and 60 s at 72 ºC. A final extension of 7 min was used 155 before storage at 4 ºC. PCR products were prepared for sequencing by treatment with ExoSAP-156 IT™ PCR Product Cleanup Reagent (Thermo Fisher, UK) and submitted to either Eurofins 157 (Luxembourg) or Source Bioscience (Nottingham, UK) for Sanger sequencing. Sequences 158 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 2, 2025. ; https://doi.org/10.1101/2025.09.02.673639doi: bioRxiv preprint 8 obtained using forward and reverse primers were aligned using MAFFT online 159 (https://mafft.cbrc.jp/alignment/server/). BLASTn searches were carried out using the NCBI 160 database (Camacho et al., 2009). 161 Full genome sequencing and analysis: The full genomes of Bacillus velezensis CM5, CM19, 162 CM35, EM5 and EM39 were sequenced and annotated. Four 10 mL bacterial overnight cultures 163 (LB) were combined, washed tw ice in PBS and diluted to OD600 = 1, before aliquoting and 164 pelleting 10 mL and re-suspending in 0.5 mL of DNA shield (Zymo Research , Leyden, NL). 165 Genome sequencing of the three strains was carried out by Microbes NG, Birmingham, UK. 166 The downstream processing of sequence data was accomplished using several bioinformatics 167 tools: Reads were adapter trimmed using Trimmomatic 0.30 with a sliding window quality 168 cutoff of Q15 (Bolger et al., 2014). De novo assembly was performed on samples using SPAdes 169 version 3.7 (Bankevich et al., 2012), and contigs were annotated using Prokka 1.11 (Seemann, 170 2014). The Artemis genome browser was used for visualization and annotation (Rutherford et 171 al., 2000). Assembly metrics were calculated by Quast (Gurevich et al., 2013). A standard 172 analysis pipeline assigned reads to the closest available reference genome using Kraken (Wood 173 and Salzberg, 2014). The RAST annotation server and the SEED annotation environment were 174 used to annotate the prokaryotic genome (Aziz et al. , 2008; Aziz et al. , 2012 ). Further 175 comparative genomics of novel B. velezensis strains was carried out using the NCBI 176 Prokaryotic Genomes Automatic Annotation Pipeline (PGAAP) (Tatusova et al., 2016). 177 The complete genome sequences of the B. velezensis strains reported in this paper have been 178 deposited in NCBI under the BioProject PRJNA1142391 with the genome accession numbers 179 CP168150 (CM19), CP168028 (CM5), CP168027 (CM35), CP168026 (EM5), CP168025 180 (EM39). To identify all genes containing single nucleotide polymorphisms (SNPs) in our novel 181 strains compared to the QST713 reference strain (Accession CP025079), we performed 182 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 2, 2025. ; https://doi.org/10.1101/2025.09.02.673639doi: bioRxiv preprint 9 comparative analyses using various computational tools and packages in R and Python. The 183 analysis workflow involved using BLASTn to detect differences in all genes present in the 184 QST713 reference genome. Genes with identified mismatches were then translated into their 185 corresponding amino acid sequences using the translate method from Biopython. These 186 translated sequences were subsequently compared using the BLASTp function to identify 187 amino acid substitutions. Coding sequences (CDS) with observed mismatches were further 188 analysed to ensure meaningful comparisons of functional protein-coding regions. BLAST Ring 189 Image Generator (BRIG) was employed for prokaryote genome comparisons (Alikhan et al., 190 2011). The potential specialised metabolite profiles of the sequenced strains was evaluated by 191 antiSMASH (Blin et al., 2021). 192 Phylogenetic analysis of B. velezensis strains: Phylogenetic analyses were performed using 193 five housekeeping genes: gyrB, pgk, rpoB, rpoD, and tuf. Genomes of comparative strains were 194 downloaded from NCBI and blasted against the annotated genes of the reference strain B. 195 velezensis FZB42. For our isolated strains of interest (CM5, CM19, CM35, CM18, CM20, 196 CM21), DNA extractions were carried out using the NucleoSpin Plant II, Mini kit for DNA 197 from plants (MACHEREY-NAGEL GmbH & Co. KG). Genes of interest were amplified by 198 PCR, carried out using Phusion® High -Fidelity DNA Polymerase master mix (NEB, Herts, 199 UK) and the primers listed in Table S4. PCR products were treated with Exo SAP as per the 200 manufacturer’s instructions and submitted to Source Bioscience (UK) for Sanger sequencing. 201 Both FWD and RVS primers were used for sequencing reactions. Concatenated sequences were 202 used to build a phylogeny using Multiple Sequence Comparison by Log -Expectation 203 (MUSCLE) (Edgar, 2004). The phylogenetic tree was built using IQTREE (Nguyen et al. , 204 2015) using 1000 bootstrap replicates and set to automatically detect the optimal number of 205 threads. The resulting tree was visualised using iTOL (Letunic and Bork, 2021). 206 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 2, 2025. ; https://doi.org/10.1101/2025.09.02.673639doi: bioRxiv preprint 10 Transformation of B. velezensis CM19 with pHAPII gfp+ 207 The plasmid pHAPII gfp+ (Cao et al., 2011) was extracted from a transformed strain of B. 208 velezensis QST713 (kindly provided by Romain Briandet, INRAE) using the NucleoSpin 209 Plasmid DNA extraction kit (Macherey -Nagel) using standard conditions. Subsequently, B. 210 velezensis CM19 cells were transformed using a room temperature electroporation protocol 211 (Morales-Ruiz et al., 2019). In brief, competent cells were made by washing and suspending a 212 stationary phase culture of CM19 in phosphate buffered sucrose electroporation buffer (PEB: 213 272 mM sucrose, 1 mM MgCl 2·6H2O, 7 mM phosphate buffer, pH 7.4). Electroporation was 214 carried out in a 1 mm cuvette containing 400 ng plasmid and 90 μl of bacterial cells using a 215 single electric pulse at 620V, 200R and 25 μF. Cells were immediately resuspended in SOC 216 medium and incubated at 37 ºC without shaking for 4 h. After plating, a single transformed 217 colony was present after 24 h incubation at 28 ºC. Validation was carried out by plasmid 218 extraction (Macherey-Nagel™ NucleoSpin™ Plasmid) and submission for sequencing with 219 Plasmidsaurus (www.plasmidsaurus.com). 220 In vitro fungal inhibition assays 221 In vitro con frontation trials: Four fungal parasites were used for biocontrol assays: 222 Trichoderma aggressivum (TAV1), Mycogone perniciosa (M25) and Cladobotryum 223 mycophilum (CM13900) isolated from commercial mushroom crops in La Rioja (Spain) ; and 224 Lecanicillium fungicola (150/1), a UK isolate of medium virulence obtained from Warwick 225 HRI, UK (Banks et al., 2019). All were stored as 2 5% glycerol stocks at -80 ºC. Before use, 226 fungi were plated onto PDA (potato dextrose agar) and grown for 3-7 days before scraping and 227 re-suspending in 30 mL MilliQ water in a 50 mL Falcon tube. Bacteria, also stored as 2 5% 228 glycerol stocks, were grown as overnight cultures in LB before use. For initial qualitative 229 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 2, 2025. ; https://doi.org/10.1101/2025.09.02.673639doi: bioRxiv preprint 11 biocontrol assessments, 10 µL of the fungal suspension was spotted on the centre of a PDA 230 plate. Four different 10 µL bacterial droplets from LB overnight cultures were spotted at a 231 distance of 20 mm from the fungal droplet. Plates were incubated in the dark at 25 ºC for 2 to 232 9 days depending on the parasite. Bacterial strains were scored as positively controlling fungal 233 parasites if a clear zone of clearing (halo) was observed around the bacterial colony. Each 234 bacterial strain was assessed at least twice. Since B. velezensis displayed biocontrol activity 235 against the four fungal parasites, further quantitative biocontrol plates were carried out using 236 just a single bacterial strain per plate. For this, 10 µL of the fungal cultures were co-inoculated 237 with bacterial droplets (at a distance of 30 m m) for between 1 and 6 days (depending on the 238 parasite). Plates were again incubated in the dark at 25 ºC for between 4 and 13 days. 239 Quantitative assessments were made by measuring zones of clearing. Each bacterial strain was 240 assessed in duplicate for at least two repeat experiments for each fungal parasite. 241 Spore germination assays Fully colonised PDA plates of the four fungal parasites were scraped 242 and resuspended in water through 8 layers of Miracloth. Using a haemocytometer, spore counts 243 were diluted to a concentration between 1 * 107 and 1 * 108 spores mL-1. A 100 µL aliquot was 244 spread over a PDA plate and allowed to dry for 20 min. Subsequently, a 10 µL droplet of an 245 overnight, stationary culture of B. velezensis CM19 pHAPII gfp (in LB) was spotted onto the 246 centre of the PDA plate and allowed to dry. After 1 -4 days of incubation at 28ºC, sections of 247 the agar plate around the bacterial droplet and halo of inhibition were cut and mounted on to a 248 microscope slide and covered with a cover slip. Fungal s amples were stained for cell death 249 using a droplet of propidium iodide (PI, Thermo Fisher, R37108) at a final concentration of 1 250 μL mL-1 and assessed using a Leica TCS SP5 confocal microscope. Samples were scanned at 251 400 Hz (512 × 512 pixel images) using an argon laser with emission at 488 nm (intensity, 17%) 252 to excite GFP (detected at 510 -530 nm) alongside a DPSS 561 nm laser (intensity, 35%) to 253 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 2, 2025. ; https://doi.org/10.1101/2025.09.02.673639doi: bioRxiv preprint 12 excite propidium iodide (detected at 625 -645 nm). Unchallenged fungal parasites (without B. 254 velezensis confrontation) grown in vitro on PDA plates, stained for cell death as described were 255 assessed as control. 256 Effect of isolated bacteria on A. bisporus mycelium growth: A. bisporus was plated onto 2% 257 agar compost medium (70 g of powdered compost suspended in 1 L of water, autoclaved, then 258 filtered through Miracloth, and autoclaved again) and incubated in the dark for 7 days. 10 µL 259 droplets of bacteria from LB overnight cultures were spotted onto the plates at a distance of 3 260 cm from the original plating position of A. bisporus. Plates were divided into quadrants, and 261 four droplets were dispensed per plate (one per quadrant) with A. bisporus H15 inoculated in 262 the centre. After a further 7 days, photographs were taken and the radius of the distance between 263 the centre of the bacterial droplet and the edge of the fungal radial growth was measured. 264 Controls consisted of four droplets of LB broth free of bacteria per plate. 