Hülle-cell-mediated protection of fungal reproductive and overwintering structures against fungivorous animals

Fungal Hülle cells with nuclear storage and developmental backup functions are reminiscent of multipotent stem cells. In the soil, Hülle cells nurse the overwintering fruiting bodies of Aspergillus nidulans. The genome of A. nidulans harbors genes for the biosynthesis of xanthones. We show that enzymes and metabolites of this biosynthetic pathway accumulate in Hülle cells under the control of the regulatory velvet complex, which coordinates development and secondary metabolism. Deletion strains blocked in the conversion of anthraquinones to xanthones are delayed in maturation and growth of fruiting bodies. Xanthones are not required for sexual development but exert antifeedant effects on fungivorous animals such as springtails and woodlice. These findings reveal a novel role of Hülle cells in establishing secure niches for A. nidulans by accumulating metabolites with antifeedant activity that protect reproductive structures from animal predators.

Aspergillus comprises more than 300 species with high relevance for biotechnology, 56 pathogenicity and post-harvest crop protection (Samson et al., 2014). The biosynthetic 57 pathway for epishamixanthone and shamixanthone was firstly identified in Aspergillus 58 nidulans (Sanchez et al., 2011) and the corresponding gene cluster consists of one 59 polyketide synthase (PKS) encoding gene mdpG and 11 "tailoring" genes (mdpA-F, 60 mdpH-L). The mdp genes are biosynthetically linked with the three xpt genes xptA, xptB 61 and xptC, which are distributed over the genome, and encode two prenyltransferases 62 and an oxidoreductase (Fig. S1). All biosynthetic genes will be referred to as the 63 mdp/xpt gene cluster. In total, more than 30 compounds belonging mostly to the 64 chemical groups of anthraquinones, benzophenones and xanthones are synthesized 65 and secondary metabolism in response to environmental stimuli (Ö. Bayram et al., 2008). 77 After spore germination, a network of vegetative hyphae is formed, which develops in 78 certain environmental conditions through asexual or sexual developmental programs to 79 spore-bearing conidiophores or sexual fruiting bodies (cleistothecia) (Busch et al., 2007;80 Etxebeste et al., 2010). Light stimulates the asexual pathway, whereas lowered oxygen 81 levels and darkness stimulate the sexual pathway, respectively (Ö. Bayram et al., 2016). 82 The cleistothecium serves as overwintering structure and contains more than 10,000 83 sexual ascospores. It is surrounded by several layers of globose Hülle cells, which have repression on sexual fruiting body and resting structure formation of other fungi. We 99 showed in a food choice experiment that the mdp/xpt cluster metabolites present in 100 wildtype protect A. nidulans from soil animal predators. These results suggest that the 101 mdp/xpt metabolites produced in wildtype Hülle cells protect the sexual fruiting body of A. 102 nidulans from fungivorous animals. 103

Results 104
Proteins encoded by the mdp/xpt cluster are located in Hülle cells and sexual 105 mycelia in Aspergillus nidulans 106 Most of the mdp/xpt genes in A. nidulans are expressed during sexual development (Ö. 107 Bayram et al., 2016). A comparative proteome study on protein extracts of whole sexual 108 tissues as well as enriched Hülle cells from wildtype A4 was conducted and indicated a 109 specific spatial and temporal accumulation of Mdp/Xpt proteins ( Fig. S2a and Proteomic 110 MS analysis data). Vegetative and asexual mycelia were used as controls. Vegetative 111 mycelia were cultivated 20 h in liquid medium and asexual and sexual tissues as well as 112 Hülle cells were harvested three, five and seven days after inoculation on plates. An LC-113 MS analysis revealed that 24 proteins were present exclusively in both sexual mycelia 114 and enriched Hülle cells but were not identified from vegetative or asexual tissues (Table  115   S1). Among them, five proteins encoded by the mdp/xpt cluster were identified, MdpG, 116 MdpL, MdpH, XptB and XptC (Table 1). To verify the localization of these proteins, the 117 final enzyme in the biosynthesis of epi-/shamixanthone, XptC, was selected as an 118 example and C-terminally fused to GFP for fluorescence microscopy. The fusion protein 119 XptC-GFP was exclusively detected in three days-old sexual hyphae as well as in 120 enriched Hülle cells but not in 20 hours-old vegetative hyphae (Fig. 1a, Fig. S2b). The 121 stability of the fusion protein XptC-GFP was verified by the α-GFP antibody in Western 122 analysis (Fig. 1b). These results suggest that at least five members of the mdp/xpt 123 cluster, MdpG, MdpL, MdpH, XptB and XptC, are specifically located to Hülle cells as 124 well as sexual hyphae. Members of the Mdp/Xpt proteins can be detected from three to 125 seven days of sexual development. 126  After two days of sexual development, wildtype and deletion strains did not produce any 167 mdp/xpt metabolites (Fig. S5). After three and five days, wildtype produced arugosin A 168 (1) and the final xanthones emericellin (2), shamixanthone (3) and epishamixanthone (4) 169 ( Fig. 2b, Fig. S5, Table S2). As expected, loss of the first two enzymes of the    plates. The germination was monitored after 48 hours at 37°C. n = 40 (± 1) with two biological replicates.

