bioRxiv preprint doi: https://doi.org/10.1101/2020.11.24.395384; this version posted November 24, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
1 Competition-based screening secures the
2 evolutionary stability of a defensive microbiome
3 1 2,† 1,3,† 4 Sarah F. Worsley , Tabitha M. Innocent , Neil A. Holmes , Mahmoud M. Al- 1 3 4 2* 5 Bassam , Barrie Wilkinson , J. Colin Murrell , Jacobus J. Boomsma , Douglas 1,5,6,* 1,3,* 6 W. Yu , Matthew I. Hutchings
7
1 8 School of Biological Sciences, University of East Anglia, Norwich Research Park, Norwich, Norfolk,
9 UK NR47TJ 2 10 Centre for Social Evolution, Section for Ecology and Evolution, University of Copenhagen,
11 Copenhagen, Denmark. 3 12 Department of Molecular Microbiology, John Innes Centre, Norwich Research Park, Norwich,
13 Norfolk, UK NR47UH 4 14 School of Environmental Sciences, University of East Anglia, Norwich Research Park, Norwich,
15 Norfolk, UK NR47TJ 5 16 State Key Laboratory of Genetic Resources and Evolution, Kunming Institute of Zoology, Chinese
17 Academy of Sciences, Kunming, Yunnan, China 650223 6 18 Center for Excellence in Animal Evolution and Genetics, Chinese Academy of Sciences, Kunming
19 Yunnan, China 650223
20
21
22 †These authors contributed equally to the manuscript
23 *Correspondence: [email protected], [email protected], [email protected],
24
25
bioRxiv preprint doi: https://doi.org/10.1101/2020.11.24.395384; this version posted November 24, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
26 Abstract.
27 Cuticular microbiomes of Acromyrmex leaf-cutting ants are exceptional because they are freely
28 colonizable, and yet the prevalence of Pseudonocardia, a native vertically transmitted
29 symbiont that controls Escovopsis fungus-garden disease, is never compromised. Game theory
30 suggests that competition-based screening can allow the selective recruitment of antibiotic-
31 producing bacteria from the environment, by fomenting and biasing competition for abundant
32 host resources. Mutual symbiont aggression benefits the host and also maintains native
33 symbiont viability. Here we use RNA-stable isotope probing (RNA-SIP)
34 to confirm predictions that Acromyrmex cuticles can maintain a range of microbial symbionts.
35 We then used dual-RNA-sequencing and bioassays to show that vertically transmitted
36 Pseudonocardia strains produce antibacterials that differentially reduce the growth rates of
37 other microbes, ultimately eliminating non-antibiotic-producing strains that might parasitize
38 the symbiosis while still allowing antibiotic-producing Streptomyces strains to survive. Open
39 cuticular microbiomes can thus maintain a specific co-evolved mutualism by restricting access
40 for other bacterial strains.
41
42
43 Keywords: antibiotics, Attini, game theory, defensive microbiome, mutualism,
44 Actinobacteria, partner, leaf-cutting ants, Pseudonocardia, interference competition,
45 horizontal acquisition, symbiosis
46
2 bioRxiv preprint doi: https://doi.org/10.1101/2020.11.24.395384; this version posted November 24, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
47 Introduction
48 The diversity of insect associated microbial communities is staggering. They may consist of
1 49 single intracellular symbionts with reduced genomes owing to coadaptation at one extreme , 2 50 to dynamic microbiomes in open host compartments such as guts at the other end of the scale .
51 Gut microbiomes have been most intensively studied in humans and other vertebrates because
52 there is increasing consensus of their vital interest for host fitness throughout ontogenetic
3, 4, 5 53 development . The stability and cooperative characteristics of complex microbiomes is a 6 54 paradox. While relentless competition is the default setting of the microbial world , hosts 7 55 appear to evolve control by holding their microbiome ecosystems on a leash , but how dynamic
56 stability under continuing turnover is achieved remains unclear. Despite an abundance of
57 microbiome research, recent reviews have concluded that “integration between theory and 8 58 experiments is a crucial ‘missing link’ in current microbial ecology” and that “our ability to 9 59 make predictions about these dynamic, highly complex communities is limited” .
60 Game theory suggests a compelling solution to the unity-in-diversity paradox by
61 showing that competition-based screening can be a powerful mechanism to maintain
62 cooperative stability. Screening is likely to work when hosts evolve to: (1) provide nutrients
63 and/or space to foment competition amongst symbionts, thus creating an attractive but
64 demanding environment, and (2) skew the competition so that the mutualistic symbionts enjoy
65 a comparative advantage. Competitive exclusion then ‘screens in’ mutualists and ‘screens out’ 10, 11, 12 66 parasitic or free-rider symbionts . Screening is conceptually clearest when the symbiont
67 trait that confers competitive superiority is the same as (or strongly correlated with) the trait
68 that benefits the host. An illustrative example of such correlated functionality was provided by
13 69 Heil who showed that ant-hosting acacia plants provide copious food bodies, which fuels the
70 production of numerous externally patrolling ant workers. The ant species whose colonies
71 invest in a greater number of, and more active, workers typically wins the plant, and the same 13 72 investment in active workers is likely to better protect the host plants against herbivores .
73 Screening has also been suggested to act in animal-microbe symbioses. For instance, 14 74 Tragust et al. showed that carpenter ants acidify their own stomachs by swallowing acidopore
75 secretions. Entomopathogenic bacteria are rapidly killed off, whereas the gut bacterial
76 symbiont Asiaia sp. (Acetobacteraceae) exhibits a lower mortality rate and maintains itself in 15 77 the midgut. Addressing a similar question, Itoh et al. used co-inoculation experiments to show
78 that environmentally recruited but co-adapted ‘native’ Burkholderia symbionts outcompete
79 non-native bacteria in the gut of their bean bug host, even though they are able to establish in
3 bioRxiv preprint doi: https://doi.org/10.1101/2020.11.24.395384; this version posted November 24, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
16 80 the absence of the ‘native’ symbiont. Finally, Ranger et al. showed that ambrosia beetles
81 selectively colonize physiologically stressed trees, which have a high ethanol titre due to
82 anaerobic respiration. The vertically transmitted fungal symbionts of these beetles have
83 evolved to detoxify the ethanol whereas competing weedy fungi remain inhibited.
84 Competition-based screening seems particularly apt for the establishment of protective
85 microbiomes when competition would screen in microbes that are similarly able to produce
86 compounds that kill competitors and thus contribute to defensive functions from the host’s 7, 10, 11, 17 87 perspective . Moreover, natural selection is expected to reinforce the correlation
88 between antibiotic production and antibiotic resistance, since production without resistance
89 would be suicidal. Antibiotic-resistant strains should thus be superior competitors in antibiotic-
11 90 filled environments, reinforcing the abundance of antibiotic producers as well . Previous
91 research has shown that the protective, cuticular microbiome of Acromyrmex echinatior
92 leafcutter ants (Formicidae, Attini) is an ideal model system to test the dynamic predictions of
93 screening theory in a context that also acknowledges the long-term ecosystem-on-a-leash
7, 18 94 perspective . These ants forage for fresh leaf fragments to provision their co-evolved fungus- 19, 20 95 garden mutualist Leucoagaricus gongylophorus . The fungal cultivar produces gongylidia, 21, 22 96 nutrient-rich swellings that are the sole food source for the queen and larvae and the 23 97 predominant food source for the workers who only ingest plant-sap in addition to fungal food .
98 However, Leucoagaricus is at risk of being parasitized by the specialized, coevolved mould
99 Escovopsis weberi, which can degrade the fungal cultivar and also cause severe ant paralysis 24, 25, 26, 27 100 and mortality . To prevent infections, leafcutter ants have evolved a range of weeding 28, 29, 30 101 and grooming behaviours , and A. echinatior and other Acromyrmex species also
102 maintain filamentous actinomycete bacteria that grow as a white bloom on the cuticles of large 26 103 workers and produce antimicrobials that inhibit the growth of E. weberi . In Panama, where
104 almost all fieldwork on this multipartite symbiosis has been carried out, the cuticle of
105 Acromyrmex workers is dominated by one of two vertically transmitted strains of
18, 31, 32, 33 106 Pseudonocardia, named P. octospinosus (Ps1) and P. echinatior (Ps2) .
107 Newly eclosed large workers are inoculated with the vertically transmitted
108 Pseudonocardia strain by their nestmates upon emergence. This blooms over the cuticle,
109 reaching maximum coverage after ca. six weeks, before shrinking back to the propleural plates 34, 35 110 as the ants mature to assume foraging tasks . Several studies have also identified other
111 actinomycete strains on the propleural plates of A. echinatior, which are presumed to be 17, 18, 36, 37, 38, 39 112 acquired from the environment . This includes species of the bacterial genus
4 bioRxiv preprint doi: https://doi.org/10.1101/2020.11.24.395384; this version posted November 24, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
36, 113 Streptomyces, which have been identified on ants maintained in laboratory-based colonies 37, 38 114 , as well as in 16S rRNA gene amplicon sequencing studies of ants sampled from their
115 native environment, although many of these could also have been close relatives of 18 116 Streptomyces . Species of this actinomycete genus produce a variety of antimicrobials so their 37, 38, 39 117 additional presence may suggest a form of multi-drug therapy against Escovopsis .