265 Toxicity of bacteria to A. bisporus sporocarp tissue: For assessment of sporocarp tissue 266 damage, 10 mm cubes of mushroom tissue were cut from mushroom caps. Bacteria from 267 overnight LB cultures were pipetted onto mushroom cubes at a rate of 100 µL per cube. Cubes 268 were photographed and assessed for browning using a 0-4 scale where: 0 was no effect; 1 light 269 tissue scarring; 2 light browning; 3 heavy browning; and 4 being either black or dark brown 270 with obvious seepage. Control treatments consisted of mushroom cubes treated with sterile LB 271 broth at a rate of 100 µL per cube. 272 Biocontrol activity of B. velezensis in a crop trial infected with L. fungicola 273 Bacterial inoculum preparation: Bacteria were grown overnight in LB broth . Cultures with 274 OD600 > 0.7 were used to inoculate a bioreactor to produce bacterial batches through liquid 275 fermentation. A 5L Biostat A fermenter (Sartorius, Germany) was then used to culture large 276 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 2, 2025. ; https://doi.org/10.1101/2025.09.02.673639doi: bioRxiv preprint 13 bacterial batches (T = 28 ºC, pH = 7, critical oxygen concentration = 10-50% of air saturation, 277 time: 7-8 h to an OD600 = 1). The resulting broth was centrifuged at 4500 rpm for 10 min at 278 10ºC (Sorvall Legend XTR Centrifuge, Thermo Scientific, Waltham, MA, USA ). The 279 supernatant was removed and the pellet washed twice with PBS (phosphate buffer, pH = 7.4) 280 before resuspension in 100 mL of 10% skim milk and 10% saccharose solution as a 281 cryoprotector. The frozen solution was freeze-dried and pulverized into ready to use water 282 soluble powder. The viable concentrations were confirmed by plating serial dilutions in sterile 283 water onto LB. 284 Crop Trial: The crop trial was conducted in an experimental growth chamber equipped with 285 climate control (Mod. IGCS 1500 HR LED, Ibercex, Spain) according to standard practices 286 used in mushroom farms. The trial was performed with a randomized block design using 32 287 blocks (0.04 m2 crop surface) of 1250 g phase III compost (Sylvan H15 - supplemented with 288 Mylo Pro and Champfood) and 750 g of black casing (CNC, Valimex or Euroveen) at 289 maximum water holding capacity. The six treatment groups are shown in Table 1 (1: Control 290 (water) + L. fungicola (Lf); 2: Control (water) – Lf; 3: CM5 + Lf; 4: CM19 + Lf; 5: CM35 + 291 Lf;6: Prochloraz -Mn (PCL) + Lf ). Treatment with prochloraz-Mn ( PCL), as a fungicide 292 control, was carried out on day 4 after casing (d4), inoculation of bacteria was carried out on 293 d5, infection with Lf strain 150/1 (Banks et al., 2019) was carried out on d10. L. fungicola 294 inoculum was prepared as described by Gea et al. (Gea et al., 2014) on the day of inoculation 295 and applied by spraying the top of the casing layer at a dose of 106 conidia m -2 using a 296 suspension of 20 mL of 2 × 106 conidia L-1. Bacterial suspensions were applied on top of the 297 casing layer using 20 mL of a 2 × 109 cfu L-1 suspension in tap water for a final rate of 109 cfu 298 m-2. 299 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 2, 2025. ; https://doi.org/10.1101/2025.09.02.673639doi: bioRxiv preprint 14 For analysis of disease and biological efficiency, healthy and diseased mushrooms and bubbles 300 were scored at the end of the first and second crop flushes. For PLFA analysis in the casing 301 materials, four timepoints were assessed: d0 (Beginning of trial), day 10 (Fully colonized 302 casing), d21 (Fructification of first flush), and d29 (End of first flush). 1: Raw casing material 303 (beginning of the trial, casing free of host), 2: Casing fully colonised by Agaricus mycelium 304 (end of spawn running, day 10), 3: Casing during mushroom fructification (fructification of 1st 305 flush, day 21); 4: Casing sample taken by the end of 1st flush (end of the harvest, day 29). 306 Analysis of PLFAs: Casing microbial biomass and structure were determined using 307 phospholipid fatty acid (PLFA) analysis (Frostegård et al., 1993) and fatty acid methyl esters 308 (FAMEs) were analysed by gas chromatography. The total microbial biomass was estimated 309 by the total sum of PLFAs and expressed as nmol g−1. Specific PLFAs were used as biomarkers 310 to quantify the relative abundance of Gram-negative bacteria (monounsaturated fatty acids and 311 cyclopropyl 17:0) , Gram-positive bacteria (iso and anteiso saturated branched chain fatty 312 acids), Actinobacteria (10 -methyl fatty acids) and fungi (18:2 ω6 cis and 16:1w5) (Zelles, 313 1999). Statistical Analysis: Total biomass of bacteria and fungi determined by PLFAs was 314 submitted for the analysis of variance (ANOVA) by previous Levene variance homogeneity 315 tests to determine significant differences between samples. Means were compared by either 316 Tukey or Games -Howell post hoc test based on whether variance of homogeneity was met, 317 respectively (p < 0.05). ANOVA and post hoc tests were performed by IBM SPSS Statistics 318 v24 software package (SPSS Inc. Chicago, USA). Principal component analysis (PCA) was 319 performed with PAST v3.15 software (Hammer et al., 2011). 320 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 2, 2025. ; https://doi.org/10.1101/2025.09.02.673639doi: bioRxiv preprint 15 TaqMan assays for B. velezensis strains (CM19, CM5 and CM35) 321 DNA extraction: Bacteria were collected with a loop from the agar surface of 90 mm diameter 322 tryptic soy agar ( TSA) plates into a 2 mL tube and used for DNA extraction. The Wizard 323 Genomic DNA purification kit (Promega) was used for DNA extraction . DNA yield was 324 determined by using the Qubit ds DNA HS assay kit with the Qubit Fluorometer (Life 325 Technologies). 326 TaqMan design: To design a TaqMan assay that specifically detects the B. velezensis target 327 strains, we made use of the whole chromosomal genome sequence. Potential target sites were 328 identified using CLC genomic workbench (Qiagen, Aarhus, DK) . The genome sequences of 329 the target were dissected in 500 bp -long sequences. These sequences were mapped to 330 sequences of B. velezensis QST713 (CFSAN0334339), a non-target strain isolated from 331 agricultural soil (Pandin et al., 2018) (CLC mapping settings: length fraction: 0.85, similarity 332 fraction: 0.85, global alignment: no), thus removing sequences that mapped to any of the non-333 target strains. To assist the final selection and design step, these (unmapped) sequences were 334 added as a track list to the target strain genome. 335 Before proceeding to the design step, the candidate regions were checked for sequence 336 similarity with non-target organisms using BLASTn (NCBI) (Camacho et al., 2009). At least 337 six sets of primer/probe combinations were designed per target on 500 bp target -specific 338 fragments using primer quest tool of Integrated DNA Technologies (IDT, Leuven, Belgium) 339 with default settings. Two sets of primer/probe combinations were selected for the design of 340 the final triplex TaqMan assay, targeting different genes, based on the kinetics. 341 Specificity of the TaqMan assays: The specificity of the assays was tested using genomic 342 DNA of 27 strains (Table S6). For most strains 0.5 ng of purified DNA was used, but for some 343 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 2, 2025. ; https://doi.org/10.1101/2025.09.02.673639doi: bioRxiv preprint 16 strains a suspension of 109- 1010 cells/mL was boiled for 10 min in 150 µL MilliQ water (MQ) 344 prior to testing. For each TaqMan assay 2 µL of DNA or the boiled suspensions was mixed 345 with 2 µL reaction mix containing PerfeCTa multiplex qPCR ToughMix 5x (Quantabio, 346 Beverly, USA), 100 nM probe and 300 nM of each forward and reverse primer. The reactions 347 were performed in a 12K Flex QuantStudio or a QuantStudio 5 Real -Time PCR system 348 (Applied biosystems) using the following conditions: 95 °C for 2 min; 40 cycles of 95 °C for 349 15 s followed by 60 °C for 60 s. Analysis of the data was done by automatic threshold 350 calculation within the Applied Biosystems software. A Ct value ≤ 35 was considered positive. 351 The specificity and usability were further tested using constituents of casing soils or casing 352 soils of various compositions. DNA was extracted from 10 g of material using the DNeasy 353 PowerMax Soil kit (Qiagen, Germany) according to the manufacturer’s protocol. TaqMan 354 assays were performed as described above with 2 µL DNA. 355 Triplex TaqMan design: The two simplex assays for the target bacteria were combined with 356 an assay that quantifies Xanthomonas campestris pv. campestris (Xcc) into a triplex TaqMan 357 (Table S7) (Köhl et al., 2011; Taparia et al., 2020). For each group of biostimulants, a triplex 358 TaqMan assay was designed with two target -specific assays, each with its own dye (Cy5 or 359 FAM) and an amplification and extraction control ( X. campestris), labelled with HEX. The 360 triplex TaqMan was performed using the same PCR conditions and materials as the simplex 361 tests, with 100 nM probe and 300 nM of each primer in a total volume of 25 µL. 362 Limit of detection of the assays: The limit of detection for the triplex assays designed for B. 363 velezensis strains CM19, CM5 and CM35 (Table S12-S15) was determined in two casing soils, 364 i.e. in peat and steamed spent casing soil. 100 µL of a ten -fold serial dilution of 3 -days old 365 cultures of the strains in Ringers in a PowerBead Pro tube was mixed with 250 mg of the 366 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 2, 2025. ; https://doi.org/10.1101/2025.09.02.673639doi: bioRxiv preprint 17 casings (4 × 107 – 4 × 101 cells g-1 of casing soil). Casing soils with 100 µL Ringers without 367 supplemented bacteria served as a control. A volume of 100 µL of a suspension of X. 368 campestris (Xcc) of 10 6 cell mL-1 was added to each sample as extraction and amplification 369 control. The soils were subsequently freeze dried overnight and shaken for 2 times 90 seconds 370 in a paint shaker. DNA was extracted using a DNeasy Power Soil Pro Kit according to the 371 manufacturer’s instructions. The triplex TaqMan assay was executed as described above with 372 0.75 µL of each primer (10 µM) and 0.75 µL of each probe (10 µM), labelled with FAM, Cy5 373 (for target specific probes) or HEX (for Xcc). 374 Surfactant activity and lipopeptide production 375 Surfactant production: B. velezensis strains were cultured overnight in Optimised Medium 376 (OM, Table S1) and standardised to OD600 = 1. Droplets (10 µL) were pipetted onto the lid of 377 a Greiner CELLSTAR® 96 well plate, allowed to sit for 2 min, and measured using an eye 378 piece graticule on a ZEISS dissecting microscope. The mean droplet diameters of three 379 independent experiments, each containing 2 droplets, were assessed for differences to an OM 380 only control using Students’ T-Test. 381 Lipopeptide fraction collection: Lipopeptides were extracted from 600 mL liquid cultures of 382 CM19, which were incubated at 30 ºC for 24 h in Optimized Medium ( Table S1). Samples 383 were centrifuged at 10,000 g for 10 min to remove the cell pellet. The resulting solution was 384 acidified by addition of concentrated HCl to pH 2.0 and precipitated overnight at 4 °C. The 385 solution was again centrifuged at 10,000 g for 15 min at 4 °C. The pellet was resuspended in 386 10 mL methanol while shaking for 2 h continuously. The resulting solution was filtered through 387 a 0.4 µm filter and concentrated by evaporation to a final volume of 1 mL methanol. The 388 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 2, 2025. ; https://doi.org/10.1101/2025.09.02.673639doi: bioRxiv preprint 18 methanol suspension was then used for the confrontation assays by saturating the discs by 389 briefly incubating them in the solution. 