Enhancement of antimicrobial diversity in situ through relaxed symbiont

1 specificity in an insect/actinomycete partnership 2 3 Rita de Cassia Pessottia, Bridget L. Hansena, Jewel N. Reasoa, Javier A. Ceja-Navarrocd, Laila El4 Hifnawib, Eoin L. Brodieef, and Matthew F. Traxler*a 5 6 aDepartment of Plant and Microbial Biology, University of California, Berkeley 7 bDepartment of Molecular and Cellular Biology, University of California, Berkeley 8 cBioengineering and Biomedical Sciences Department, Biological Systems and Engineering Division, 9 Lawrence Berkeley National Laboratory, Berkeley, CA, USA 10 dInstitute for Biodiversity Science and Sustainability, California Academy of Sciences, Berkeley, CA, USA 11 eEcology Department, Earth and Environmental Sciences, Lawrence Berkeley National Laboratory, 12 Berkeley, CA, USA 13 fDepartment of Environmental Science, Policy and Management, University of California, Berkeley, CA, 14 USA 15 16 17


78
The exploration of these insect/actinomycete symbioses has provided key insights into the ecology of 79 microbial specialized metabolites. Importantly, analyses of the leafcutter ant and beewolf systems have 80 shown the co-evolution of the insect hosts and actinomycete symbionts, suggesting that these relationships, 81 and the molecules involved, have remained durable over tens of millions of years (Kaltenpoth et al., 2014; 82 Li et al., 2018). Both leafcutter ants and beewolves have specialized structures for maintaining their 83 actinomycete symbionts, which facilitate vertical transmission and high symbiont fidelity (Kaltenpoth et al.,  for the chemical repertoires found in these systems? Thus, we were motivated to identify an 89 insect/actinomycete symbioses that i) utilized a different mechanism of vertical transmission, and ii) enabled 90 direct detection of microbially-produced specialized metabolites in situ.

91
With this in mind, we assessed Odontotaenius disjunctus, a subsocial passalid beetle commonly found 92 in decomposing logs across eastern North America, and its frass (fecal material), as a model system for 93 studying the ecology of actinomycete specialized metabolism. Frass is an abundant and easily sampled 94 material in O. disjunctus galleries, and it is an important nutrient source in this system for both adult and 95 larval survival, and pupal chamber construction (Biedermann & Nuotclà, 2020;Mason & Odum, 1969; 96 Schuster & Schuster, 1985;Valenzuela-González, 1992 González, 1992) (Fig. 1C). Given the high nutrient content of frass, and the complex parental behaviors 132 associated with it, we drew parallels between this system and the other insect/actinomycete systems 133 described above. Thus, we hypothesized that O. disjunctus galleries, and frass specifically, might contain 134 actinomycete symbionts that have the potential to provide chemical defense to their hosts and the food 135 source on which their brood subsist.

136
To investigate if actinomycetes were associated with O. disjunctus galleries, we sampled material from 137 22 galleries across eastern North America (Fig. 1E, Supplementary Table S1). Samples included freshly 138 produced frass from live beetles and larvae, and frass and wood from within the galleries. Pupal chamber 139 material was also sampled when available and in this case, pupae were also gently sampled with a swab.

140
Using two selective media to enrich for actinomycetes, we isolated 339 bacterial strains (Supplementary   141   Table S2) and assayed their ability to inhibit growth of the Gram-positive bacterium Bacillus subtilis and the 142 fungal pathogen Candida albicans. We found that the frequency of bioactivity was surprisingly high among 143 these isolates. Specifically, 76.1% of the collection displayed activity against B. subtilis and/or C. albicans 144 (Fig. 1D, Table S2). These findings support the notion that coprophagy is a mode of vertical transmission of these 238 microbes, since the larvae are thought to exclusively consume frass (Valenzuela-González, 1992). Beyond 239 this, an analysis of metagenomic data previously generated by members of our team confirms that