118 However, these putative functions remain enigmatic because Streptomyces symbionts were
18 35 119 never found on the callow workers that execute hygienic and defensive fungus-garden tasks .
120 In terms of resources, the propleural plates have a high concentration of tiny subcuticular 40 121 glands, which are presumed to supply the cuticular microbiome with resources . These plates
122 can thus be conjectured to create the food-rich but antibiotic-laden demanding environment
123 that competition-based screening assumes, because the vertically transmitted native 11 124 Pseudonocardia symbiont always colonizes the propleural plates first . We have previously
125 shown that both P. octospinosus and P. echinatior encode and make antibacterial compounds 31 126 that inhibit multiple unicellular bacteria, but not Streptomyces species . However, the other
127 key elements of the screening hypothesis have remained untested.
128 The present study provides a number of tests of competition-based screening, using the
129 cuticular microbiome of A. echinatior as an open symbiotic ecosystem model where the host
130 holds the resource leash. Within these constraints, the single co-adapted and dominant
131 defensive Pseudonocardia symbiont competes with an unspecified number of additional
132 symbionts, which have unknown functions but are predicted by screening theory to keep the
133 microbiome cooperative by excluding parasitic or free-riding microbes. We used RNA-SIP
134 (stable isotope probing) to show that the ants provide a public food resource on their cuticles
135 that is indeed consumed by multiple bacterial species. We then use dual RNA sequencing to
136 show, in vivo, that mutualistic P. octospinosus and P. echinatior strains express antibacterial
137 biosynthetic gene clusters (BGCs) on the ant cuticle. Next, we show that diffusible metabolites
138 of these Pseudonocardia species exhibit broad-spectrum antibacterial activity in vitro, but only
139 weakly inhibit Streptomyces species, which we separately and directly show are resistant to a
140 range of antibiotics. Finally, we use in-vitro competition experiments to show that slower-
141 growing Streptomyces species can competitively exclude faster-growing non-antibacterial-
142 producing species, but only when grown on media infused with Pseudonocardia metabolites.
143 Together, these results let us conclude that competition-based screening is a plausible
144 mechanism for the environmental acquisition of additional strains with similar mutualistic
145 inclinations as the native symbiont, and that screening thus maintains and dynamically
5 bioRxiv preprint doi: https://doi.org/10.1101/2020.11.24.395384; this version posted November 24, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
146 stabilizes the defensive microbiome initiated by the single native Pseudonocardia symbiont.
147 Our results underline that a host leash of a symbiotic microbiome can be restricted to a single
148 co-adapted symbiont that further shapes the microbiome to its own and the host’s interests,
149 provided its metabolites are hostile enough to avoid exploitation by non-beneficial invaders.
Results
150 The host provides public resources to its cuticular microbiome. RNA-SIP tracks the flow
151 of heavy isotopes from the host to the RNA of microbial partners that metabolize host-derived 41, 42, 43 152 resources . Labelled and unlabelled RNA within a sample are separated and used as
153 templates for 16S rRNA amplicon sequencing, so that the bacterial taxa that do (and do not)
43 154 use host-supplied resources can be identified . Three replicate groups of 22 mature worker 12 13 155 ants were fed a 20% (w/v) solution of either C or C glucose for 10 days and then propleural
156 plates were dissected out for total RNA extraction (Supplementary Figure 1). Control feeding
157 experiments demonstrated that a fluorescently labelled glucose-water diet was not transferred
158 to the surface of the propleural plate region of the ants during feeding, and the ants were only
159 ever observed to feed using their mandibles such that their propleural plates never came into
160 direct contact with the liquid diet (Supplementary Figure 2).
161 Following cDNA synthesis, 16S rRNA gene amplicon sequencing showed that
162 filamentous actinomycetes dominate the propleural plate samples, making up 76.6 % and 78.0 12 13 163 % of unfractionated cDNA samples from C and C fed ants, respectively (Figure 1). The 12 13 164 most abundant bacterial genera were Pseudonocardia, 35.8% and 38.1% in C and C
165 samples respectively, and Streptomyces, 19.7% and 20.5%, respectively. Wolbachia made up
166 22.8% and 19.6%, respectively. Wolbachia are known to be associated with the thoracic
167 muscles of A. echinatior worker ants where they may have an unspecified mutualistic 44, 45 168 function . We therefore assume these reads came from residual ant tissue on the dissected
169 propleural plates and do not consider them further. More than 95% of the Pseudonocardia 16S
170 rRNA gene reads in unfractionated RNA samples were identical in sequence to a single P. 18, 31 171 octospinosus mutualist strain . The presence of Streptomyces bacteria is consistent with
172 previous studies and supports the hypothesis that additional antibiotic-producing 17, 36, 37, 38, 39, 46 173 actinomycetes can be horizontally acquired by these large workers after always 18 174 first being inoculated with the native Pseudonocardia symbiont .
175
176
6 bioRxiv preprint doi: https://doi.org/10.1101/2020.11.24.395384; this version posted November 24, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
100
100 75 100
Phylum Genus actinotalea 75 PhylumActinobacteria 75 Bacteroidetes gordonia 50 acidobacteria microbacterium actinobacteriaFirmicutes Proteobacteria micrococcus bacteroidetes Tenericutes okibacterium chloroflexi 50 50 other Relative abundance (%) abundance Relative cyanobacteria propioniferax deinococcus_thermus pseudonocardia 25 firmicutes rhodococcus proteobacteria Relative abundance (%) abundance Relative Relative abundance (%) abundance Relative streptomyces tenericutes 25 25 tsukamurella wolbachia
0 0 0 12C ants 13C ants 12C ants 13C ants 12C ants 13C ants Treatment Treatment Treatment A B
177
178 Figure 1. Frequencies of bacteria at different taxonomic levels in unfractionated RNA samples 12 13 179 from the propleural plates of ants provided with a 20% (w/v) solution of either C- or C-
180 labelled glucose. (A) Phylum resolution, showing that >75% were actinobacteria. The
181 remaining reads almost all corresponded to Wolbachia (Proteobacteria) so the three other phyla 12 13 182 do not appear in the bars (total relative abundance of 0.6% and 0.4% in C and C fractions,
183 respectively). As Wolbachia is not part of the cuticular microbiome (see text) these results
184 show that the cuticular microbiome is completely dominated by actinobacteria. (B) Genus-
185 level resolution showing that the cuticular microbiomes are dominated by the native
186 Pseudonocardia symbiont followed by appreciable fractions of horizontally acquired
187 Streptomyces species and a series of other actinobacteria at low prevalence.
188
189 Caesium trifluoroacetate density gradient ultracentrifugation was used to separate the 13 12 13 190 C-labelled “heavier” RNA from the un-labelled C “lighter” RNA within the C-fed samples 12 191 (Supplementary Figure 1). RNA samples from the control C dietary treatment also underwent
192 density gradient ultracentrifugation. The resulting gradients were fractionated by buoyant
193 density (Supplementary Figures 3 & 4), and cDNA from each fraction was used in quantitative
194 RT-PCR reactions. This demonstrated that 16S rRNA gene copies had shifted to higher 13 195 buoyant densities in samples derived from the C dietary treatment, which is consistent with 13 196 the heavier C isotope being incorporated into the RNA of cuticular bacteria, due to the
197 metabolism of labelled host resources. Fractions spanning peaks across the buoyant density
198 gradient were selected for 16S rRNA gene amplicon sequencing (Supplementary Figure 4),
199 which confirmed that the actinobacterial genera had shifted to higher buoyant densities under
7 bioRxiv preprint doi: https://doi.org/10.1101/2020.11.24.395384; this version posted November 24, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
13 200 the C treatment (Figure 2A and B). Pseudonocardia and Streptomyces sequences had
201 respective relative abundances of 36.30% ± 4.17 (SE) and 19.18 ± 0.33 in fraction 7, and
13 202 32.30% ± 4.11 and 17.64 ± 0.26 in fraction 8 of the C samples (Figure 2B and C). These -1 203 fractions had an average buoyant density of 1.797 g ml ± 0.001 SE (fraction 7) and 1.805 g -1 204 ml ± 0.001 SE (fraction 8) and coincided with the peak in 16S rRNA gene copy number
205 identified by qPCR (Supplementary Figures 3 & 4).
12 206 In contrast, bacterial sequences from the C samples were barely detectable by qPCR
207 in fraction 8 (buoyant density of 1.801 g ml-1 ± 0.003) (Supplementary Figure 4). 12 208 Pseudonocardia and Streptomyces sequences were observed in fraction 7 of C samples
209 (Figure 2A and B), but only 1.59% ± 1.20 of the bacterial 16S rRNA sequences could be
210 amplified in qPCR experiments (Supplementary Figure 4). Peaks in 16S rRNA gene copy 12 211 number instead occurred in fractions 5-6 of C samples, which had average buoyant densities -1 212 of 1.780-1.788 g ml (Supplementary Figures 3 & 4). The other horizontally acquired taxa 12 13 213 showed similar shifts to higher buoyant densities between the C and C treatment groups;
214 for example, the genus Microbacterium made up 8.71% ± 1.43 of fraction 7, and 12.40% ±
215 0.68% of fraction 8, respectively (Figure 2C). Note that all these percentages become ca. 20%
216 higher after excluding Wolbachia, i.e. when considering only the cuticular microbiome.