390 Extraction of agar plates for targeted metabolic profiling : The B. velezensis strains CM5, 391 CM19 and CM35 were cultivated on L B agar plates for 5 days at 28 º C. For each strain, five 392 agar plates were extracted and pooled. Firstly, bacterial agar plates and negative control plates 393 (unseeded LB agar plates) were crushed and suspended in 20 mL HPLC-grade MeOH (J. T. 394 Baker, Deventer, The Netherlands). The mixture was transferred to a beaker, 30 mL of fresh 395 MeOH was added (total volume 50 mL) and further processed in an ultrasonic bath (Sonorex 396 Longlife, Bandelin, Berlin, Germany). After 10 minutes, the mixture was filtered and the wash 397 repeated. The filtrate was concentrated to dryness under reduced pressure at 35 °C in a rotary 398 evaporator (Laborota 4001 Efficient, Heidolph, Schwabach, Germany) operating at 100 rpm. 399 The residue obtained was dissolved in 1 mL HPLC-grade MeOH, centrifuged at 8000 rpm for 400 5 min, and measured by HPLC-HRMS. 401 High-performance liquid chromatography -high resolution mass spectrometry (HPLC -402 HRMS): The extracts were analysed on a LTQ Orbitrap XL mass spectrometer (Thermo 403 Scientific, USA) equipped with a HESI-II source coupled to Agilent (Santa Clara, USA) 1200 404 HPLC system including pump, PDA detector, column oven and autosampler. A Luna C 18 (2) 405 column (50 × 3 mm, 3 μm particle size) from Phenomenex (Torrance, USA) was used for the 406 HPLC-HRMS and an Aeris PEPTIDE XB-C18 (150 × 2.1 mm, 3.6 µm particle size) for HRMSn 407 measurements with a H 2O (+0.1% FA) (A) and CH 3OH (+0.1% FA) (B) gradient (flow rate 408 350 μL min-1). The gradient program was optimized and set as follows: 5% B isocratic for 2 409 min, linear gradient from 5% to 100% B over 24 min, 100% B isocratic for 6 min, the system 410 returned within 0.5 min to initial conditions of 95% A and was equilibrated for 3 min. For 411 selected target fengycins and their HRMSn measurements, the gradient was slightly modified: 412 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 2, 2025. ; https://doi.org/10.1101/2025.09.02.673639doi: bioRxiv preprint 19 5% B isocratic for 2 min, linear gradient from 5% to 90% B over 8 min and then over 20 min 413 up to 100%, the system returned within 0.5 min to initial conditions of 95% A and was 414 equilibrated for 3 min. The mass spectrometer w as operated in positive mode with a nominal 415 mass resolving power of 60000 at m/z 400 with a scan rate of 1 Hz, and N-butyl 416 benzenesulfonamide was used as lock mass ([M + H] + ; m/z 214.08963 for full scans and a 417 mass range from m/z 160-1600. Helium served as collision gas and N2 was used as sheath and 418 auxiliary gas. The HRMSn measurements were performed with collision-induced dissociation 419 of 42 eV. An authentic reference standard of fengycin (≥ 90%) was procured from Sigma -420 Aldrich (Steinheim, Germany). The analyses were performed using Xcalibur software v. 2.2 421 SP1.48. (Thermo Scientific, Bremen, Germany). The acquired masses were sorted by intensity 422 (I > 1.00E3). A maximum mass tolerance of 2 ppm was accepted. Background subtraction was 423 performed as required using Xcalibur software v. 2.2 SP1.48. (Thermo Scientific, Bremen, 424 Germany). 425

426 Bacillus velezensis strains isolated from mushroom casing microbiome display 427 antimicrobial activity towards mushroom mycoparasites. 428 One hundred and forty-one strains belonging to t he cultivable bacterial microbiome from the 429 peat casing layer and basidiomes of Agaricus bisporus were isolated and classified using 430 BLASTn (blast.ncbi.nlm.nih.gov) based on their 16 S rDNA sequences (Table S2 ). A wide 431 range of bacterial species were isolated of which Bacillus spp. and Pseudomonas spp. were 432 most highly represented ( Figure 1A). Initial a ntifungal confrontation assays (Table S3) 433 indicated that strains belonging to the species B. velezensis showed a particularly high degree 434 of antagonism towards four major fungal parasites of A. bisporus: Trichoderma aggressivum, 435 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 2, 2025. ; https://doi.org/10.1101/2025.09.02.673639doi: bioRxiv preprint 20 Mycogone perniciosa, Lecanicillium fungicola, and Cladobotryum mycophilum - these strains 436 were selected for further phenotypic analysis (Figure 1B and Table 2). 437 Nine strains of B. velezensis , and a control strain CM26 (BLASTn 98.47% Bacillus 438 oceanisediminis) which failed to show antagonistic activity in initial assays , were tested for 439 antifungal activity against four fungal parasites of the cultivated button mushroom Agaricus 440 bisporus in confrontation assays: L. fungicola (150/1); T. aggressivum (TAV1); M. perniciosa 441 (M25); and C. mycophilum (CM13900). All of the B. velezensis strains tested produced 442 statistically significant levels of inhibition with only CM26 showing significantly different 443 inhibition (Figure 2A, 4B; ONE WAY ANOVA, C. mycophilum p < 0.001, L. fungicola p < 444 0.001, M. perniciosa p = 0.139, T. aggressivum p < 0.001. Letters are results of post -hoc 445 TUKEY test showing significant differences p < 0.05). 446 B. velezensis CM19 was selected as a representative strain and transformed with the plasmid 447 pHAPII gfp. Transformed bacteria were confirmed to express GFP and used to examine the 448 effect of CM19 on conidial and hyphal viability and morphology in vitro (Pandin et al., 2017). 449 Live dead staining with propidium iodide (PI) w as used to study the effect of CM19 on spore 450 survival and germination (Figure 2C, D, E). Spores were classified as dead if either fully stained 451 with PI or observed to be hollow/lysed cells. In many cases spores were deemed to have 452 germinated but resulting hyphae were also found to be stained or lysed. 453 For a strain to be an effective biocontrol agent against mycoparasitic fungi it is important that 454 strains are not only antagonistic to mycoparasites, but also show low toxicity to hyphae and 455 sporocarps of A. bisporus. Strains were tested for antagonism to A. bisporus in both 456 confrontation assays and in mushroom cube assays, where cubes of sporocarp tissue were 457 exposed to dense suspensions of bacteria, a standard assay for mushroom pathogenic bacteria 458 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 2, 2025. ; https://doi.org/10.1101/2025.09.02.673639doi: bioRxiv preprint 21 such as Pseudomonas tolaasii, which was used as a positive control (Godfrey et al., 2001). The 459 majority of the selected strains exhibited limited antagonism to A. bisporus and did not cause 460 significant browning and collapse of sporocarp tissue , making them promising candidates as 461 biocontrol agents (Figure 2F and Figure S1). 462 Mushroom c asing isolates of Bacillus velezensis are closely related to the commercial 463 biocontrol strain B. velezensis QST713 464 Further phylogenetic analysis was carried out for five B. velezensis strains isolated from casing 465 alongside the commercially used ‘Serenade’ B. velezensis strain QST713 using primers for the 466 housekeeping genes rpoB, tuf, rpoD, pgk and gyrB (Table S4 ). The nucleotide sequences 467 obtained were used to build a phylogenetic tree of the isolated B. velezensis strains alongside 468 other Bacillus spp. for which sequence data is available in the NCBI database. The results 469 showed a clear delineation into three clades: B. amyloliquefaciens, B. velezensis and B. subtilis 470 Figure 1B, which is consistent with previous reports (Fan et al., 2017; Pandin et al., 2018). 471 Four of the strains (CM5, CM19, CM35 and EM5) , were found to be within the B. velezensis 472 clade and to be very closely related to each other and to the commercial biocontrol agent 473 QST713. 474 Five B. velezensis casing isolates (CM5, CM19, CM35, EM5 and EM39) were selected for 475 genome sequencing, which confirmed that these four strains had very high levels of homology 476 compared to QST713, returning a percentage identity of 99.99% when analysed using NCBI 477 PGAAP (Table 2). A full breakdown of all genes containing SNPs compared to QST713 and 478 how these affect the relevant amino acid sequences can be found in Table S5. 479 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 2, 2025. ; https://doi.org/10.1101/2025.09.02.673639doi: bioRxiv preprint 22 Fengycins are the most abundant anti-microbial lipopeptides produced by 480 casing isolates of Bacillus velezensis in vitro. 481 We used the prediction software antiSMASH to identify potential gene clusters involved in 482 specialised metabolite biosynthesis that could be involved in the antagonistic behaviour 483 displayed by the casing isolates of B. velezensis (Figure 2). This analysis identified multiple 484 gene clusters (Table 3) with homology to clusters previously shown to encode the enzymatic 485 machinery for synthesis of the nonribosomal peptides surfactin , fengycin, bacillibactin, and 486 bacilysin and the antimicrobial polyketides bacillaene, difficidin, and macrolactin. Surfactin 487 and fengycin are lipopeptides known to have both antimicrobial and surfactant activity (Sur et 488 al., 2018; Gilliard et al., 2024) while bacillibactin is an iron-chelating siderophore (Qin et al., 489 2019). 490 Surfactant production by the selected B. velezensis strains was evaluated using a droplet 491 diameter assay (Akpa et al., 2001). All B. velezensis strains had significantly different droplet 492 diameters compared to the LB-only control (A), confirming that these strains produce factors 493 both with antimicrobial and surfactant activity. 494 The identity of the specialised metabolites produced by CM5, CM19 and CM35 was further 495 investigated using bioinformatics- and literature-guided targeted metabolic profiling by high -496 performance liquid chromatography-high-resolution mass spectrometry (HPLC-HRMS). 