243
Other isolates fell into areas of the tree that held higher phylogenetic diversity, and these strains 244 produced a wider array of compounds, including cycloheximide, polycyclic tetramate macrolactams (PTMs) 245 (e.g. alteramides), nigericin, piericidin, nactins, and novobiocin. The higher phylogenetic diversity of these 246 isolates suggests that they may represent transient members of the frass microbiota that have been more The fact that multiple microbial isolates from geographically remote galleries were found to produce the 267 same compounds, together with the in situ detection of these compounds, indicate that antimicrobials made 268 by actinomycetes often coexist in the frass environment. Therefore, we sought to explore chemical 269 interactions (i.e. synergism and antagonism) between a subset of the most commonly identified molecules 270 across our in situ and in vitro investigations. This list included the ionophore families of the nactins and 271 filipins, the angucyclinone STA-21 (a Stat3 inhibitor (Song et al., 2005)) and actinomycin X2 (a transcription 272 inhibitor (El-Naggar et al., 1999)). During our fieldwork, we collected an O. disjunctus carcass that was 273 partially covered with fungal biomass (Fig. 6A). We identified this material as a strain of Metarhizium 274 anisopliae (strain P287), an entomopathogenic fungus with a broad host range (Zimmermann, 1993). We 275 utilized M. anisopliae P287 as a target to investigate chemical interactions between the selected 276 compounds. Using the Bliss Independence model (Bliss, 1939), we found multiple instances of compound 277 interactions, including synergistic, antagonistic, and additive effects ( Fig. 5 and Supplementary Fig. S7).

279
B). The actinomycin X2/filipin result is notable since these compounds are usually made in concert by the 280 same organism (S. padanus). In contrast, actinomycin X2 displayed an antagonistic effect when tested in 281 combination with nactins (Fig. 5C). The combination of filipins and STA-21 (Fig. 5D)

311
We first tested the ability of the three streptomycetes to inhibit the two M. anisopliae strains in a plate-312 based assay. This assay showed that S. padanus P333 and S. scopuliridis P239 were able to produce 313 robust zones of inhibition against both M. anisopliae strains, while S. californicus P327 did not (Fig. 6B).

314
We next asked whether or not these Streptomyces isolates could inhibit the growth of the M. anisopliae 315 strains while growing in frass. To do so, we inoculated known quantities of spores of each microbe into 316 microtubes containing 3 mg of sterilized dry frass. The water used as the inoculation vehicle supplied 317 moisture, and the tubes were incubated at 30°C, which is close to the average temperature observed in O.

318
disjunctus galleries (see Supplementary Table 1), for seven days. The microbes were inoculated in 319 14 different combinations including: 1) a single microbe per tube, 2) one Streptomyces strain + one M. 320 anisopliae strain, and 3) combinations of two Streptomyces strains. Also, each microbe was inoculated into 321 empty microtubes as a control to assess growth promoted by frass.

322
All microorganisms were able to use frass as a substrate for growth, including both Metarhizium 323 anisopliae strains whose growth was enhanced ~14-20 fold compared to the no frass control (Fig. 6C, 324 Supplementary Fig. S9). We note that even though environmental frass often contains multiple 325 antimicrobials (e.g. Fig. 2), the heterogeneous nature of this material, plus autoclaving during preparation, 326 likely means that any native antimicrobials were at low concentration and/or inactivated in these microbial 327 growth assays. All three Streptomyces strains strongly inhibited M. anisopliae P016 and P287 growth in 328 frass (p<0.001, Fig. 6D). We next asked whether or not each Streptomyces strain produced its known 329 antimicrobials while growing in these frass assays. Metabolomics analysis of crude extracts of the frass 330 material revealed the presence of the actinomycins and filipins produced by S. padanus, nactins and 331 alteramides produced by S. californicus, and angucyclinones produced by S. scopuliridis, matching the 332 compounds produced in vitro by these three Streptomyces (Fig. 6F, Supplementary Fig. S10,11). These 333 results again highlight frass as an active site for production of antimicrobials, consistent with the notion that 334 these molecules likely inhibit M. anisopliae growth. However, we note that other molecules not identified 335 here could also play a role in this inhibition, as could competition for space and/or nutrients.