15 217
60 A Pseudonocardia 20 B Streptomyces 10 Treatment 15 40 12C 13C Treatment Treatment 5 1012C 12C 13C 13C
Relative abundance (%) abundance Relative 20 5 Relative abundance (%) abundance Relative Relative abundance (%) abundance Relative 0 1 2 3 4 5 6 7 8 9 10 11 0 0 1 2 3 4 5 6 7 8 9 10 11 1 2 3 4 5 6 7 8 9 10 11 Buoyant density fraction Buoyant density fraction Buoyant density fraction
15 C Microbacterium D Wolbachia 20 Relative abundanceRelative(%) 10 15 Treatment Treatment 12C10 12C 13C 13C 5 5 Relative abundance (%) abundance Relative Relative abundance (%) abundance Relative 0 0 1 2 3 4 5 6 7 8 9 10 11 1 2 3 4 5 6 7 8 9 10 11
Buoyant density fraction Buoyant density fraction Buoyant density fraction 218
219 Figure 2. Average frequencies of the genera Pseudonocardia (A), Streptomyces (B),
220 Microbacterium (C), and Wolbachia (D) across fractions of the buoyant density gradient of
13 12 221 samples taken from ants fed on a C (blue) or C (red) glucose diet. Higher fraction numbers
8 bioRxiv preprint doi: https://doi.org/10.1101/2020.11.24.395384; this version posted November 24, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
-1 13 222 represent greater buoyant densities (g ml ), and only bacteria from the C-treated ants were 13 223 found in the heavier fractions on the right, indicating uptake of heavier C from the ant host.
224 Error bars indicate standard errors with n=3 replicate samples per treatment.
225 13 226 In addition to the shifts in buoyant density, indicating uptake of C by multiple bacterial genera 13 227 from the ant, several of the genera were also more abundant in the heavy fractions of the C
228 samples (which correspond with the peak in 16S rRNA gene copy number identified by qPCR, 12 229 Supplementary Figure 4) than in the heaviest fractions of the C samples (Supplementary
230 Figure 5), implying that they were particularly effective in their use of carbon-based resources
231 from their host ants. For example, Streptomyces species increased from an average frequency
12 13 232 of 14.8% ± 0.33% (SE) in C heavy fractions to 18.5% ± 0.29% in C heavy fractions
233 (Supplementary Figure 5). Interestingly, although Pseudonocardia was abundant in the 13 234 heaviest fractions of the C samples, its frequency decreased from an average of 53.65% 12 13 235 ±1.95% to 34.30 ±3.78% between the heaviest fractions of the C treatment and C treatment
236 (Supplementary Figure 5). This corresponded with a concurrent increase in the abundance of 13 237 other genera and a more even representation of genera in the heavier C fractions. These
238 increases in the other genera (all actinobacteria) are further evidence that ant-derived resources
239 were not solely being made available to the Pseudonocardia (which otherwise would have
13 240 dominated the C heavy fractions) but are publicly available to all bacteria on the cuticle. 12 241 The frequency of Wolbachia also increased from 9.86% ± 1.60% in the heavy C 13 242 fractions, to 19.49% ±3.53% in the C heavy fractions (Figure 2D; Supplementary Figure 5).
243 Since Wolbachia are extracellular endosymbionts, this finding supports the interpretation that
244 resources were supplied to cuticular bacteria by the ant hosts and not taken directly from the
245 glucose water. This interpretation is also supported by isotope ratio mass spectrometry (IRMS), 13 246 which showed that surface-washed ants incorporate a significant amount of the C from their
247 glucose diet into their bodies (Supplementary Figure 6), and by direct fluorescent microscopy
248 demonstrating that the glucose water was not transferred to the propleural plate (Supplementary
249 Figure 2).
250
251 Antibacterial BGCs are expressed by Pseudonocardia on the ant cuticle. We previously
252 generated high-quality genome sequences for five P. octospinosus and five P. echinatior strains
253 isolated from A. echinatior ant colonies and identified several BGCs (biosynthetic gene 31 254 clusters) in each of their genomes that are associated with antimicrobial activity . To establish
9 bioRxiv preprint doi: https://doi.org/10.1101/2020.11.24.395384; this version posted November 24, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
255 if these BGCs are expressed in vivo on the ant cuticle, total RNA was extracted and sequenced
256 from the propleural plates of ants in the captive colonies Ae088 (which hosts a vertically
257 transmitted P. echinatior strain) and Ae1083 (which hosts a P. octospinosus strain)
258 (Supplementary Table 1). A single RNA extraction was carried out for each colony, with each
259 sample consisting of the pooled propleural plates of 80 individual ants. RNA samples were 47 260 sequenced using a dual-RNA sequencing approach , and the resulting reads were quality
48 261 filtered and mapped to either the A. echinatior genome (~98% of reads, Supplementary Table 31 262 2) or their corresponding Pseudonocardia reference genomes (~1% of reads, Supplementary
263 Table 2). Both Pseudonocardia species showed very similar patterns of gene expression in
264 vivo, with genes involved in the production of secondary metabolites, including antibiotics (as
265 classified by KEGG) being expressed at similar levels by both Pseudonocardia strains on the
266 cuticle of A. echinatior ants (Supplementary Figure 7). 31 267 BGCs that are shared by the P. octospinosus and P. echinatior strains (Supplementary
268 Table 3) displayed remarkably similar patterns of in situ expression on the propleural plates
269 (Figure 4A). For both Pseudonocardia species, the most highly expressed BGCs encoded
270 proteins responsible for the synthesis of the compound ectoine and a putative carotenoid
271 terpene pigment (cluster D and F, Figure 4A). Such compounds are known to provide
272 protection against abiotic stressors such as desiccation and high concentrations of free radicals
49, 50, 51 273 which are often associated with biofilms . Also expressed on the propleural plates is a
274 shared BGC encoding a putative bacteriocin (Cluster E, Figure 4A), which belongs to a family
275 of ribosomally-synthesized post-translationally modified peptide (RiPP) antibiotics produced 52, 53, 54 276 by many species of bacteria ; these are known to prevent the formation of biofilms by
52, 55 277 other microbial species . In contrast, a shared Type 1 PKS gene cluster, encoding nystatin- 31, 37 278 like antifungal compounds , had very low expression in the P. echinatior strain and was not
279 expressed at all in the P. octospinosus strain (cluster C, Figure 4A). This suggests that
280 additional cues, such as direct exposure to E. weberi may be required to activate this BGC,
281 since both Pseudonocardia strains produce inhibitory antifungals when confronted with E.
282 weberi in vitro (Supplementary Figure 8). A Pseudonocardia strain isolated from Acromyrmex 37 283 ants has also previously been shown to produce nystatin-like compounds in vitro . The most
284 highly expressed BGCs unique to P. echinatior (cluster P, Figure 4C) or P. octospinosus
285 (cluster J, Figure 4B) are both predicted to encode bacteriocins.
286 Taken together, the results of the RNA-SIP and dual-RNA sequencing experiments are 18, 32 11, 12 287 consistent with both previous empirical research and the screening hypothesis : the ant
10 bioRxiv preprint doi: https://doi.org/10.1101/2020.11.24.395384; this version posted November 24, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
40 288 host provides public resources to its cuticular microbiome via glandular secretions for which
289 additional ectosymbionts compete once workers become exposed. This is always after the
290 native Pseudonocardia mutualists have established and gained dominance, which invariably
291 creates a demanding cuticular environment for any additional strain to invade.
292
500 AAe1083 450 Ae088 Acromyrmex echinatior 400 ant colony code 350 500 300 Ae1083 (associated with 250 450 Ae088 500 P. octospinosus) (RPKM) (RPKM) 200 Ae1083 400 150 450 Ae088
kilobase of exon model per million reads model per million reads kilobase of exon (associated with BGCs 100 350 400 P. echinatior) 50
Reads per 300 0 350 Oligosaccharide Terpene T1PKS Terpene Bacteriocin Ectoine (cluster A) (cluster B) (Cluster C) (cluster D) (clusterE) (cluster F) 300250 120 100 (RPKM) (RPKM) Pseudonocardia B C 250200 100
80 (RPKM) 200150
RPKM for 80 kilobase of exon model per million reads model per million reads kilobase of exon 60 150100
60 RPKM kilobase of exon model per million reads model per million reads kilobase of exon (RPKM) 10050 40 40 Reads per 500 20 Oligosaccharide Terpene T1PKS Terpene Bacteriocin Ectoine 20 Reads per 0 (cluster A) (cluster B) (Cluster C) (cluster D) (clusterE) (cluster F) Reads per kilobase of exon model per million reads
0 0 Oligosaccharide Terpene T1PKS Terpene Bacteriocin Ectoine Other Other NRPS Bacteriocin NRPS Other NRPS Other T1PKS-NRPS (clusterBacteriocin A) Terpene(clusterBacteriocin B) (Cluster C) (cluster D) (clusterE) (cluster F) (cluster G) (cluster H) (cluster I) (cluster J) (cluster K) (cluster L) (cluster M) (cluster N) (cluster O) (cluster P) (cluster Q) (cluster R)
293
294 Figure 3. Expression (in reads per kilobase of transcript per million mapped reads, RPKM) of
295 (A) biosynthetic gene clusters that are shared between the two Pseudonocardia species, (B)
296 biosynthetic gene clusters that are unique to P. octospinosus (colony Ae1083) and (C)
297 biosynthetic gene clusters that are unique to P. echinatior (colony Ae088). Biosynthetic gene
298 cluster codes relate to Table S3. Results from each colony were derived from a pool of 80 ants.