497 Targeted metabolic profiling revealed production of fengycin (1xCO,1xOH,Ala:R1=13C) 498 (Figure 4) by all three bacterial strains under the cultivation conditions tested. Fengycins A and 499 B differ in the amino acid alanine (A) or valine (B) at the sixth position. Identification of the 500 fengycin was achieved by comparison with an authentic reference standard and referencing the 501 mass spectral fragmentation pathway with literature (Wang et al., 2004; Ma et al., 2016). Other 502 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 2, 2025. ; https://doi.org/10.1101/2025.09.02.673639doi: bioRxiv preprint 23 target compounds from antiSMASH prediction (see Table S9), including macrolactins , were 503 not produced under the tested conditions (i.e., < limit of detection (LOD)). 504 The application of Bacillus velezensis strains to mushroom crops was 505 ineffective in limiting disease caused by the dry bubble parasite 506 Lecanicillium fungicola. 507 Three B. velezensis strain (CM5, CM19, CM35) were taken forward for crop trials. Application 508 doses were selected based on preliminary results observed using a small growth chamber and 509 the application dose approved for the commercial product Serenade ASO® (Bacillus velezensis 510 QST 713) in Spain: 5.4 x 108 cfu/100 kg compost against Trichoderma aggressivum (MAPA, 511 2025). The crop trials were designed to assess (i) effects of the treatment on the biological 512 efficiency of the crop, and (ii) biocontrol effects against the fungal parasite L. fungicola. 513 Disease was successfully established following inoculation with LF and statistically significant 514 differences between control groups (+LF and –LF) were detected (Figure 5A), confirming that 515 artificial inoculation with LF affected yield. The control inoculated with LF ( Figure 5A) was 516 the least productive in terms of biological efficiency, and treatment with a single dose of CM19 517 was the most productive. However, differences between control and B. velezensis treatments 518 were not found to be significant (p > 0.05) in the first flush, second flush or total cropping cycle 519 (Figure 5A). As expected, the use of prochloraz-Mn (PCL) was found to significantly reduce 520 disease in both the first and second flush. 521 Phospholipid fatty acid analysis (PLFA) 522 To further examine whether supplementation with B. velezensis affected the mushroom casing 523 microbiota, phospholipid fatty acid (PLFA) analysis was carried out to evaluate the relationship 524 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 2, 2025. ; https://doi.org/10.1101/2025.09.02.673639doi: bioRxiv preprint 24 between microbial biomass and community structure in the commercial casing substrate during 525 the crop trial at different stages of mushroom cropping . PLFA quantifies total microbial 526 biomass and specific microbial groups ( gram-positive bacteria, gram-negative bacteria, 527 Actinomycetes, and total fungi) to provide a general overview of microbial structure and 528 dynamics (52) (Table S8). 529 The total microbial biomass in the casing soil increased significantly over time at the different 530 stages of the crop. Specifically, microbial PLFAs in the casing soil were 3.4, 5.8 and 9.5 times 531 higher in the control treatment at the stages of mycelia penetration in the casing layer (initial 532 cropping stage), first flush primordia formation (start of fructification) , and first flush 533 mushroom growth (end of harvest) , respectively, compared to the casing soil (raw material) 534 (Figure 6A). 535 The proportion of PFLAs associated with bacterial populations in the casing soil decreased 536 over time while the proportion of fungal PLFAs increased. Fungal PLFAs were relatively more 537 abundant than bacterial PLFAs by the end of the first flush (day 29). Gram-negative and gram-538 positive bacterial fatty acids were higher during the initial cropping stage compared to the other 539 stages (Figure 6B,C). However, the relative abundance of gram-negative bacterial PLFAs was 540 higher than gram -positive ones a cross all cropping stages . Actinomycete populations were 541 significantly lower compared to the other bacterial groups (gram-negative and gram-positive 542 bacteria) during all stages of mushroom growth (Figure 6D). 543 The casing soils treated with B. velezensis strains showed significant differences in microbial 544 biomass and composition within the four microbial groups analysed and across the different 545 cropping stages (Figure 6E). The highest relative abundance of bacterial PLFAs at the flushing 546 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 2, 2025. ; https://doi.org/10.1101/2025.09.02.673639doi: bioRxiv preprint 25 stage was observed in the control treatment supplemented with L. fungicola (+), while the 547 highest relative abundance of fungal PLFAs was observed in treatment 5 (strain CM35). 548 TaqMan assays for B. velezensis indicate low persistence of biostimulants 549 following supplementation to mushroom crops 550 To investigate whether the lack of a positive biocontrol effect for B. velezensis in crop could 551 be linked to low persistence of the biostimulants in mushroom casing we developed a specific 552 TaqMan assay to track the population density of B. velezensis when supplemented to casing 553 soil. The whole genome sequence of B. velezensis strains CM19, CM5 and CM35 are almost 554 identical and only showed a limited number of SNPs (Table S5). It was technically impossible 555 to differentiate the strains in a TaqMan assay based on a single SNP , and thus we decided to 556 design an assay for a subgroup of this species that includes the strains CM5, CM19 and CM35, 557 together with other closely related B. velezensis strains. Two B. velezensis subgroup-specific 558 assays, targeting different genes, were combined into a triplex assay with a specific assay for 559 Xanthomonas ca mpestris pv. campestris, which was used as extraction and amplification 560 control (Table S7). 561 The two B. velezensis subgroup-specific assays in the triplex assay gave a positive result with 562 the three target strains and with CM18, CM20, CM21 and EM5, which all clustered in the same 563 subgroup (Figure S2 B; Table S14 ), but not with other B. velezensis strains outside this 564 subgroup, such as B. velezensis EM39. 565 Despite showing high specificity, the assay showed low sensitivity , giving a positive result 566 only for cells present at 104 cfu/g of casing soil when used to detect cells present in exponential 567 cultures, and 10 7 cfu/g of soil for samples from stationary cultures (Figure 7C), potentially 568 indicating low efficiency of DNA extraction from spores. When used to detect B. velezensis in 569 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 2, 2025. ; https://doi.org/10.1101/2025.09.02.673639doi: bioRxiv preprint 26 casing samples, the bacteria could only be detected on the day of applicati on (days 1 and 18 570 after casing; Figure 7A, B), suggesting low persistence and/or conversion to spores following 571 supplementation. 572

573 The horticultural activity of button mushroom cultivation include s different agronomic 574 operations: a) production of a selective substrate through a process of composting; b) 575 inoculation with the selected A. bisporus strain; c) casing of colonised substrate, commonly 576 with non-sterile, peat-based material; d) fruiting and harvest (Carrasco et al., 2021). The casing 577 layer is a microbe-rich environment where mushroom fructification occurs, driven by the action 578 of beneficial bacteria (Braat et al., 2022), and is also considered to be the main infection site 579 for mycoparasites, the most damaging biotic disorders of mushroom crops (Carrasco et al., 580 2021; Gea et al., 2021). The microbiota present in casing soil therefore plays a key role in both 581 promotion of mushroom fructification and disease development or disease suppression. The 582 aim of this work was to isolate and characterize bacteria that are naturally resident in casing 583 soil or live in association with mushroom basidiomes, to show antifungal activity of these 584 bacteria towards economically important mycoparasites; to investigate the mechanistic basis 585 of this antifungal activity, and to investigate the potential of these bacteria as biocontrol agents 586 in crop. 587 Several strains of B. velezensis, including CM5, CM19 and CM35 (included in a published 588 patent (Carrasco and Preston, 2023 )), isolated from commercial casing material, and B. 589 velezensis EM5 and EM39 , isolated from mushroom basidiomes, have been identified and 590 demonstrated to possess antifungal activity towards four of the most damaging mycoparasites 591 (C. mycophilum, L. fungicola, M. perniciosa and T. aggressivum ). The antifungal activity 592 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 2, 2025. ; https://doi.org/10.1101/2025.09.02.673639doi: bioRxiv preprint 27 observed in our strains is consistent with the potential biocontrol activity reported in other 593 strains, including B. velezensis Kos isolated from mushroom casing environment at Teagasc, 594 Ireland (Kosanović et al., 2021; Clarke et al., 2022a, 2022b ) and the commercial biocontrol 595 agent B. velezensis QST713, isolated from soil in a peach tree orchard in Fresno County, 596 California (USA) (Anastassiadou et al., 2021). 597 Importantly, B. velezensis CM5, CM19 and CM35 showed limited toxicity to A. bisporus in 598 vitro and did not show a deleterious effect on the mushroom mycelium at the dose applied in 599 crop (109 cfu m-2). However, compared to an established chemical treatment (PCL), none of 600 the three strains provided effective control of the dry bubble parasite L. fungicola in a crop trial 601 (Figure 5B). According to previous published results, the commercial formulation s of B. 602 velezensis strain QST 713 (Serenade®, Bayer Crop Science, Valencia, Spain) and Bacillus 603 amyloliquefaciens subsp. plantarum strain D747 (Amylo-X®, Certis Europe, Alicante, Spain) 604 also offered poor control of wet bubble disease in crop trials, with their effectiveness further 605 reduced under high pressure of parasite inoculum (Navarro et al. , 2023 ). Similarly, B. 606 velezensis QST 713 did not control cobweb disease in a crop trial artificially infected by C. 607 mycophilum isolate 1546 while B. velezensis Kos 96h culture filtrate reached 30–40% efficacy 608 (Clarke et al., 2024). This indicates that parasite identity, inoculum density and disease pressure 609 may all affect the outcome of control treatments, and that beneficial effects may largely due to 610 antimicrobial compounds present in the culture (or culture filtrate) at the time of application . 