336
Next, we investigated if the Streptomyces strains were capable of inhibiting each other during growth 337 on frass. When we co-inoculated pairs of streptomyces on frass, the growth of S. padanus P333 was not 338 affected by either S. californicus P327 or S. scopuliridis (Fig. 6E). However, S. padanus P333 strongly 339 inhibited the growth of S. californicus P327. It was not possible to assay S. scopuliridis P239 growth via 340 plate counts in the presence of the other Streptomyces due to its vulnerability to the antimicrobials they 341 produced in vitro. However, we noted that production of the angucyclinone STA-21, which is produced by 342 S. scopuliridis P239, was dramatically reduced when S. scopuliridis P239 and S. padanus P333 where co-343 inoculated in frass, suggesting that S. padanus likely had a negative impact on S. scopuliridis P239 in this 344 treatment (Supplementary Fig. S11). Collectively, these findings offer direct evidence that in frass,        symbionts that are likely vertically transmitted. The idea that coprophagy may serve as a mechanism for 476 vertical transmission is further supported by our findings that multiple representatives of these clades were 477 isolated directly from fresh frass produced by larvae and adult beetles. Beyond this, data from our previous 478 metagenomic analysis indicates that Streptomyces are present throughout the beetle digestive tract, with 479 a notable enrichment in the posterior hindgut. Our phylogenetic analysis also indicates that frass contains 480 diverse actinomycetes that are transient or recently acquired members of this system. Thus, the O.

496
we found that they were strongly synergistic. Likewise, actinomycin X2 was also robustly synergistic with 497 the most commonly detected antimicrobial in frass, STA-21 (an angucyclinone). Thus, synergism may 498 potentiate the antimicrobial activity of multiple molecules produced by single strains, as well as molecules 499 produced across species. These findings are aligned with previous work that has suggested that some 500 insects, such as beewolves, might make use of cocktails of synergistic antimicrobials akin to 'combination 501 therapy' (Engl et al., 2018;Schoenian et al., 2011). In contrast, we also found multiple instances of 502 molecular antagonism, including between filipins and STA-21, and between actinomycin and nactins, which 503 were the second most frequently detected antimicrobial in frass samples. While antagonism between 504 molecules in frass may lead to diminished potency in the short term, emerging evidence indicates that 505 antagonism can guard against the evolution of antimicrobial resistance (Chait et al., 2007). Collectively, our 506 in vitro results, and the distributions of antimicrobials we detected in situ, lead us to speculate that 507 actinomycetes in frass likely produce an ever-shifting landscape of antimicrobial combinations, where their 508 activities are constantly enhanced or dampened, but also buffered against the development of pathogen 509 resistance. We hypothesize that such an environment may present a more challenging target for would-be

665
The type of compound interaction was determined by calculating the Bliss predicted value for independent 666 effect (EAB,Bliss) and Bliss excess (b) followed by a t-test using a method described elsewhere (Folkesson et  streptomycetes. Each microbe was also added to empty microtubes as a growth control, and some 676 microtubes containing frass were inoculated with water as a sterility control. In the case of multiple microbes 677 per tube, spores of each microbe were pre-mixed before adding them to the frass. Therefore, each 678 treatment had its own initial inoculum, which was plated to verify the exact initial concentration of each 679 microbe in each treatment by CFU count. All tubes were vortexed for 3 seconds and spun down for another 680 3 seconds. Microtubes were then incubated at 30ºC for one week. After the incubation time, 100 µL of a 681 solution of 0.03% of Tween80 was added to each tube and vortexed for 30 sec, left at RT for 1h, and 682 vortexed again for another 30 sec to detach cells from the frass. An aliquot of each tube was serially diluted 683 and plated for CFU count. The rest of the material was extracted with ethyl acetate (aqueous phase) and 684 methanol (frass material). Crude extracts were submitted for metabolomics analysis using the same 685 pipeline described above. In both cases of initial and final CFU counts, M. anisopliae counts were performed 686 using PDA plates supplemented with apramycin (25 µg/mL) to suppress the growth of the co-inoculated 687 streptomycete, and streptomycetes counts were performed using ISP2-agar plates. 688 689