299
300 Pseudonocardia antibacterials create a demanding environment for non-antibiotic-
301 producing bacteria. Next we compared the growth rates of antibiotic-producing Streptomyces
302 strains and non-antibiotic-producing bacteria, which were all isolated from soil or fungus-
303 growing ant nests (Supplementary Table 1), on antibiotic-infused and control media. The
304 antibiotic-infused media were created by growing lawns of 17 Pseudonocardia isolates (Table
11 bioRxiv preprint doi: https://doi.org/10.1101/2020.11.24.395384; this version posted November 24, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
305 1) on SFM agar, while control media were inoculated with 20% glycerol. After a six-week
306 incubation period, the agar medium was flipped to reveal a surface for colonisation. The non-
307 producer strains grew more quickly on the undemanding control media while the antibiotic-
308 producers grew more quickly on the demanding Pseudonocardia-infused media, producing a
2 309 highly significant statistical interaction effect (n = 975, c = 45.86, df = 2, p < 0.0001; Fig. 4).
310 There was also a significant main effect of Pseudonocardia genotype, with both non-producers
311 and producers exhibiting a lower growth rate on P. echinatior (Ps2)-infused media than on P.
2 312 octospinosus (Ps1)-infused media (linear mixed-effects model, n = 915, c = 24.55, df = 1, p
313 < 0.0001, control-media data omitted for this analysis). This outcome is consistent with the
18 314 observation by Andersen et al. that Acromyrmex colonies hosting Ps2-dominated cuticular
315 microbiomes were less prone to secondary invasion by other bacteria.
316
317
318 Figure 4. Growth-rate experiments. Bacterial colony sizes after 5 days at 30 ºC. Boxplots
319 indicate medians ± one quartile. The white section shows growth rates on control media, and
320 the grey section shows growth rates on the Pseudonocardia-infused media. Red boxes
12 bioRxiv preprint doi: https://doi.org/10.1101/2020.11.24.395384; this version posted November 24, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
321 represent non-antibiotic-producer strains, and blue boxes represent antibiotic-producing 56 322 Streptomyces strains. A linear mixed-effects model , including 17 Pseudonocardia strains
323 and 20 inoculated bacterial species as random intercepts, was used to test for interaction and
324 main effects of growth media (Control vs. Ps1.infused vs. Ps2.infused) and antibiotic
325 production (Non.producers vs. Producers): lme4::lmer(Growth.score ~
326 Growth.media*Antibiotic.production+(1|Ps.strain/Plate)+(1|In.strain). One
327 non-producer strain (Staphylococcus epidermidis) grew more rapidly than all other strains
328 (open triangle points), demonstrating the need to control for correlated residuals.
329
330
331 To test the hypothesis that producer strains are generally resistant to antibiotics, which 11 332 would confer competitive superiority in an antibiotic-infused host environment , we grew ten
333 producer strains (all Streptomyces spp.) and ten non-producer strains (Supplementary Table 1)
334 in the presence of eight different antibiotics (Supplementary Table 4), representing a range of
335 chemical classes and modes of action. After seven days, Lowest Effective Concentration (LEC,
336 lowest concentration with inhibitory effect) and Minimum Inhibitory Concentration (MIC,
337 lowest concentration with no growth) scores were assigned on a Likert scale of 1–6, where a 57 338 score of 1 was no resistance and a score of 6 was resistance above the concentrations tested .
339 Antibiotic-producer strains exhibited greater levels of resistance, measured by both LEC
340 (Wilcoxon two-sided test, W = 94.5, p = 0.0017) and MIC scores (W = 80, p = 0.0253); p-
341 values corrected for two tests (Fig. 5).
342 We also performed growth rate experiments and measured antibiotic resistance profiles
343 with resident non-producer strains that were directly isolated from cuticular microbiomes.
344 These strains had significantly slower growth rates overall, even on control media without
345 antibiotics, suggesting that they may only be transient environmental contaminants
346 (Supplementary Figure 9). Resident non-producer strains also demonstrated high levels of
347 resistance (Supplementary Figure 10) as expected, given that they were isolated from ant
348 cuticles.
349
13 bioRxiv preprint doi: https://doi.org/10.1101/2020.11.24.395384; this version posted November 24, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
A B
6 6
4 4 Lowest Inhibitory (LEC) score Lowest Concentration Minimum InhibitoryMinimum (MIC) score Concentration
2 2
350 Antibiotic−producers Non−producers Antibiotic−producers Non−producers
351 Figure 5. Antibiotic resistance profiles (A LEC and B MIC) for producers and non-producer
352 strains (Supplementary Table 1), based upon each strain’s mean growth score across the eight
353 tested antibiotics (details in S3). Boxplots indicate medians (notches) ± one quartile. ‘Rabbit
354 ears’ in B indicate that the medians are also the highest values.
355
356 Pseudonocardia antibacterials allow Streptomyces to competitively exclude non-
357 antibiotic-producing bacteria. Finally, to test whether producer strains have a competitive
358 advantage in the demanding environment created by Pseudonocardia, we pairwise-competed
359 two of the Streptomyces producer strains, named S2 and S8 (Supplementary Table 1), against
360 each of 10 environmental non-producer strains on normal and on antibiotic-infused media. In
361 the latter case, Pseudonocardia was again grown on agar plates before turning the agar over
362 and coinoculating producer and non-producer test strains. On normal growth media,
363 Streptomyces were more likely to lose to non-producers, but when grown on Pseudonocardia-
2 364 infused media, Streptomyces were more likely to win (S8 on Ps1-infused media: n = 129, c =
365 103.6, df = 1, p < 0.0001; S2 on Ps2-infused media: general linear mixed-effects model, n =
2 366 94, c = 87.9, df = 1, p < 0.0001; Figure 6).
14 bioRxiv preprint doi: https://doi.org/10.1101/2020.11.24.395384; this version posted November 24, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
367
368 Figure 6. Pairwise competition experiment, scoring the frequency of producer wins. (A)
369 Representative images of agar plates at five days post-inoculation showing examples of the
370 three competitive outcomes: Win (producer S2 vs. non-producer St3 on Ps2 media), Loss
371 (producer S8 vs. non-producer St3 on control media), and Draw (producer S2 vs. non-producer
372 St3 on control media) (strain details in Supplementary Table 1). (B) Barcharts of competitive
373 outcomes for the two Streptomyces producer strains (S8 and S2; Supplementary Table 1). Each
374 Streptomyces strain was individually competed against ten different non-antibiotic-producer
375 strains. Streptomyces is more likely to win on Pseudonocardia-infused media. A general linear
376 mixed-effects model, with ten non-antibiotic-producer strains as a random intercept, was used
15 bioRxiv preprint doi: https://doi.org/10.1101/2020.11.24.395384; this version posted November 24, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
377 to test for effect of medium (Control vs. Ps-infused) on competitive outcome (Win vs. Loss)
378 lme4::glmer(outcome ~ medium + (1 | non.producer.strain), family = binomial).
Discussion
379 We tested screening theory using the external (cuticular) microbiome of the leafcutter ant
380 Acromyrmex echinatior as an experimental model. We used RNA-SIP to show for the first time
381 that an animal host is directly feeding its microbiome by secreting a public resource that all
382 surface-resident bacteria can use (Figure 2), and we confirm that the public resource is indeed
383 used by a number of environmentally acquired bacterial species. We further demonstrated, in
384 two separate captive colonies, that both P. octospinosus and P. echinatior strains express
385 antibacterial biosynthetic gene clusters (BGCs) on the ant cuticle (Figure 3). We next showed
386 that the two species of actinobacteria have broad-spectrum antibacterial activity against
387 environmental isolates in vitro but, importantly, have a weaker effect on Streptomyces species
388 (Figure 4), consistent with these Streptomyces species being resistant to a range of antibiotics,
389 which is typical for this genus (Figure 5). Finally, we used in vitro competition experiments to
390 demonstrate that the Streptomyces species can competitively exclude faster-growing bacteria
391 that do not make antibiotics, but only when the competing species are grown on media infused
392 with Pseudonocardia metabolites (Figure 6).
393 Taken together, these results are consistent with the hypothesis that competition-based
394 screening is a plausible mechanism for the environmental acquisition and assembly of a multi-
395 species defensive microbiome, provided that the vertically transmitted Pseudonocardia
396 symbionts become established first to create a demanding environment. This priority
397 establishment is invariably the case because callow large workers are inoculated by their 34 398 nestmates within 24 hours of emerging from their pupae , meaning that Pseudonocardia is 11 399 never competitively excluded from the microbiome, as predicted by Scheuring & Yu and 18 400 empirically shown by Andersen et al. . The new results reported here confirm the tractability
401 of the cuticular leafcutter-ant microbiome, which is accessible to experimentation and for
402 which the adaptive benefit to the ant hosts is clearly defined and explicitly testable: defence 25, 37, 58, 59 403 against specialized Escovopsis pathogens . Our present results remain consistent with
404 Streptomyces strains being superior contenders for secondary acquisition, in the context of, but
405 without being a threat to, the vertical transmission across generations of the native symbionts
406 by dispersing gynes that found new colonies after being inseminated.