611 Our trial used a high inoculum density of L. fungicola compared to the natural parasite pressure 612 in commercial facilities . It is possible that at a lower inoculum density, a higher dose of B. 613 velezensis applied to the crop , or by optimising culture conditions and inoculum design to 614 increase production of antimicrobial compounds, both prior to and upon application to the crop, 615 a better disease control will be achieved. However, in pilot assays, at the highest doses used (> 616 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 2, 2025. ; https://doi.org/10.1101/2025.09.02.673639doi: bioRxiv preprint 28 1010 cfu/m2), B. velezensis was observed to have a potential deleterious effect on A. bisporus 617 colonisation of casing soil (our unpublished results), and therefore it is likely be necessary to 618 further optimise the dose and formulation of B. velezensis applied to the crop to achieve optimal 619 results. 620 Selective TaqMan assays were developed to study the population dynamics of B. velezensis in 621 crop. The availability of two TaqMan assays per target organism in combination with an assay 622 for extraction and amplification allows specific and reliable detection of the targets in multiplex 623 set-up. However, the LOD was ≤ 4 × 105 cells per mL in casing soils, which only allows studies 624 of the population dynamics in situations in which relatively high densities of the B. velezensis 625 are supplemented to or persist in the casings. Additionally, we observed low efficiency 626 detection of bacteria that had entered stationary phase , potentially linked to challenges with 627 efficient extraction of DNA from spores. 628 When the TaqMan assays were applied to crop trial samples, we found that the target strains 629 could only be reliably detected on the day of application . This suggests that their population 630 declines rapidly following inoculation, or that the bacteria have entered stationary phase and 631 formed spores, in which case the TaqMan assay may significantly under-report their abundance 632 in casing samples. Since B. velezensis must adapt to a complex environment, optimizing the 633 formulation or antimicrobial activity of B. velezensis based products is key to improve stability, 634 shelf-life and deliver biocontrol activity (Kenfaoui et al., 2024).. 635 As previously reported (Carrasco et al., 2019, 2020), we observed that microbiome structure 636 changes along the process of mushroom cultivation. PLFA analyses confirmed that the 637 cultivation stage of A. bisporus correlated with significant changes in the composition of the 638 microbial communities in the casing, where the bacterial population, particularly gram-639 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 2, 2025. ; https://doi.org/10.1101/2025.09.02.673639doi: bioRxiv preprint 29 negative bacteria, have been reported to play a fundamental role during the early development 640 of the mushroom primordia. The highest microbial biomass was found in the casing soil at the 641 end of fructification , indicating changes in microbial structure associated with mushroom 642 mycelium growth. The increase in the microbial population, with respect to both bacterial and 643 fungal biomass, correlated with the mycelial growth of A. bisporus , in accordance with 644 previous results (Chen et al. , 2009; Carrasco et al ., 2020 ). The increase in the bacterial 645 population may be attribut able to metabolites released during the development of fruiting 646 bodies, which alter nutrient composition and influence the diversity of microbial communities 647 in mushroom casing (Zhao et al., 2022). 648 Interestingly, application of B. velezensis strains in crop had a positive effect on both the gram-649 negative and gram -positive communities suggesting that in addition to directly inhibiting 650 mycoparasitic fungi, it could have a synergistic effect on mushroom growth by promoting the 651 proliferation of disease-suppressing or fructification -promoting bacteria, perhaps through an 652 anti-competitor effect or by releasing nutrients through lysis of susceptible organisms . The 653 higher amount of gram negative compared to gram positive bacteria in all cropping stages is 654 consistent with the hypothesis that gram negative bacteria play a key role in stimulating the 655 initiation of mushroom sporophore s (Chen et al., 2009). Gram-negative bacteria, particularly 656 members of the Pseudomonadales such as Pseudomonas putida have been identified as major 657 microbial stimulants for sporophore formation in the casing layer, potentially by metabolizing 658 volatile compounds in the casing soil that would otherwise inhibit sporophore initiation, 659 thereby creating conditions conducive to sporophore formation (Rainey, 1989; Noble et al., 660 2003). 661 Actinomycetes are sensitive to environmental changes, such as shifts in temperature, pH, and 662 moisture levels, which occur during mushroom cultivation (Lacey, 1997; Akond et al., 2016). 663 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 2, 2025. ; https://doi.org/10.1101/2025.09.02.673639doi: bioRxiv preprint 30 These fluctuations c ould adversely affect their survival and proliferation, leading to lower 664 populations compared to more resilient bacterial groups. Besides, the casing layer is rich in 665 microbial diversity, leading to competition for nutrients and space. Actinomycetes may be 666 outcompeted by faster-growing bacteria, such as Pseudomonas spp., resulting in their reduced 667 presence during mushroom growth stages (Rainey, 1989; Noble et al., 2003). 668 B. velezensis CM5, CM19, CM35, EM5 and EM39 all showed antimicrobial activity against 669 mycoparasites causing cobweb (C. mycophilum), dry bubble (L. fungicola), wet bubble (M. 670 perniciosa) and green mould disease (T. aggressivum) in vitro. Similarly, Clarke et al. (Clarke 671 et al., 2022a, 2022b) have reported that B. velezensis Kos has antimicrobial activity against C. 672 mycophilum and L. fungicola . Genome sequencing identified several gene clusters in these 673 strains that could direct the production of a range of non -ribosomal peptide and polyketide 674 antifungal compounds. However, analyses of the specialised metabolites produced in vitro 675 identified lipopeptides known as fengycins as the major non-ribosomal peptides produced by 676 these strains in pure culture under the conditions tested. It therefore seems likely that fengycins 677 are the primary factors responsible for both the anti-fungal activity and the surfactant activity 678 observed for these bacteria in vitro. However, we cannot exclude the possibility that some of 679 the anti-fungal compounds produced by these bacteria are only produced in the presence of 680 other microorganisms, or that these bacteria express a different arsenal of anti-microbial factors 681 during colonisation of mushroom casing or sporocarps. Previous studies have shown that many 682 gene clusters encoding anti -microbial factors remain silent in standard culture conditions . 683 These cryptic biosynthetic genes may only be induced during co -culture or specific 684 environmental conditions (Ochi , 2017). 685 The chemical structure of the fengycin analogues produced has been determined for CM5, 686 CM19 and CM35 when cultured in vitro. Interestingly, despite the high degree of similarity 687 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 2, 2025. ; https://doi.org/10.1101/2025.09.02.673639doi: bioRxiv preprint 31 between these strains at a genomic level CM19 showed higher production of the fengycin 688 analogue in vitro, suggesting that epigenetic differences or one or more of the SNPs detected 689 (Table S16) could be responsible for this differential production. 690 Fengycin is known to exert its anti-fungal activity by assembling into pore-forming complexes 691 that target fungal cell membranes (Zakharova et al., 2019) in an interaction that depends on the 692 lipid composition of these membranes. The observation that fengycins are the major anti-fungal 693 factor produced by these strains in vitro and that these strains exhibit greater a ntagonism 694 towards the four mycoparasitic fungi, all of which belong to the order Hypocreales (Division: 695 Ascomycota) compared to A. bisporus (order Agaricales, Division Basidiomycota), raises the 696 prospect that this differential toxicity could rel ate to intrinsic differences in membrane 697 properties between fungi within these two clades . This suggests that biocontrol agents 698 specifically producing fengycins and selected or engineered to not produce broadly active anti-699 fungal compounds could prove particularly effective as biocontrol agents for mushroom 700 cultivation. This also raises the exciting prospect that fengycins or other lipid-specific, 701 membrane-targeting compounds could be further developed as selective fungicides capable of 702 targeting specific groups of fungi. Research is already underway to develop approaches to 703 produce fengycin through fermentation on an industrial scale (Yin et al., 2024). Additional 704 research could be carried out to determine which of the different analogues of fengycin 705 produced by B. velezensis and related bacteria display the greatest activity and specificity in 706 order to focus production on these metabolites. 707 Although fengycin remains the most likely candidate as the causal agent of the antimicrobial 708 activity observed in vitro, we have not yet been able to validate this hypothesis by knock-out 709 mutagenesis of the fengycin biosynthetic cluster , as the strains described in this study have 710 proven to be recalcitrant to transformation. However, the very low efficiency , but successful 711 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 2, 2025. ; https://doi.org/10.1101/2025.09.02.673639doi: bioRxiv preprint 32 transformation of CM19 with pHAPII gfp suggests that with further optimisation of 712 transformation protocols this might be possible. Additionally, we have not yet been able to 713 detect expression of fengycin biosynthetic genes or the production of fengycin when these 714 strains are inoculated into casing (data not shown) , consistent with the rapid decline in 715 detectable bacteria and the lack of suppression of L. fungicola observed in crop trials . 716 Therefore, in order to fully realise the potential of fengycin-producing bacteria as biocontrol 717 agents it will be necessary to focus on the development of fengycins as selective fungicides or 718 further optimise the formulation and properties of these strains in the crop environment to 719 maximise survival and fengycin production. 720