The intact mdp/xpt cluster is required for accurate sexual development in
The effect of the extracted metabolites of the mdp/xpt deletion strains on cleistothecia 289 development of A. nidulans wildtype was examined to analyze whether the delayed 290 maturation of cleistothecia correlates with the accumulated cluster metabolites. 291 Metabolites of wildtype and mdp/xpt deletion strains were extracted after five days of 292 sexual development, solved in methanol and loaded onto paper discs separately. SM 293 loaded paper discs were placed on an agar plate inoculated with A. nidulans wildtype 294 spores. Pure methanol served as control. After five days of sexual incubation, the 295 cleistothecia development on each paper disc was monitored. Metabolites of wildtype 296 and ΔmdpG, ΔmdpF, ΔmdpL, ΔmdpD, ΔxptA, ΔxptB and ΔxptC displayed no effect on 297 cleistothecia development. Metabolites of ΔmdpC exhibited obvious effects on the 298 wildtype cleistothecia development producing significantly smaller and immature 299 cleistothecia ( Fig. 5e and Fig. S8). This verified that the metabolites of ΔmdpC 2,ω-300 dihydroxyemodin (5), ω-hydroxyemodin (6) and emodin (7), or at least one of them, 301 negatively influence cleistothecia development. In summary, the incomplete epi-

316
Error bar indicates standard deviation with five biological replicates; amount of cleistothecia of wt was set 317 to 100%. e) Box plot of cleistothecia diameter of A. nidulans wt grown on paper discs loaded with 318 extracted SMs of wt and mdp/xpt deletion strains. Strains were sexually grown for five days. SMs were 319 extracted, solved in MeOH and loaded on paper discs separately (pure MeOH was used as blank control).

320
Paper discs were placed on agar plates inoculated with 1 x 10 5 conidia of A. nidulans wt. Cleistothecia on 321 the paper discs were collected after five days of sexual growth (n ≥ 65). All significance tests are in 322 comparison to wt, ***/ ** P < 0.005 / 0.05, two-tailed t test. after seven days of surface cultivation (Teichert et al., 2020). When S. macrospora was 345 exposed to SMs of A. nidulans ΔmdpC, ΔmdpL and ΔmdpD, perithecia formation was 346 repressed, resulting in a halo surrounding the SM loaded paper disc. The biggest halo 347 was observed for ∆mdpC metabolites (Fig. 6, Fig. S9). Verticillium dahliae and V.

369
In order to identify the active components, commercially available pure chemicals were 370 tested. 75 µg of ω-hydroxyemodin (6), emodin (7) and chrysophanol (8) were loaded 371 separately onto paper discs and placed on agar plates inoculated with A. nidulans, S. 372 macrospora, V. dahliae and V. longisporum wildtypes (Fig. 7). S. macrospora, V. dahliae 373 and V. longisporum produced fewer fruiting bodies or resting structures, respectively, 374 when exposed to emodin (7)  Besides competition with other soil-borne microorganisms, fungi are facing the risk to be 392 attacked by fungivores. Therefore, we examined whether metabolites from the mdp/xpt 393 cluster protect A. nidulans from predators. The mdpC and mdpG complementation 394 strains (mdpC com , mdpG com ), which produce all mdp/xpt cluster metabolites like 395 wildtype (Fig. S10), the non-producing strain ∆mdpG as well as the emodins 396 accumulating strain ∆mdpC were offered to animal predators in a food choice 397 experiment (Fig. 8). Animals representing distant arthropod lineages were selected: the 398 mealworm larvae Tenebrio molitor (insect), the collembolan Folsomia candida (primitive 399 arthropod) and the woodlouse Trichorhina tomentosa (crustacean). Two different fungal 400 cultures were placed onto opposite sides of a Petri dish, and active animals were placed 401 onto the center area. The animals feeding on each fungal culture were counted along 402 with time. 403 T. molitor showed no obvious food preference facing the mdp/xpt cluster metabolites 404 (Fig. 8a). F. candida and T. tomentosa displayed a strong avoidance for A. nidulans 405 producing the final cluster products including epi-/shamixanthone (Fig. 8b-c). The 406 springtails and isopods gradually gathered on fungal cultures not producing the final 407 products, where they remained over 24 hours, indicating that the final cluster products 408 deter animal predators from feeding. Emodins, which are accumulated in ∆mdpC, have 409   Here, we showed that the mdp/xpt 450 metabolites are produced in Hülle cells (Fig. 3) and that they deter the animal predators 451 F. candida and T. tomentosa from feeding (Fig. 8b-c). This suggests a novel additional 452 role for Hülle cells to preserve survival of cleistothecia. The mdp/xpt metabolites are 453 produced as soon as the first Hülle cells occur to protect the developing, young fruiting 454 body and are still present when the cleistothecia are mature, guaranteeing long-term 455 protection (Fig. 2b, Fig. S5 and Fig. 5a). macrospora and two different species of Verticillium, but not of A. nidulans itself (Fig. 7). 493 inhibiting the sexual fruiting body or resting structure formation. We suggest that the 511 mdp metabolite emodin is produced to compete other fungi in order to maintain enough 512 space and nutrients for the own sexual fruiting body development. 513 Besides emodin, other mdp/xpt SMs seem to be involved in the repression of fruiting 514 and resting structure formation, since SMs of ΔmdpD, which do not contain emodin, are 515 involved in repression in S. macrospora, V. dahliae and V. longisporum (Fig. 6). Further, 516 enriched intermediates in the mdp/xpt deletion strains ΔmdpC, ΔmdpL, ΔmdpD, ΔxptA 517 and ΔxptB delayed the fruiting body maturation of A. nidulans (Fig. 5a). Whereas the 518 delay was gradually rescued along with the decrease of accumulated intermediates in 519 ΔmdpL, ΔmdpD, ΔxptA and ΔxptB (Fig 2B and Fig. S5), ΔmdpC showed an additional 520 reduction in cleistothecia size even after 25 days (Fig. S7). Emodin and ω-521 hydroxyemodin do not show an effect on cleistothecia size when externally added to A. 522 nidulans wildtype (Fig. 7). Therefore, 2,ω-dihydroxyemodin or a combination of the 523 different anthraquinone SMs might be the responsible compounds for the reduction in 524 cleistothecia size. Smaller cleistothecia have lower energy and material costs. The 525 accumulation of those intermediates might be an internal signal for incomplete 526 xanthone/arugosin biosynthesis, which leads to fruiting bodies unprotected from animal 527 predators. 528 In conclusion, our results suggest that the mdp/xpt pathway in A. nidulans was recruited 529 for the protection of fruiting bodies from predation. This occurred by (i) connecting the 530 anthraquinone producing mdp cluster with xpt genes converting anthraquinones into structures from animal predators (Fig. 9), which guarantees a long-term survival of A. 537 nidulans. 538 539 540 Fig. 9 The mdp/xpt cluster metabolites establish a secure niche for A. nidulans. 541 The velvet complex VelB-VeA-LaeA regulates sexual development of A. nidulans and the expression of