16 bioRxiv preprint doi: https://doi.org/10.1101/2020.11.24.395384; this version posted November 24, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
407 Our results are of broad significance because they provide mechanistic evidence for the 7 408 combined applicability of the evolutionary-ecosystem-on-a-leash concept and screening
409 theory specifying the ecological dynamics of tilting the competitive landscape in favour of 11 410 other antibiotic-producing bacteria . The evolutionary perspective suggests that, depending on
411 their degree of openness, host-associated microbiomes may have both native core members
412 that co-adapt with host environments, while the ecological perspective addresses the
413 mechanism by which successfully establishing bacterial colonists are unlikely to be parasites
414 or free-riders. Studies in other symbioses appear to support this dual evolutionary and 13, 15, 16 415 ecological view , both for an array of mutualistic symbioses with multicellular partners
416 and for microbiomes more specifically. The combined evolutionary and ecological perspective
417 of our present study is remarkably parallel to the priority effects documented for vertically
418 transmitted (actinomycete) Bifidobacterium species in the milk glycobiomes that inoculate the 10, 60 419 gut microbiome of human neonates . Similar processes may also play out during the 61, 62, 63 420 establishment of plant-root microbiomes and could be open to manipulation in efforts to
64 421 improve crop yields .
422 Future work on the Acromyrmex model system should aim at characterising the
423 antibacterial molecules made by the symbiotic Pseudonocardia strains in vitro and in vivo and
424 matching these compounds to the BGCs expressed on the ants. This will be challenging
425 because ant-derived Pseudonocardia strains grow poorly on agar plates and not at all in liquid
426 culture. Additionally, Imaging Mass Spectrometry has, as yet, not been possible on the ants
427 themselves.
428
429 Materials and Methods.
430 Ant colony collection and maintenance. Colonies of A. echinatior (Hymenoptera,
431 Formicidae, Attini) were collected from the Gamboa area of the Soberania National Park,
432 Panama, between 2001 and 2014. Colonies Ae1083 and Ae088 (Supplementary Table 1) were
433 maintained under controlled temperature conditions (25ºC) at UEA and fed a daily diet of
434 bramble and laurel leaves. Additional colonies were maintained at the University of
435 Copenhagen in rearing rooms at ca 25ºC and 70% relative humidity, where they were fed with
436 bramble leaves and occasional supplements of apple and dry rice.
437
13 438 RNA SIP C feeding experiment. Six replicate groups of 22 mature worker ants with visible
439 bacterial growth on their propleural plates were selected from colony Ae1083 (Supplementary
17 bioRxiv preprint doi: https://doi.org/10.1101/2020.11.24.395384; this version posted November 24, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
440 Table 1) and placed into 9cm petri dishes containing a 2x2 cm square of cotton wool soaked in
441 water. Following 24 hours of starvation, three replicate groups of 22 ants were supplied with
13 442 300 µl of a 20% C glucose solution (w/v, Sigma Aldrich) and the remaining three groups
12 443 were supplied with 300 µl of a 20% C glucose solution (w/v, Sigma Aldrich) for 10 days.
444 Glucose solutions were supplied to ants in microcentrifuge tube caps and were refreshed every
13 13 445 three days. To confirm uptake of the C isotope by ants fed on the C labelled diet, a further
446 5 ants were fed on each type of glucose diet; these were submitted for Isotope Ratio Mass
447 Spectrometry (IRMS) analysis (Supplementary Note 1), which enables the relative abundance 13 12 448 of each stable isotope ( C and C) to be quantified in a sample. To confirm that the glucose
449 water diet did not spread over the ant cuticle, a 20% glucose solution containing a non-toxic -1 450 fluorescent green drain tracing dye (5 mg ml , Hydra) was fed to the ants. Ants were sampled
451 just after taking a feed, and after 6 and 24 hours of being exposed to the dye, to trace the spread
452 of the solution over time. The ants were imaged using fluorescent microscopy (Supplementary
453 Figure 2, Supplementary Note 1).
454
455 Cuticular dissection and RNA extraction. At the end of the 10-day feeding experiment, the
456 propleural plates of the propleura were removed from the ventral exoskeleton using a dissection
457 microscope and fine sterile tweezers. Propleural plates from each of the 22 ants in each group
458 were placed together in lysis matrix E tubes (MP Biomedicals) on dry ice, before being snap
459 frozen in liquid nitrogen. A modified version of the Qiagen RNeasy Micro Kit protocol was
460 used for all RNA extractions (Supplementary Note 1). The quantity and purity of all RNA
461 samples was checked using a nanodrop spectrophotometer and a Qubit™ RNA HS assay kit
462 (Invitrogen™).
463
464 Density gradient ultracentrifugation and fractionation. Density gradient ultracentrifugation
13 465 was carried out to separate C labelled (“heavy”) from un-labelled (“light”) RNA within the
466 same sample (Supplementary Note 1). Following centrifugation, samples were divided into 12
467 fractions using a peristaltic pump to gradually displace the gradient, according to an established 43 468 protocol . The refractive index of fractions was measured to confirm the formation of a linear
469 buoyant density gradient (Supplementary Figure 3). RNA was precipitated from fractions by
470 adding 1 volume of DEPC-treated sodium acetate (1M, pH 5.2), 1 µl (20 µg) glycogen (from
471 mussels, Sigma Aldrich) and 2 volumes of ice cold 96% ethanol. Fractions were incubated over
472 night at -20°C then centrifuged for 30 minutes at 4°C and 13,000 x g, before washing with 150
18 bioRxiv preprint doi: https://doi.org/10.1101/2020.11.24.395384; this version posted November 24, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
473 µl of ice cold 70% ethanol and centrifuging for a further 15 minutes. Pellets were then air-dried
474 for 5 minutes and re-suspended in 15 µl of nuclease free water.
475
476 Quantifying 16S rRNA gene copy number across RNA SIP fractions. The RNA in each
477 fraction was converted to cDNA by following the manufacturer’s instructions for Superscript
478 II (Invitrogen) with random hexamer primers (Invitrogen). 16S rRNA gene copy number was
479 then quantified across cDNA fractions using qPCR (Supplementary Note 1).
480
481 Sequencing and analysis. PCR was used to amplify the 16S rRNA gene in each of the fractions
482 that spanned the peaks in 16S rRNA gene copy number, identified via qPCR. This was done
483 using the primers PRK341F and 518R (Supplementary Table 1). One unfractionated sample 13 12 484 was also created for ants under each of the C or C dietary treatments, by pooling equal
485 quantities of unfractionated cDNA from each of the 3 replicate groups and using this as
486 template for PCR amplification. The resulting PCR products were purified using the Qiagen
487 MinElute™ gel extraction kit and submitted for 16S rRNA gene amplicon sequencing using
488 an Illumina MiSeq at MR DNA (Molecular Research LP), Shallowater, Texas, USA. Sequence 65, 66 489 data was then processed by MR DNA using their custom pipeline . As part of this pipeline,
490 paired-end sequences were merged, barcodes were trimmed, and sequences of less than 150 bp
491 and/or with ambiguous base calls were removed. The resulting sequences were denoised, and
492 OTUs were assigned by clustering at 97% similarity. Chimeras were removed, and OTUs were
493 taxonomically assigned using BLASTn against a curated database from GreenGenes, RDPII,
67 494 and NCBI . Plastid-like sequences were removed from the analysis. Upon receipt of the 16S
495 rRNA gene sequencing data from MR DNA, OTU assignments were verified using QIIME2 68 496 and BLASTn, and statistical analysis was carried out using R 3.2.3 . OTUs assigned as
497 Pseudonocardia were blasted against the 16S rRNA gene sequences for P. echinatior and P.
31 498 octospinosus to confirm the relative abundance of each of these vertically-transmitted strains
499 in the samples. All 16S rRNA gene amplicon sequencing data from this experiment has been
500 deposited in the European Nucleotide Archive (ENA) public database under the study
501 accession number PRJEB32900.
502
503 Detecting the expression of Pseudonocardia BGCs in situ.
504 RNA extraction and sequencing from ant propleural plate samples. The propleural plates
505 of Acromyrmex echinatior ants were dissected (as described above) from individual mature
19 bioRxiv preprint doi: https://doi.org/10.1101/2020.11.24.395384; this version posted November 24, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
506 worker ants that had a visible growth of Pseudonocardia bacteria on their cuticle. A pool of 80
507 ant cuticles were sampled from each of the colonies Ae1083 and Ae088, respectively
508 (Supplementary Table 1), after which RNA was extracted as described above. The quantity,
509 purity and integrity of all RNA samples was checked using a nanodrop spectrophotometer and
510 Qubit™ RNA HS assay kit (Invitrogen™), as well an Experion™ bioanalyser with a
511 prokaryotic RNA standard sensitivity analysis kit (Bio-Rad, California, USA). One µg of RNA
512 from each of the propleural plate samples was sent to Vertis Biotechnologie AG (Freising-
513 Weihenstephan, Germany) where samples were processed and sequenced using a Dual RNA 47 514 sequencing approach as described in . Single-end sequencing (75 bp) was performed using
515 an Illumina NextSeq500 platform. All sequencing reads have been deposited in the ENA public
516 database under the study accession number PRJEB32903. Upon receipt, the resulting RNA
517 reads from each sample were quality filtered and aligned to both the A. echinatior reference 48 518 genome and a representative genome for either P. octospinosus (from Ae707), or P.