721 In this study we set out to investigate whether endogenous casing bacteria have the potential to 722 be effective biocontrol agents against mycoparasitic fungi . B. velezensis strains isolated from 723 casing and basidiomes were demonstrated to display a high level of antifungal activity against 724 major fungal parasites, including L. fungicola, the causal agent of dry bubble. We identified a 725 fengycin analogue as the compound most likely to be responsible for the anti -fungal activity 726 exhibited by these strains in vitro. However, these strains have not been demonstrated to be 727 capable of suppressing the development of dry bubble when inoculated onto mushroom crops 728 that have been artificially infected with this parasite. The lack of antifungal activity observed 729 in crop could be due to a number of factors : the doses of mycoparasite and biocontrol agent 730 used, the level of fengycin present in or produced by the bacterial treatments applied to the 731 crop and the limited persistence of these strains in the casing material following inoculation. 732 In this sense, further optimisation of the bacteria formulation in relation to the composition and 733 dynamics of the microbiota that could compete for space or nutrients in the casing could be the 734 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 2, 2025. ; https://doi.org/10.1101/2025.09.02.673639doi: bioRxiv preprint 33 object of future research. Nevertheless, the specificity shown by these bacteria towards 735 mycoparasitic fungi over their mushroom host raises exciting possibilities to develop selective 736 biocontrol of mycoparasitic fungi. Understanding the composition and dynamics of the 737 microbiota in the casing during commercial cultivation of the fungus A. bisporus could be 738 essential to design novel biocontrol agents against mushroom mycoparasites. 739 Tables 740 Table 1: Experimental blocks used in crop trial 741 Treatment Code Dose Infected (Lecanicillium fungicola 150/1) Control Control+ 20 ml Tap water 106 conidia m-2 (10 days after casing) Control Control- 20 ml Tap water - Bacillus velezensis CM5 CM5 109 cfu* m-2 106 conidia m-2 (10 days after casing) Bacillus velezensis CM19 CM19 109 cfu m-2 106 conidia m-2 (10 days after casing) Bacillus velezensis CM35 CM35 109 cfu m-2 106 conidia m-2 (10 days after casing) Prochloraz-Mn PCL 1 g m-2 106 conidia m-2 (10 days after casing) *cfu: colony forming unit. 742 743 Table 2: Summary of genomic data for B. velezensis casing isolates compared to QST713 (BLASTn (Nucleotide BLAST); BLASTp (Protein-Protein BLAST)). QST713 CM5 CM19 CM35 EM5 EM39 Similarity to 1QST713 (%) 100 99.99 99.99 99.99 99.99 98.38 Genes (total) 4,238 4,256 4,255 4,256 4,256 3,906 GC Content 46 45.84 46 45.85 45.84 46.43 Genome Size 4233757 4240819 4240819 4240818 4240818 3929792 2CDSs (total) 4,047 4,137 4,136 4,137 4,137 3,788 Genes (coding) 4,047 4,049 4,048 4,048 4,048 3,686 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 2, 2025. ; https://doi.org/10.1101/2025.09.02.673639doi: bioRxiv preprint 34 CDSs (with protein) 4,047 4,049 4,048 4,048 4,048 3,686 Genes (RNA) 107 119 119 119 119 118 3rRNAs (5S, 16S, 23S) 9, 8, 8 10, 9, 9 10, 9, 9 10, 9, 9 10, 9, 9 9, 9, 9 Complete rRNAs (5S, 16S, 23S) 9, 8, 8 10, 9, 9 10, 9, 9 10, 9, 9 10, 9, 9 9, 9, 9 4tRNAs 77 86 86 86 86 86 5ncRNAs 5 5 5 5 5 5 Pseudo Genes (total) 84 88 88 89 89 102 CDSs (without protein) 84 88 88 89 89 102 Pseudo Genes (ambiguous residues) 0 of 84 0 of 88 0 of 88 0 of 89 0 of 89 0 of 102 Pseudo Genes (frameshifted) 48 of 84 51 of 88 51 of 88 52 of 89 52 of 89 61 of 102 Pseudo Genes (incomplete) 58 of 84 61 of 88 61 of 88 61 of 89 61 of 89 65 of 102 Pseudo Genes (internal stop) 14 of 84 11 of 88 11 of 88 11 of 89 11 of 89 9 of 102 Pseudo Genes (multiple problems) 30 of 84 29 of 88 29 of 88 29 of 89 29 of 89 30 of 102 1Results for all strains are from analyses using NCBI PGAAP (Tatusova et al., 2016). Quality Assessment of 744 Bacillus genomes using QUAST can be found in Table S 1; 2CDS: Coding sequences; 3rRNA: ribosomal RNA; 745 4tRNA: transfer RNA; 5ncRNA: non-coding RNA. 746 747 Table 3: Comparative AntiSMASH analysis of gene clusters predicted to encode genes involved in 748 specialised metabolite synthesis between Bacillus velezensis CM5, CM19, CM35, EM5, EM39 and 749 QST713. 750 Most similar cluster Type QST713 CM5 CM19 CM35 EM5 EM39 bacillaene Polyketide+1NRP 100% 100% 100% 100% 100% 100% bacillibactin NRP 100% 100% 100% 100% 100% 100% bacilysin Other 100% 100% 100% 100% 100% 100% butirosin A/B Saccharide 7% 7% 7% 7% 7% 7% difficidin Polyketide 100% 100% 100% 100% 100% 100% fengycin NRP 100% 100% 100% 100% 100% 100% locillomycin NRP+Polyketide 28% 28% 28% 28% 28% - macrolactin H Polyketide 100% 100% 100% 100% 100% 100% subtilin 2RiPP: Lanthipeptide 100% 100% 100% 100% 100% - surfactin NRP:Lipopeptide 82% 82% 82% 82% 82% 82% .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 2, 2025. ; https://doi.org/10.1101/2025.09.02.673639doi: bioRxiv preprint 35 1NRP: Non -Ribosomal Peptides ( built by non -ribosomal peptide synthetases (NRPSs) ); 2RiPPs: Ribosomally 751 synthesized and Post-translationally modified Peptides. 752 Figures 753 754 Figure 1: A) Cultivable bacterial isolates from three types of casing layer and endofungal bacteria from 755 basidiomes of Agaricus bisporus identified using 16S rDNA sequencing ; B) Phylogenetic tree of 756 isolated Bacillus strains. The phylogeny was built using concatenated genomes comprising the five 757 housekeeping genes ( gyrB, pgk, rpoB, rpoD, and tuf) aligned to the reference strain B. velezensis 758 FZB42. Strains highlighted in bold were isolated and phenotypically characterised in this study; C) 759 Visualization of Bacillus genomes using circular map and BLAST comparison using B LAST Ring 760 Generator (BRIG) (42); Putative specialised metabolite biosynthetic clusters compared to QST713 are 761 shown by red marks in the outer ring. 762 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 2, 2025. ; https://doi.org/10.1101/2025.09.02.673639doi: bioRxiv preprint 36 763 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 2, 2025. ; https://doi.org/10.1101/2025.09.02.673639doi: bioRxiv preprint 37 Figure 2: A) Representative images of halos formed during antagonism between B. velezensis CM19 764 and the four fungal parasites of the cultivated button mushroom Agaricus bisporus in confrontation 765 assays: L. fungicola (150/1); T. aggressivum (TAV1); M. perniciosa (M25); and C. mycophilum 766 (CM13900). B) Graph shows the mean and standard error of zones of inhibition (mm) of the three 767 biological replicates of the antagonism assay. Letters indicate groups of statistically different conditions 768 (one-way ANOVA and Tukey’s HSD test α = 0.05). C & D) Representative images of unchallenged 769 fungal parasites grown in vitro on PDA plates – images taken using confocal microscope at t=0 (C), 770 and t=18 h (D). E) Representative confocal microscope images of fungal parasites in vitro in the 771 presence of B. velezensis CM19 transformed with the plasmid pHAPII gfp (green). Conidial and hyphal 772 viability and morphology were assessed in vitro using PDA plates. Live dead staining with propidium 773 iodide (PI; pink) was used to study the effect of CM19 on spore survival and germination (C, D, E). 774 Spores were classified as dead if either fully stained with PI or hollow/lysed cells. In many cases spores 775 were deemed to have germinated but resulting hyphae were also found to be stained or lysed. F) 776 Representative images of mushroom sporocarp cubes 24h after inoculation with LB, B. velezensis 777 CM35, B. velezensis EM39, and P. tolaasii. Graph of results can be viewed in Figure S1. 778 779 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 2, 2025. ; https://doi.org/10.1101/2025.09.02.673639doi: bioRxiv preprint 38 780 781 Figure 3: Surfactant production in B. velezensis strains. Data shown is the mean droplet diameter of 782 supernatant from overnight LB cultures . Droplet diameters were measured 2 minutes after pipetting 783 onto microscope slides. Graph shows the mean and standard error of 5 experiments, each containing 5 784 droplets per treatment. Letters indicate groups of statistically different conditions (one -way ANOVA 785 and Tukey’s HSD test α = 0.05). 786 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 2, 2025. ; https://doi.org/10.1101/2025.09.02.673639doi: bioRxiv preprint 39 787 Figure 4: HPLC-HRMS analysis of fengycin produced by B. velezensis strains CM5, CM19 and CM35 788 cultivated on LB agar. (A) Extracted ion chromatograms ( [M+H]+, m/z 1463.804, ±2 ppm) show the 789 production of fengycin by all the three strains. The highest physiological amounts of the compound 790 were produced by B. velezensis CM19, followed by strains CM5 and CM35. The left insert shows the 791 phenotypic characteristics of the strains grown on LB agar. The chemical structure of fengycin is shown 792 on the right. (B) HRMS spectrum of fengycin produced by the bacteria. 793 794 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 2, 2025. ; https://doi.org/10.1101/2025.09.02.673639doi: bioRxiv preprint 40 795 796 Figure 5: Crop trial of B. velezensis biocontrol activity against Lecanicillium fungicola 150/1. Blocks 797 were artificially infected with Lecanicillium fungicola 150/1 and treated with CM5, CM19 and CM35, 798 using prochloraz -Mn (PCL) as a positive control treatment and water as a negative control : A) 799 Biological efficiency (BE) of the crop ( Kg of mushroom per 100 Kg of compost in dry weight) 800 recovered from the different treatments at flush 1 (F1), flush 2 (F2) and for the total cropping cycle; B) 801 Diseased mushrooms (%) collected from the different treatments at F1, F2 and for the total cropping 802 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 2, 2025. ; https://doi.org/10.1101/2025.09.02.673639doi: bioRxiv preprint 41 cycle. Graphs show the mean and standard error of three biological replicates. Letters indicate groups 803 of statistically different conditions (one-way ANOVA and Tukey’s HSD test α = 0.05). 804 805 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 2, 2025. ; https://doi.org/10.1101/2025.09.02.673639doi: bioRxiv preprint 42 806 Figure 6: A: Quantification of casing microbial biomass by analysis of PLFAs . Four timepoints were 807 assessed: d0 (Beginning of trial; grey), day 10 (Fully colonized casing, blue), d21 (Fructification of first 808 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 2, 2025. ; https://doi.org/10.1101/2025.09.02.673639doi: bioRxiv preprint 43 flush, green), and d29 (End of first flush, orange). Treatment 1: Control +LF (+); Treatment 2: Control 809 -LF (-); Treatment 3: CM5 +LF; Treatment 4: CM19 +LF; Treatment 5: CM35 +LF; Treatment 6: PCL 810 +LF). Letters in uppercase highlight differences between treatments at the same time point (Tukey Test, 811 p > 0.05) . Error bars show SE. B: Abundance of Gram -positive bacteria. C: Abundance of Gram -812 negative bacteria. D: Abundance of Actinomycetes. E: Abundance of fungal PLFAs. All data shown 813 uses nmol g-1. 