Transformation of E. coli and A. nidulans 638
E. coli transformation was performed using the heat-shock method (Inoue et al., 1990). 639 A. nidulans transformation was performed using the polyethylene glycol-mediated 640 protoplast fusion method (Punt et al., 1992). Successful transformation was further 641 verified by Southern hybridization (Southern et al., 1975). The recyclable marker 642 cassette was eliminated from the genome by culturing the fungus on 1% xylose MM 643 plate (Hartmann et al., 2010).

Protein digestion with trypsin and LC-MS analysis 665
Approximately 80 μg of protein were separated by SDS-PAGE for 60 minutes at 200 V. 666 The gel was stained (Neuhoff et al., 1988) and the lanes were excised and subjected to 667 tryptic digestion (Shevchenko et al., 1996). Digested peptides were desalted by using 668 and sexually grown for two, three, five, seven and 10 days. A 5.7 cm 2 agar piece of the 698 colonies was cut into small pieces and covered with 5 ml of ethyl acetate in 50 ml tube. 699 Tubes were shaken at 200 rpm at room temperature for 30 min followed by 10 min 700 highest level ultra-sonication in a Bandelin Sonorex TM Digital 10P ultrasonic bath 701 (Bandelin Electronic GmbH & Co. KG, Berlin, Germany). 3 ml of ethyl acetate phase was 702 transferred to a glass tube and evaporated. 703 SM sample was resuspended in 700 µl methanol and centrifuged for 10 min (13,000 g, 704 4°C) to remove particles. 500 µl of supernatant was transferred into the LC-MS vial. LC-705 MS was performed by using a Q ExactiveTM Focus orbitrap mass spectrometer coupled 706 to a Dionex Ultimate 3000 HPLC (Thermo Fisher Scientific). 5 µl of SM sample was 707 injected into the HPLC column (Acclaim TM 120, C 18 , 5 µm, 120 Å, 4.6 x 100 mm). The

Effect of secondary metabolite on fungi 734
SMs of mdp/xpt gene deletion strains and A. nidulans wildtype AGB552, extracted from 735 nine point-inoculated colonies, were dissolved in 450 µl of methanol. 30 µl of mixed 736 solution was loaded on the filter paper disc (Φ = 9 mm) individually. 30 µl of pure 737 methanol was used as a blank control. 75 µg of pure ω-hydroxyemodin (ChemFaces, 738 Wuhan, China), emodin (VWR, Darmstadt, Germany) and chrysophanol (VWR) were 739 dissolved in methanol and loaded on the paper disc for following tests. Approximately 20 animals were placed onto the center of the Petri dish and the number 762 on each side was counted over a period of 24 hours. The experiment was performed 763 with three biological and four technical replicates. 764 The food choice experiment with the isopod T. tomentosa was carried out as described 765 for F. candida with n = 8 animals.