519 echinatior (from Ae706) (Supplementary Table 1; Supplementary Note 1). Reads that mapped
520 to the Pseudonocardia genomes were then mapped back to the ant genome (and vice versa) to
521 check that reads did not cross-map between the two genomes. Following alignment, uniquely
522 mapped reads were counted and read counts per annotated coding sequence were converted to
523 reads per kilobase of transcript per million mapped reads (RPKM) (Supplementary Note 1).
524
525 Expression analysis. In order to investigate the expression levels of different functional groups
31 526 of genes, protein sequences of every annotated gene in each Pseudonocardia genome 69 527 (Supplementary Table 1) were extracted and uploaded to BlastKOALA . Assigned K numbers
528 were classified into five main KEGG pathway categories (and their associated sub-categories)
529 using the KEGG Pathway Mapper tool. Each gene, with its associated K number and category
530 assignments, was then matched to its RPKM value from the RNA sequencing dataset so that
531 the expression levels of different KO categories could be established. To investigate the
532 expression of biosynthetic gene clusters (BGCs) by Pseudonocardia on the ant cuticle, 31 533 reference genome sequences for P. octospinosus and P. echinatior (Supplementary Table 1)
534 were uploaded to antiSMASH version 4.0, which predicts the presence and genomic location 70 535 of BGCs based on sequence homology to known clusters . RPKM values were then generated
536 for each predicted BGC, based on the length of the predicted cluster and read counts for genes
537 situated within it.
538
20 bioRxiv preprint doi: https://doi.org/10.1101/2020.11.24.395384; this version posted November 24, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
539 Escovopsis bioassays. The Pseudonocardia strains PS1083 and PS088 (Supplementary Table
540 1) were isolated from the propleural plates of individual large Acromyrmex echinatior workers
541 taken from colonies Ae1083 and Ae088 (colonies used in RNA-seq experiments, see
542 Supplementary Table 1), respectively. Each of the strains were bioassayed against the
543 specialized fungal pathogen Escovopsis weberi strain CBS 810.71, which was acquired from
544 the Westerdijk Fungal Biodiversity Institute (Supplementary Table S1). Isolation of the
545 Pseudonocardia strains and bioassays are described in detail in Supplementary Note 1.
546
547 Collection and isolation of bacterial strains. Nineteen strains of Pseudonocardia (11 strains
548 of P. echinatior and 8 of Pseudonocardia octospinosus, Supplementary Table 1) were isolated
549 from the cuticles of individual Acromyrmex echinatior worker ants across 18 different colonies
550 and genotypes, as described in Supplementary Note 1. Ten of these isolated Pseudonocardia 31 551 strains were previously genome-sequenced by Holmes et al. . The 10 environmental
552 antibiotic-producer strains (all in the genus Streptomyces) were taken from general collections
553 in the Hutchings lab and are a mixture of isolates from either soil environments or from worker
554 ants taken from captive colonies (Supplementary Table 1). The 10 environmental non-producer
555 strains were obtained from the Hutchings lab (two strains) and from the ESKAPE suite (eight
556 strains with varying origins (human skin, soil, etc.) used to test antibiotic resistance or efficacy
557 in clinical/research settings (Supplementary Table 1).
558
559 Individual growth rate and pairwise competition experiments. Antibiotic-infused media
560 was created by growing lawns of 17 Pseudonocardia isolates (Supplementary Table 1) on SFM
561 agar, whilst control SFM media was inoculated with 20% glycerol. After a six-week incubation
562 period, the agar medium was flipped to reveal a new surface for colonisation. All producer and
563 non-producer strains (Supplementary Table 1) were then inoculated onto the plates to compare
564 their growth rates on each type of media (Supplementary Note 1). After 5 days of incubation
565 at 30 °C, photographs were taken and growth area was measured. (Supplementary Note 1).
566 Antibiotic-producer strains were also grown in direct competition against non-producers, both
567 on normal control medium and on medium infused with Pseudonocardia secondary
568 metabolites. Plates were incubated at 30 °C for 5 days before imaging. Images were then scored
569 with respect to the producer as: win, draw or loss (Supplementary Note 1).
570
21 bioRxiv preprint doi: https://doi.org/10.1101/2020.11.24.395384; this version posted November 24, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
571 Antibiotic resistance assays. To test whether producer and non-producer strains differed in
572 their resistance to antibiotics, each strain (Supplementary Table 1) was grown in the presence
573 of eight different antibiotics, representing a range of chemical classes and modes of action
574 (Supplementary Table 4). Producers and non-producers were inoculated onto plates and
575 incubated at 30 °C for 7 days, then photographed and scored on a Likert scale of 1–6.
576 (Supplementary note 1).
577
578 Statistical analyses. R scripts and data used for statistical analyses of Figures 4, 5, 6, and S9
579 are available at github.com/dougwyu/Worsley_et_al_screening_test_R_code
580
581 Acknowledgements. This work was supported by a NERC PhD studentship to SFW (NERC
582 Doctoral Training Programme grant NE/L002582/1), NERC grants NE/J01074X/1 to MIH and
583 DWY, NE/M015033/1 and NE/M014657/1 to MIH, JCM, DWY, and BW, European Research
584 Council Advanced grant 323085 to JJB, and Marie Curie Individual European Fellowship grant
585 627949 to TI. DWY was also supported by the Key Research Program of Frontier Sciences,
586 Chinese Academy of Sciences (QYZDY-SSW-SMC024) and the Strategic Priority Research
587 Program, Chinese Academy of Sciences (XDA20050202, XDB31000000). We thank the
588 Smithsonian Tropical Research Institute for logistical help and facilities for all work in
589 Gamboa, Panama, and the Autoridad Nacional del Ambiente y el Mar for permission to sample
590 and export ants. We thank Simon Moxon for advice on RNA sequencing analysis and Paul
591 Disdle for his assistentance with IRMS analysis. The authors declare no conflicts of interest.
592 Figure Legends
593 Figure 1. Frequencies of bacteria at different taxonomic levels in unfractionated RNA samples 12 13 594 from the propleural plates of ants provided with a 20% (w/v) solution of either C- or C-
595 labelled glucose. (A) Phylum resolution, showing that >75% were actinobacteria. The
596 remaining reads almost all corresponded to Wolbachia (Proteobacteria) so the three other phyla 12 13 597 do not appear in the bars (total relative abundance of 0.6% and 0.4% in C and C fractions,
598 respectively). As Wolbachia is not part of the cuticular microbiome (see text) these results
599 show that the cuticular microbiome is completely dominated by actinobacteria. (B) Genus-
600 level resolution showing that the cuticular microbiomes are dominated by the native
601 Pseudonocardia symbiont followed by appreciable fractions of horizontally acquired
602 Streptomyces and a series of other actinobacteria at low prevalence.
22 bioRxiv preprint doi: https://doi.org/10.1101/2020.11.24.395384; this version posted November 24, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
603
604 Figure 2. Average frequencies of the genera Pseudonocardia (A), Streptomyces (B),
605 Microbacterium (C), and Wolbachia (D) across fractions of the buoyant density gradient of 13 12 606 samples taken from ants fed on a C (blue) or C (red) glucose diet. Higher fraction numbers -1 13 607 represent greater buoyant densities (g ml ), and only bacteria from the C-treated ants were 13 608 found in the heavier fractions on the right, indicating uptake of heavier C from the ant host.
609 Error bars indicate standard errors with n=3 replicate samples per treatment.
610
611 Figure 3. Expression (in reads per kilobase of transcript per million mapped reads, RPKM) of
612 (A) biosynthetic gene clusters that are shared between the two Pseudonocardia species, (B)
613 biosynthetic gene clusters that are unique to P. octospinosus (colony A.e1083) and (C)
614 biosynthetic gene clusters that are unique to P. echinatior (colony A.e088). Biosynthetic gene
615 cluster codes relate to Table S3. Results from each colony were derived from a pool of 80 ants.
616
617 Figure 4. Growth-rate experiments. Bacterial colony sizes after 5 days at 30 ºC. Boxplots
618 indicate medians ± one quartile. The white section shows growth rates on control media, and
619 the grey section shows growth rates on the Pseudonocardia-infused media. Red boxes
620 represent non-antibiotic-producer strains, and blue boxes represent antibiotic-producing
56 621 Streptomyces strains. A linear mixed-effects model , including 17 Pseudonocardia strains
622 and 20 inoculated bacterial species as random intercepts, was used to test for interaction and
623 main effects of growth media (Control vs. Ps1.infused vs. Ps2.infused) and antibiotic
624 production (Non.producers vs. Producers): lme4::lmer(Growth.score ~
625 Growth.media*Antibiotic.production+(1|Ps.strain/Plate)+(1|In.strain). One
626 non-producer strain (Staphylococcus epidermidis) grew more rapidly than all other strains
627 (open triangle points), demonstrating the need to control for correlated residuals.
628
629 Figure 5. Antibiotic resistance profiles (A LEC and B MIC) for producers and non-producer
630 strains (Supplementary Table 1), based upon each strain’s mean growth score across the eight
631 tested antibiotics (details in S3). Boxplots indicate medians (notches) ± one quartile. ‘Rabbit
632 ears’ in B indicate that the medians are also the highest values.