814 815 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 2, 2025. ; https://doi.org/10.1101/2025.09.02.673639doi: bioRxiv preprint 44 816 Figure 7: Population dynamics of Bacillus velezensis CM19 following application to mushroom casing 817 as determined by a quantitative TaqMan assay. T0 = after cac-ing (day 0), T1 = directly after application 818 of B. velezensis CM19 (day 3), T2 = End of spawn run (Day 9), T3 = Beginning of the 1st flush (Day 819 14), T4 = End of the 1st flush (Day 17), T5 = Directly after the second application of B.velezensis CM19 820 (Day 18), T6 = Beginning of the 2nd flush (Day 22), T7 = End of the 2nd flush (Day 25). A: Results 821 using the HP2.21 (FAM) TaqMan probe; B) Results using the HP2.19 (Cy5) TaqMan probe C ) 822 Electrophoresis gel showing PCR results from PCR reaction using HP2.19 TaqMan primers (35 cycles) 823 for DNA extracted from B. velezensis CM19 cells (from LB medium) in exponential growth stage (6h) 824 and stationary cultures (24 h). Numbers indicate the number of bacteria added to the casing soil before 825 extraction (cells added /g of soil). 826 827 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 2, 2025. ; https://doi.org/10.1101/2025.09.02.673639doi: bioRxiv preprint 45 Author statements 828 Authors 829 • William Kay; [email protected] ; orcid.org/0000-0002-2561-4856 830 • Jaime Carrasco; [email protected]; orcid.org/0000-0002-2051-1823 , 831 • Gail Preston; [email protected] ; orcid.org/0000-0003-3882-4438 832 • Souvik Kusari: [email protected]; orcid.org/0000-0002-4685-0794 833 • Marjon Krijger; [email protected] 834 • M. José Carpio; [email protected]; orcid.org/0000-0003-1752-9608 835 • Thomas Barnes; [email protected] 836 • M. Sonia Rodríguez Cruz; [email protected]; orcid.org/0000-0001-6748-837 3391 838 • Jan van der Woolf; [email protected]; orcid.org/0000-0002-9196-5419 839 • Till Bebenroth; [email protected] 840 Author Contributions 841 • Conceptualization: JC WK SK GP 842 • Data curation: WK JC SK MK MJC TB 843 • Formal analysis: JC MLT MT AT GP 844 • Funding acquisition: JC SK MSRC JVW GP 845 • Investigation: WK JC SK MK MJC TB MSRC JVW TB GP 846 • Methodology: WK JC SK MK MJC 847 • Project administration: JC SK MSRC JVW GP 848 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 2, 2025. ; https://doi.org/10.1101/2025.09.02.673639doi: bioRxiv preprint 46 • Resources: WK JC SK MK MJC TB MSRC JVW TB GP 849 • Software: WK JC SK MK MJC TB MSRC JVW TB GP 850 • Supervision: JC SK MSRC JVW GP 851 • Validation: WK JC SK MK MJC TB MSRC JVW TB GP 852 • Visualization: WK JC GP 853 • Writing – original draft: WK JC GP 854 • Writing – review & editing: WK JC SK MK MJC TB MSRC JVW TB GP 855 Conflicts of interest 856 There are no conflicts of interests. All the authors have read and approved the manuscript and 857 all are aware of its submission to the Journal. The paper has not been submitted in any other 858 journal. 859 Funding information 860 The project leading to this report has received funding from the European Union’s Horizon 861 2020 research and innovation programme GA: 101000651 (BIOSCHAMP) and the Marie 862 Sklodowska-Curie IF GA: 742966 (MYCOBIOME). JC is the recipient of a Ramon y Cajal 863 contract [RYC2021 -032796-I], funded by MCIN/AEI/10.13039/ 501100011033 and the 864 European Union “NextGenerationEU”/PRTR”. 865 Ethical approval 866 No ethics statement required since the work has been accomplished with commercial 867 samples and wild strains from fungal specimens. 868 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 2, 2025. ; https://doi.org/10.1101/2025.09.02.673639doi: bioRxiv preprint 47

869 B. velezensis CM5, CM19 and CM35 are protected as biocontrol agents against mushroom 870 parasites in a patent published as EP4200400A1; WO2021255181A1PCT. 2023/06/28. The 871 project leading to this report has received funding from the European Union’s Horizon 2020 872 research and innovation programme GA: 101000651 (BIOSCHAMP) and the Marie 873 Sklodowska-Curie IF GA: 742966 (MYCOBIOME). JC is the recipient of a Ramon y Cajal 874 contract [RYC2021 -032796-I], funded by MCIN/AEI/10.13039/ 501100011033 and the 875 European Union “NextGenerationEU”/PRTR”. 876

877 1. Wang Q, Juan J, Xiao T, Zhang J, Chen H, Song X, et al. The physical structure of 878 compost and C and N utilization during composting and mushroom growth in Agaricus 879 bisporus cultivation with rice, wheat, and reed straw -based composts. Appl Microbiol 880 Biotechnol. 2021;105:3811-23. 881 2. Dias ES, Zied DC, Pardo-Gimenez A. Revisiting the casing layer: Casing materials and 882 management in Agaricus mushroom cultivation. Ciência e Agrotecnologia. 883 2021;45:e0001R21. 884 3. Fones HN, Bebber DP, Chaloner TM, Kay WT, Steinberg G, Gurr SJ. Threats to global 885 food security from emerging fungal and oomycete crop pathogens. Nature Food. 886 2020;1(6):332-42. 887 4. Gea FJ, Navarro MJ, Santos M, Diánez F, Carrasco J. Control of fungal diseases in 888 mushroom crops while dealing with fungicide resistance: a review. Microorganisms. 889 2021;9(3):585. 890 5. Sonnenberg AS, Baars JJ, Gao W, Visser RG. Developments in breeding of Agaricus 891 bisporus var. bisporus: progress made and technical and legal hurdles to take. Appl Microbiol 892 Biotechnol. 2017;101:1819-29. 893 6. Kosanović D, Potočnik I, Duduk B, Vukojević J, Stajić M, Rekanović E, et al. 894 Trichoderma species on Agaricus bisporus farms in Serbia and their biocontrol. Annals of 895 Applied Biology. 2013;163(2):218-30. 896 7. Commision E. Commission Implementing Regulation (EU) 2019/677 of 29 April 2019 897 concerning the non -renewal of the approval of the active substance chlorothalonil, in 898 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 2, 2025. ; https://doi.org/10.1101/2025.09.02.673639doi: bioRxiv preprint 48 accordance with Regulation (EC) No 1107/2009 of the European Parliament and of the Council 899 concerning the placing of plant protection products on the market, and amending Commission 900 Implementing Regulation (EU) No 40/2011. Official Journal of the European Union. 901 2019;114:15-7. 902 8. Zhang L, Sun C. Fengycins, cyclic lipopeptides from marine Bacillus subtilis strains, 903 kill the plant -pathogenic fungus Magnaporthe grisea by inducing reactive oxygen species 904 production and chromatin condensation. Applied Environmental Microbiology. 2018;84(18). 905 9. Carrasco J, Preston GM. Growing edible mushrooms: a conversation between bacteria 906 and fungi. Environmental Microbiology. 2020;22(3):858-72. 907 10. Rabbee MF, Hwang B-S, Baek K-H. Bacillus velezensis: a beneficial biocontrol agent 908 or facultative phytopathogen for sustainable agriculture. Agronomy. 2023;13(3):840. 909 11. Kenfaoui J, Dutilloy E, Benchlih S, Lahlali R, Ait -Barka E, Esmaeel Q. Bacillus 910 velezensis: a versatile ally in the battle against phytopathogens —insights and prospects. Appl 911 Microbiol Biotechnol. 2024;108(1):439. 912 12. Pandin C, Le Coq D, Deschamps J, Védie R, Rousseau T, Aymerich S, et al. Complete 913 genome sequence of Bacillus velezensis QST713: A biocontrol agent that protects Agaricus 914 bisporus crops against the green mould disease. Journal of Biotechnology. 2018;278:10-9. 915 13. Gao L, Guo J, Fan Y, Ma Z, Lu Z, Zhang C, et al. Module and individual domain 916 deletions of NRPS to produce plipastatin derivatives in Bacillus subtilis . Microbial Cell 917 Factories. 2018;17:1-13. 918 14. Desmyttere H, Deweer C, Muchembled J, Sahmer K, Jacquin J, Coutte F, et al. 919 Antifungal activities of Bacillus subtilis lipopeptides to two Venturia inaequalis strains 920 possessing different tebuconazole sensitivity. Frontiers in Microbiology. 2019;10:2327. 921 15. Zihalirwa Kulimushi P, Argüelles Arias A, Franzil L, Steels S, Ongena M. Stimulation 922 of fengycin-type antifungal lipopeptides in Bacillus amyloliquefaciens in the presence of the 923 maize fungal pathogen Rhizomucor variabilis. Frontiers in Microbiology. 2017;8:850. 924 16. Mantil E, Buznytska I, Daly G, Ianoul A, Avis TJ. Role of lipid composition in the 925 interaction and activity of the antimicrobial compound fengycin with complex membrane 926 models. The Journal of Membrane Biology. 2019;252:627-38. 927 17. Pandin C, Darsonval M, Mayeur C, Le Coq D, Aymerich S, Briandet R. Biofilm 928 formation and synthesis of antimicrobial compounds by the biocontrol agent Bacillus 929 velezensis QST713 in an Agaricus bisporus compost micromodel. Applied and Environmental 930 Microbiology. 2019;85(12). 931 18. Kosanovic D, Dyas M, Grogan H, Kavanagh K. Differential proteomic response of 932 Agaricus bisporus and Trichoderma aggressivum f. europaeum to Bacillus velezensis 933 supernatant. European Journal of Plant Pathology. 2021;160(2):397-409. 934 19. Clarke J, Grogan H, Fitzpatrick D, Kavanagh K. Analysis of the effect of Bacillus 935 velezensis culture filtrate on the growth and proteome of Cladobotryum mycophilum. Fungal 936 Biology. 2022;126(1):11-9. 937 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 2, 2025. ; https://doi.org/10.1101/2025.09.02.673639doi: bioRxiv preprint 49 20. Clarke J, McGuinness B, Fitzpatrick D, Kavanagh K, Grogan H. Response of the 938 mushroom pathogen Cladobotryum mycophilum to the fungicides prochloraz and metrafenone 939 and two Bacillus -based biological control agents in mushroom crop trials. Crop Protection. 940 2024;177:106530. 941 21. Clarke J, Grogan H, Fitzpatrick D, Kavanagh K. Characterising the proteomic response 942 of mushroom pathogen Lecanicillium fungicola to Bacillus velezensis QST 713 and Kos 943 biocontrol agents. European Journal of Plant Pathology. 2022;163(2):369-79. 944 22. Novikova I, Titova J. Antifungal activity of Industrial Bacillus Strains against 945 Mycogone perniciosa, the causative agent of wet bubble disease in white button mushrooms. 946 Microorganisms. 2023;11(8):2056. 947 23. Steller S, Vollenbroich D, Leenders F, Stein T, Conrad B, Hofemeister J, et al. 948 Structural and functional organization of the fengycin synthetase multienzyme system from 949 Bacillus subtilis b213 and A1/3. Chemistry & Biology. 1999;6(1):31-41. 950 24. Yin Y, Wang X, Zhang P, Wang P, Wen J. Strategies for improving fengycin 951 production: a review. Microbial Cell Factories. 2024;23(1):144. 952 25. Sang-Cheol L, Kim S-H, Park I-H, Chung S-Y, Chandra MS, Yong-Lark C. Isolation, 953 purification, and characterization of novel fengycin S from Bacillus amyloliquefaciens LSC04 954 degrading-crude oil. Biotechnology and Bioprocess Engineering. 2010;15:246-53. 955 26. Villegas-Escobar V, Ceballos I, Mira JJ, Argel LE, Orduz Peralta S, Romero -Tabarez 956 M. Fengycin C produced by Bacillus subtilis EA-CB0015. Journal of Natural Products. 957 2013;76(4):503-9. 958 27. Kalamara M, Spacapan M, Mandic‐Mulec I, Stanley‐Wall NR. Social behaviours by 959 Bacillus subtilis : quorum sensing, kin discrimination and beyond. Mol Microbiol. 960 2018;110(6):863-78. 961 28. Na H -E, Heo S, Kim T, Lee G, Lee J -H, Jeong D -W. ComQXPA quorum -sensing 962 systems contribute to enhancing the protease activity of Bacillus velezensis DMB05 from 963 fermented soybeans. International Journal of Food Microbiology. 2023;401:110294. 