633
634 Figure 6. Pairwise competition experiment, scoring the frequency of producer wins. (A)
635 Representative images of agar plates at five days post-inoculation showing examples of the
23 bioRxiv preprint doi: https://doi.org/10.1101/2020.11.24.395384; this version posted November 24, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
636 three competitive outcomes: Win (producer S2 vs. non-producer St3 on Ps2 media), Loss
637 (producer S8 vs. non-producer St3 on control media), and Draw (producer S2 vs. non-producer
638 St3 on control media) (strain details in Supplementary Table 1). (B) Barcharts of competitive
639 outcomes for the two Streptomyces producer strains (S8 and S2; Supplementary Table 1). Each
640 Streptomyces strain was individually competed against ten different non-antibiotic-producer
641 strains. Streptomyces is more likely to win on Pseudonocardia-infused media. A general linear
642 mixed-effects model, with ten non-antibiotic-producer strains as a random intercept, was used
643 to test for effect of medium (Control vs. Ps-infused) on competitive outcome (Win vs. Loss)
644 lme4::glmer(outcome ~ medium + (1 | non.producer.strain), family = binomial).
645 Supporting Information
646
647 Supplementary Note 1: Additional information regarding methods used for IRMS analysis,
648 RNA extractions and the analysis of RNA sequencing, in addition to detailed methods for the
649 growth-rate and pairwise competition experiments.
650
651 Supplementary Figure 1: Overview of methodology used for RNA stable isotope probing of
652 the propleural plate microbiome of Acromyrmex echinatior leafcutter ants.
653
654 Supplementary Figure 2: The extent to which a 20% (w/v) glucose water diet containing
655 green fluorescent dye (A) was distributed over the ant body directly after taking a feed (B), 6
656 hours after feeding (C) and 24 hours after feeding (D). DT = fluorescence shining through from
657 the ant digestive tract, PP = propleural plates with growth of actinobacteria.
658
659 Supplementary Figure 3: The relationship between RNA-SIP fraction number and buoyant -1 660 density (in g ml ). Fractions were generated from density gradients containing RNA isolated
12 13 661 from the propleural plates of A. echinatior ants fed on either a C (blue) or a C (orange)
662 glucose water diet. Points represent averages (three samples each of 22 ants per dietary
663 treatment) ± standard error.
664
665 Supplementary Figure 4: 16S rRNA gene copy number across different fractions of buoyant
666 density gradients, as determined via qPCR. Gene copy number in each fraction is displayed as
24 bioRxiv preprint doi: https://doi.org/10.1101/2020.11.24.395384; this version posted November 24, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
667 a percentage of total copy number in each sample. There were three replicate samples of 22 12 13 668 ants fed a C (light grey) or C (dark grey) glucose diet. Diamond symbols represent fractions
669 that were sent for 16S rRNA gene amplicon sequencing and yellow diamonds represent those
670 designated as “heavy” fractions under the different treatments.
671
672 Supplementary Figure 5: The relative abundances (%) of genera in the “heavy” fractions of
12 13 673 buoyant density gradients for C and C samples, respectively. Heavy fractions were
674 determined via qPCR, which identified peaks in 16S rRNA gene copy number in both sets of
675 samples (see Supplementary Figure 4). Genera reported in Figure 1 (in unfractionated samples)
676 were also recovered in fractionated samples; Pseudonocardia was the most prevalent genus,
677 followed by Streptomyces,Wolbachia and Microbacterium. Additional actinomycete genera
678 were recovered at lower abundances. Genera labelled as “other” were recovered in trace
679 amounts.
680
13 12 13 681 Supplementary Figure 6: The atom percentage of C in ants fed either a C or C labeled
682 20% (w/v) glucose diet for 10 days, as determined by Isotope Ratio Mass Spectrometry (IRMS)
683 analysis.
684
685 Supplementary Figure 7. Expression levels (in reads per kilobase of transcript per million
686 mapped reads, RPKM) of Kegg orthology pathway categories. (A) Pseudonocardia
687 octospinosus (colony Ae088) and (B) Pseudonocardia echinatior (colony Ae1083) on the
688 propleural plates of Acromyrmex echinatior ants. N= 1 sample of 80 pooled ants per colony.
689
690 Supplementary Figure 8. The bioactivity of Pseudonocardia isolates (A) P. echinatior PS088
691 and (B) P. octospinosus PS1083 against the specialized fungus-garden pathogen Escovopsis
692 weberi (Supplementary Table 1).
693
694 Supplementary Figure 9. Individual growth-rate experiments of Acromyrmex-resident, non-
695 producer strains, assessed as bacterial colony sizes after 5 days at 30 ºC, with the boxplots
696 indicating medians ± one quartile. The white section shows growth rates on control media, and
697 the grey section shows growth rates on the Pseudonocardia-infused media. Red boxes
698 represent non-producer strains, and blue boxes represent producer strains. For analysis, a linear
699 mixed-effects model, including Pseudonocardia strain (n=17) and inoculated bacterial species
25 bioRxiv preprint doi: https://doi.org/10.1101/2020.11.24.395384; this version posted November 24, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
700 (n=20) as random factors, was used to test for interactions and main effects of growth media
701 (Control vs. Ps1-infused vs. Ps2-infused) and antibiotic production (Non producers vs.
2 702 Streptomyces). There was no significant interaction effect (c = 2.64, df = 2, p = 0.27), but both
703 main effects were highly significant. The resident non-producers isolated from cuticular
2 704 microbiomes had significantly slower growth rates on all media (c = 20.96, df = 1, p < 0.0001)
705 including the control media without antibiotics (Fig. 1B). This suggests that they are unable to
706 outcompete producer strains on the cuticle of Acromyrmex ants and raises the question why
707 these non-producer species can persist at all. Bacterial growth was also generally slower on
2 708 Ps2-infused media than on Ps1-infused media (c = 21.43, df = 1, p < 0.000 when analyzed in
709 a balanced design without the control-media.
710
711 Supplementary Figure 10. Antibiotic resistance profiles for producer, non-producer, and
712 resident non-producer strains (Supplementary Table 1). Boxplots indicate medians (notches)
713 ± one quartile. For analysis, we calculated each strain’s mean growth score across the eight
714 tested antibiotics (reducing from n = 155 to n = 20; details in S3). Producers showed higher
715 levels of resistance than did non-producers for both measurements: Wilcoxon two-sided test
716 (wilcox.test), W = 94.5, p = 0.0017 for LEC (A) and W = 80, p = 0.0253 for MIC (B), after
717 correction for multiple testing. Producers and Resident Non-producers showed no difference
718 in resistance levels (p = 0.44 and 0.25). The ‘rabbit ears’ in B indicate that the medians are also
719 the highest values.
720
721 Supplementary Table 1: Details of ant colonies, bacterial strains and fungal strains, reference
722 genomes and primers used in experiments. * Pseudonocardia strains that have been genome-
723 sequenced. † Pseudonocardia strains that were only used in the growth-rate experiment with
724 non-producers.
725
726 Supplementary Table 2: The total number of RNA-sequencing reads (and percentage of total
727 reads in brackets) originating from propleural plate samples taken from the ant colonies
728 Ae1083 or Ae088, respectively, that successfully aligned to the A. echinatior genome
729 (Supplementary Table 1) and to the genomes of the Pseudonocardia species associated with
730 the ant colony of origin (P. octospinosus or P. echinatior, respectively).
731
26 bioRxiv preprint doi: https://doi.org/10.1101/2020.11.24.395384; this version posted November 24, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
732 Supplementary Table 3: Secondary metabolite BGCs in the Pseudonocardia mutualist 31 733 genomes (table adapted from Holmes et al. ) and their associated expression values (in reads
734 per kilobase of transcript per million mapped reads, RPKM) in RNA-sequencing experiments.
735 Yellow rows are BGCs shared between strains. Green and blue represent BGCs that are unique
736 to P. octospinosus and P. echinatior, respectively.
737
738 Supplementary Table 4: Media recipes and antibiotics used in this study.
739 References
740 1. McCutcheon JP, Boyd BM, Dale C. The Life of an Insect Endosymbiont from the Cradle 741 to the Grave. Curr Biol 29, R485-R495 (2019).
742
743 2. Brucker RM, Bordenstein SR. The hologenomic basis of speciation: gut bacteria cause 744 hybrid lethality in the genus Nasonia. Science 341, 667-669 (2013).
745
746 3. Rosshart SP, et al. Wild Mouse Gut Microbiota Promotes Host Fitness and Improves 747 Disease Resistance. Cell 171, 1015-1028 e1013 (2017).
748
749 4. Sommer F, Backhed F. The gut microbiota--masters of host development and 750 physiology. Nat Rev Microbiol 11, 227-238 (2013).
751
752 5. Maynard CL, Elson CO, Hatton RD, Weaver CT. Reciprocal interactions of the intestinal 753 microbiota and immune system. Nature 489, 231-241 (2012).
754
755 6. Granato ET, Meiller-Legrand TA, Foster KR. The Evolution and Ecology of Bacterial 756 Warfare. Curr Biol 29, R521-R537 (2019).
757
758 7. Foster KR, Schluter J, Coyte KZ, Rakoff-Nahoum S. The evolution of the host 759 microbiome as an ecosystem on a leash. Nature 548, 43-51 (2017).
760
761 8. Widder S, et al. Challenges in microbial ecology: building predictive understanding of 762 community function and dynamics. ISME J 10, 2557-2568 (2016).
763
764 9. Hoye BJ, Fenton A. Animal host-microbe interactions. J Anim Ecol 87, 315-319 (2018).