964 29. Carrasco J, Tello ML, de Toro M, Tkacz A, Poole P, Pérez -Clavijo M, et al. Casing 965 microbiome dynamics during button mushroom cultivation: implications for dry and wet 966 bubble diseases. Microbiology. 2019;165(6):611-24. 967 30. Aslani MA, Harighi B, Abdollahzadeh J. Screening of endofungal bacteria isolated 968 from wild growing mushrooms as potential biological control agents against brown blotch and 969 internal stipe necrosis diseases of Agaricus bisporus. Biological Control. 2018;119:20-6. 970 31. Stackebrandt E, Goodfellow M. Nucleic acid techniques in bacterial systematics. 971 Wiley, editor. Chichester ; New York: Wiley; 1991. xxix,329p. p. 972 32. Camacho C, Coulouris G, Avagyan V, Ma N, Papadopoulos J, Bealer K, et al. BLAST+: 973 architecture and applications. BMC Bioinformatics. 2009;10:1-9. 974 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 2, 2025. ; https://doi.org/10.1101/2025.09.02.673639doi: bioRxiv preprint 50 33. Bolger AM, Lohse M, Usadel B. Trimmomatic: a flexible trimmer for Illumina 975 sequence data. Bioinformatics. 2014;30(15):2114-20. 976 34. Bankevich A, Nurk S, Antipov D, Gurevich AA, Dvorkin M, Kulikov AS, et al. 977 SPAdes: a new genome assembly algorithm and its applications to single -cell sequencing. 978 Journal of Computational Biology. 2012;19(5):455-77. 979 35. Seemann T. Prokka: rapid prokaryotic genome annotation. Bioinformatics. 980 2014;30(14):2068-9. 981 36. Rutherford K, Parkhill J, Crook J, Horsnell T, Rice P, Rajandream M-A, et al. Artemis: 982 sequence visualization and annotation. Bioinformatics. 2000;16(10):944-5. 983 37. Gurevich A, Saveliev V, Vyahhi N, Tesler G. QUAST: quality assessment tool for 984 genome assemblies. Bioinformatics. 2013;29(8):1072-5. 985 38. Wood DE, Salzberg SL. Kraken: ultrafast metagenomic sequence classification using 986 exact alignments. Genome biology. 2014;15(3):1-12. 987 39. Aziz RK, Bartels D, Best AA, DeJongh M, Disz T, Edwards RA, et al. The RAST 988 Server: rapid annotations using subsystems technology. BMC Genomics. 2008;9(1):1-15. 989 40. Aziz RK, Devoid S, Disz T, Edwards RA, Henry CS, Olsen GJ, et al. SEED servers: 990 high-performance access to the SEED genomes, annotations, and metabolic models. PloS one. 991 2012;7(10):e48053. 992 41. Tatusova T, DiCuccio M, Badretdin A, Chetvernin V, Nawrocki EP, Zaslavsky L, et al. 993 NCBI prokaryotic genome annotation pipeline. Nucleic Acids Research. 2016;44(14):6614-24. 994 42. Alikhan N-F, Petty NK, Ben Zakour NL, Beatson SA. BLAST Ring Image Generator 995 (BRIG): simple prokaryote genome comparisons. BMC Genomics. 2011;12(1):402. 996 43. Blin K, Shaw S, Kloosterman AM, Charlop -Powers Z, Van Wezel GP, Medema MH, 997 et al. antiSMASH 6.0: improving cluster detection and comparison capabilities. Nucleic Acids 998 Research. 2021;49(W1):W29-W35. 999 44. Edgar RC. MUSCLE: multiple sequence alignment with high accuracy and high 1000 throughput. Nucleic Acids Research. 2004;32(5):1792-7. 1001 45. Nguyen L-T, Schmidt HA, Von Haeseler A, Minh BQ. IQ-TREE: a fast and effective 1002 stochastic algorithm for estimating maximum-likelihood phylogenies. Molecular Biology and 1003 Evolution. 2015;32(1):268-74. 1004 46. Letunic I, Bork P. Interactive Tree Of Life (iTOL) v5: an online tool for phylogenetic 1005 tree display and annotation. Nucleic Acids Research. 2021;49(W1):W293-W6. 1006 47. Cao Y, Zhang Z, Ling N, Yuan Y, Zheng X, Shen B, et al. Bacillus subtilis SQR 9 can 1007 control Fusarium wilt in cucumber by colonizing plant roots. Biology and Fertility of Soils. 1008 2011;47:495-506. 1009 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 2, 2025. ; https://doi.org/10.1101/2025.09.02.673639doi: bioRxiv preprint 51 48. Morales-Ruiz E, López -Ceballos A, Maldonado -Mendoza IE. Transformation of the 1010 rhizospheric Bacillus cereus sensu lato B25 strain using a room-temperature electrocompetent 1011 cells preparation protocol. Plasmid. 2019;105:102435. 1012 49. Banks AM, Aminuddin F, Williams K, Batstone T, Barker GL, Foster GD, et al. 1013 Genome sequence of Lecanicillium fungicola 150-1, the causal agent of dry bubble disease. 1014 Microbiology Resource Announcements. 2019;8(19):10.1128/mra. 00340-19. 1015 50. Gea FJ, Carrasco J, Diánez F, Santos M, Navarro MJ. Control of dry bubble disease 1016 (Lecanicillium fungicola ) in button mushroom ( Agaricus bisporus ) by spent mushroom 1017 substrate tea. European Journal of Plant Pathology. 2014;138:711-20. 1018 51. Frostegård Å, Tunlid A, Bååth E. Phospholipid fatty acid composition, biomass, and 1019 activity of microbial communities from two soil types experimentally exposed to different 1020 heavy metals. Applied and Environmental Microbiology. 1993;59(11):3605-17. 1021 52. Zelles L. Fatty acid patterns of phospholipids and lipopolysaccharides in the 1022 characterisation of microbial communities in soil: a review. Biology and Fertility of Soils. 1023 1999;29:111-29. 1024 53. Hammer O, Harper DA, Ryan PD. Palaeontological statistics software package for 1025 education and data analysis. Palaeontologia Electronica. 2001;4(1):1-9. 1026 54. Taparia T, Krijger M, Hodgetts J, Hendriks M, Elphinstone JG, Van der Wolf J. Six 1027 multiplex TaqManTM -qPCR assays for quantitative diagnostics of pseudomonas species 1028 causative of bacterial blotch diseases of mushrooms. Frontiers in Microbiology. 2020;11:989. 1029 55. Köhl J, Vlaswinkel M, Groenenboom‐de Haas B, Kastelein P, Van Hoof R, Van der 1030 Wolf J, et al. Survival of pathogens of Brussels sprouts (Brassica oleracea Gemmifera Group) 1031 in crop residues. Plant Pathology. 2011;60(4):661-70. 1032 56. Pandin C, Le Coq D, Canette A, Aymerich S, Briandet R. Should the biofilm mode of 1033 life be taken into consideration for microbial biocontrol agents? Microbial Biotechnology. 1034 2017;10(4):719-34. 1035 57. Godfrey S, Harrow S, Marshall J, Klena J. Characterization by 16S rRNA sequence 1036 analysis of pseudomonads causing blotch disease of cultivated Agaricus bisporus. Applied and 1037 Environmental Microbiology. 2001;67(9):4316-23. 1038 58. Fan B, Blom J, Klenk H-P, Borriss R. Bacillus amyloliquefaciens, Bacillus velezensis, 1039 and Bacillus siamensis form an “operational group B. amyloliquefaciens” within the B. subtilis 1040 species complex. Frontiers in Microbiology. 2017;8:237721. 1041 59. Sur, S, Romo, TD, Grossfield, A . Selectivity and mechanism of fengycin, an 1042 antimicrobial lipopeptide, from molecular dynamics. The Journal of Physical Chemistry B. 1043 2018;122(8):2219-2226. 1044 60. Gilliard, G, Demortier, T, Boubsi, F, Jijakli, MH, Ongena, M, De Clerck, C, Deleu, M. 1045 Deciphering the distinct biocontrol activities of lipopeptides fengycin and surfactin through 1046 their differen tial impact on lipid membranes. Colloids and Surfaces B: Biointerfaces . 1047 2024;239:113933. 1048 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 2, 2025. ; https://doi.org/10.1101/2025.09.02.673639doi: bioRxiv preprint 52 61. Qin, Y, He, Y, She, Q, Larese-Casanova, P, Li, P, Chai, Y. Heterogeneity in respiratory 1049 electron transfer and adaptive iron utilization in a bacterial biofilm. Nature Communications. 1050 2019;10(1):3702. 1051 62. Akpa E, Jacques P, Wathelet B, Paquot M, Fuchs R, Budzikiewicz H, et al. Influence 1052 of culture conditions on lipopeptide production by Bacillus subtilis. Applied Biochemistry and 1053 Biotechnology. 2001;91(1):551-61. 1054 63. Wang J, Liu J, Wang X, Yao J, Yu Z. Application of electrospray ionization mass 1055 spectrometry in rapid typing of fengycin homologues produced by Bacillus subtilis. Letters in 1056 Applied Microbiology. 2004;39(1):98-102. 1057 64. Ma Y, Kong Q, Qin C, Chen Y, Chen Y, Lv R, et al. Identification of lipopeptides in 1058 Bacillus megaterium by two-step ultrafiltration and LC–ESI–MS/MS. Amb Express. 2016;6:1-1059 15. 1060 65. Carrasco J, Zied DC, Navarro MJ, Gea FJ, Pardo-Giménez A. Commercial cultivation 1061 techniques of mushrooms. Advances in Macrofungi: CRC Press; 2021. p. 11-40. 1062 66. Braat N, Koster MC, Wösten HA. Beneficial interactions between bacteria and edible 1063 mushrooms. Fungal Biology Reviews. 2022;39:60-72. 1064 67. Carrasco J, Preston G, inventors; Asociación Profesional De Productores De Sustratos 1065 Y Hongos De La Rioja, Navarra Y Aragón, assignee. Bacteria2023. 1066 68. Anastassiadou M, Arena M, Auteri D, Brancato A, Bura L, Carrasco Cabrera L, et al. 1067 Peer review of the pesticide risk assessment of the active substance Bacillus amyloliquefaciens 1068 strain QST 713 (formerly Bacillus subtilis strain QST 713). EFSA Journal. 2021;19(1):e06381. 1069 69. Navarro MJ, Santos M, Diánez F, Gea FJ. Chemical and biological control of wet 1070 bubble disease (Hypomyces perniciosus) in mushroom crops. Agronomy. 2023;13(7):1672. 1071 70. Carrasco J, García‐Delgado C, Lavega R, Tello ML, De Toro M, Barba‐Vicente V, et 1072 al. Holistic assessment of the microbiome dynamics in the substrates used for commercial 1073 champignon (Agaricus bisporus) cultivation. Microbial Biotechnology. 2020;13(6):1933-47. 1074 71. Wei-Ming C, Huai -Ying Y, Wei -Lin F, Qun -Li J, Yue -Yan L, Nan -Yi L, et al. 1075 Microbial community structure of casing soil during mushroom growth. Pedosphere. 1076 2009;19(4):446-52. 1077 72. Zhao Y, Li T, Shao P, Sun J, Xu W, Zhang Z. Variation in bacterial community 1078 structure in rhizosphere and bulk soils of different halophytes in the yellow river delta. 1079 Frontiers in Ecology and Evolution. 2022;9:816918. 1080 73. Rainey PB. The involvement of Pseudomonas putida in basidiome initiation of the 1081 cultivated mushroom Agaricus bisporus. 1989. 1082 74. Noble R, Fermor TR, Lincoln S, Dobrovin -Pennington A, Evered C, Mead A, et al. 1083 Primordia initiation of mushroom ( Agaricus bisporus ) strains on axenic casing materials. 1084 Mycologia. 2003;95(4):620-9. 1085 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 2, 2025. ; https://doi.org/10.1101/2025.09.02.673639doi: bioRxiv preprint 53 75. Lacey J. Actinomycetes in compost. Annals of Agricultural and Environmental 1086 Medicine. 1997;4(1). 1087 76. Akond MA, Jahan MN, Sultana N, Rahman F. Effect of temperature, pH and NaCl on 1088 the isolates of actinomycetes from straw and compost samples from Savar, Dhaka, Bangladesh. 1089 American Journal of Microbiology and Immunology. 2016;1(2):10-5. 1090 77. Ochi K. Insights into microbial cryptic gene activation and strain improvement: 1091 principle, application and technical aspects. The Journal of Antibiotics. 2017;70(1):25-40. 1092 78. Zakharova AA, Efimova SS, Malev VV, Ostroumova OS. Fengycin induces ion 1093 channels in lipid bilayers mimicking target fungal cell membranes. Scientific Reports. 1094 2019;9(1):16034. 1095 1096 1097 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 2, 2025. ; https://doi.org/10.1101/2025.09.02.673639doi: bioRxiv preprint