765
27 bioRxiv preprint doi: https://doi.org/10.1101/2020.11.24.395384; this version posted November 24, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
766 10. Archetti M, Ubeda F, Fudenberg D, Green J, Pierce NE, Yu DW. Let the right one in: a 767 microeconomic approach to partner choice in mutualisms. Am Nat 177, 75-85 (2011).
768
769 11. Scheuring I, Yu DW. How to assemble a beneficial microbiome in three easy steps. Ecol 770 Lett 15, 1300-1307 (2012).
771
772 12. Boza G, Worsley SF, Yu DW, Scheuring I. Efficient assembly and long-term stability of 773 defensive microbiomes via private resources and community bistability. Plos Comput 774 Biol 15, e1007109 (2019).
775
776 13. Heil M. Let the best one stay: screening of ant defenders by Acacia host plants 777 functions independently of partner choice or host sanctions. Journal of Ecology 101, 778 684-688 (2013).
779
780 14. Tragust S, et al. Formicine ants swallow their highly acidic poison for gut microbial 781 selection and control. eLife 9, e60287 (2020).
782
783 15. Itoh H, et al. Host-symbiont specificity determined by microbe-microbe competition 784 in an insect gut. Proc Natl Acad Sci U S A 116, 22673-22682 (2019).
785
786 16. Ranger CM, et al. Symbiont selection via alcohol benefits fungus farming by ambrosia 787 beetles. Proc Natl Acad Sci U S A 115, 4447-4452 (2018).
788
789 17. Sen R, Ishak HD, Estrada D, Dowd SE, Hong E, Mueller UG. Generalized antifungal 790 activity and 454-screening of Pseudonocardia and Amycolatopsis bacteria in nests of 791 fungus-growing ants. Proc Natl Acad Sci USA 106, 17805 (2009).
792
793 18. Andersen SB, Hansen LH, Sapountzis P, Sorensen SJ, Boomsma JJ. Specificity and 794 stability of the Acromyrmex-Pseudonocardia symbiosis. Mol Ecol 22, 4307-4321 795 (2013).
796
797 19. Chapela IH, Rehner SA, Schultz TR, Mueller UG. Evolutionary History of the Symbiosis 798 Between Fungus-Growing Ants and Their Fungi. Science 266, 1691 (1994).
799
800 20. Mueller UG, Rehner SA, Schultz TR. The evolution of agriculture in ants. Science 281, 801 2034-2038 (1998).
802
28 bioRxiv preprint doi: https://doi.org/10.1101/2020.11.24.395384; this version posted November 24, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
803 21. Currie CR. A community of ants, fungi, and bacteria: a multilateral approach to 804 studying symbiosis. Annu Rev Microbiol 55, 357-380 (2001).
805
806 22. De Fine Licht HH, Boomsma JJ, Tunlid A. Symbiotic adaptations in the fungal cultivar 807 of leaf-cutting ants. Nat Commun 5, 5675 (2014).
808
809 23. Sapountzis P, Zhukova M, Shik JZ, Schiott M, Boomsma JJ. Reconstructing the functions 810 of endosymbiotic Mollicutes in fungus-growing ants. Elife 7, (2018).
811
812 24. Reynolds HT, Currie CR. Pathogenicity of Escovopsis weberi: The parasite of the attine 813 ant-microbe symbiosis directly consumes the ant-cultivated fungus. Mycologia 96, 814 955-959 (2004).
815
816 25. Heine D, et al. Chemical warfare between leafcutter ant symbionts and a co-evolved 817 pathogen. Nat Commun 9, 2208 (2018).
818
819 26. Worsley SF, et al. Symbiotic partnerships and their chemical interactions in the 820 leafcutter ants (Hymenoptera: Formicidae). Myrmecological News 27, 59-74 (2018).
821
822 27. de Man TJ, et al. Small genome of the fungus Escovopsis weberi, a specialized disease 823 agent of ant agriculture. Proc Natl Acad Sci U S A 113, 3567-3572 (2016).
824
825 28. Currie CR, Stuart AE. Weeding and grooming of pathogens in agriculture by ants. Proc 826 Biol Sci 268, 1033-1039 (2001).
827
828 29. Fernandez-Marin H, et al. Functional role of phenylacetic acid from metapleural gland 829 secretions in controlling fungal pathogens in evolutionarily derived leaf-cutting ants. 830 Proc Biol Sci 282, 20150212 (2015).
831
832 30. Fernandez-Marin H, Zimmerman JK, Rehner SA, Wcislo WT. Active use of the 833 metapleural glands by ants in controlling fungal infection. Proc Biol Sci 273, 1689-1695 834 (2006).
835
836 31. Holmes NA, et al. Genome analysis of two Pseudonocardia phylotypes associated with 837 Acromyrmex leafcutter ants reveals their biosynthetic potential. Front Microbiol 7, 838 2073 (2016).
839
29 bioRxiv preprint doi: https://doi.org/10.1101/2020.11.24.395384; this version posted November 24, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
840 32. Poulsen M, Cafaro M, Boomsma JJ, Currie CR. Specificity of the mutualistic association 841 between actinomycete bacteria and two sympatric species of Acromyrmex leaf- 842 cutting ants. Mol Ecol 14, 3597-3604 (2005).
843
844 33. Currie CR, Scott JA, Summerbell RC, Malloch D. Fungus-growing ants use antibiotic- 845 producing bacteria to control garden parasites. Nature 398, 701-704 (1999).
846
847 34. Marsh SE, Poulsen M, Pinto-Tomás A, Currie CR. Interaction between workers during 848 a short time window is required for bacterial symbiont transmission in Acromyrmex 849 leaf-cutting ants. PloS one 9, e103269 (2014).
850
851 35. Poulsen M, Bot ANM, Currie CR, Nielsen MG, Boomsma JJ. Within-colony transmission 852 and the cost of a mutualistic bacterium in the leaf-cutting ant Acromyrmex 853 octospinosus. Funct Ecol 17, 260-269 (2003).
854
855 36. Kost C, Lakatos T, Böttcher I, Arendholz W-R, Redenbach M, Wirth R. Non-specific 856 association between filamentous bacteria and fungus-growing ants. 857 Naturwissenschaften 94, 821-828 (2007).
858
859 37. Barke J, et al. A mixed community of actinomycetes produce multiple antibiotics for 860 the fungus farming ant Acromyrmex octospinosus. BMC Biol 8, (2010).
861
862 38. Haeder S, Wirth R, Herz H, Spiteller D. Candicidin-producing Streptomyces support 863 leaf-cutting ants to protect their fungus garden against the pathogenic fungus 864 Escovopsis. Proc Natl Acad Sci USA 106, 4742-4746 (2009).
865
866 39. Schoenian I, Spiteller M, Ghaste M, Wirth R, Herz H, Spiteller D. Chemical basis of the 867 synergism and antagonism in microbial communities in the nests of leaf-cutting ants. 868 Proc Natl Acad Sci U S A 108, 1955-1960 (2011).
869
870 40. Currie CR, Poulsen M, Mendenhall J, Boomsma JJ, Billen J. Coevolved crypts and 871 exocrine glands support mutualistic bacteria in fungus-growing ants. Science 311, 81- 872 83 (2006).
873
874 41. Dumont MG, Murrell JC. Stable isotope probing - linking microbial identity to function. 875 Nat Rev Microbiol 3, 499-504 (2005).
876
30 bioRxiv preprint doi: https://doi.org/10.1101/2020.11.24.395384; this version posted November 24, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
877 42. Neufeld JD, Wagner M, Murrell JC. Who eats what, where and when? Isotope-labelling 878 experiments are coming of age. ISME J 1, 103-110 (2007).
879
880 43. Whiteley AS, Thomson B, Lueders T, Manefield M. RNA stable-isotope probing. Nat 881 Protoc 2, 838-844 (2007).
882
883 44. Andersen SB, Boye M, Nash DR, Boomsma JJ. Dynamic Wolbachia prevalence in 884 Acromyrmex leaf-cutting ants: potential for a nutritional symbiosis. J Evol Biol 25, 885 1340-1350 (2012).
886
887 45. Sapountzis P, Nash DR, Schiott M, Boomsma JJ. The evolution of abdominal 888 microbiomes in fungus-growing ants. Mol Ecol 28, 879-899 (2019).
889
890 46. Seipke RF, et al. A single Streptomyces symbiont makes multiple antifungals to support 891 the fungus farming ant Acromyrmex octospinosus. PLoS One 6, e22028 (2011).
892
893 47. Westermann AJ, et al. Dual RNA-seq unveils noncoding RNA functions in host- 894 pathogen interactions. Nature 529, 496-501 (2016).
895
896 48. Nygaard S, et al. The genome of the leaf-cutting ant Acromyrmex echinatior suggests 897 key adaptations to advanced social life and fungus farming. Genome Res 21, 1339- 898 1348 (2011).
899
900 49. Gershenzon J, Dudareva N. The function of terpene natural products in the natural 901 world. Nat Chem Biol 3, 408-414 (2007).
902
903 50. Richter TK, Hughes CC, Moore BS. Sioxanthin, a novel glycosylated carotenoid, reveals 904 an unusual subclustered biosynthetic pathway. Environ Microbiol 17, 2158-2171 905 (2015).
906
907 51. Nupur LN, Vats A, Dhanda SK, Raghava GP, Pinnaka AK, Kumar A. ProCarDB: a database 908 of bacterial carotenoids. BMC Microbiol 16, 96 (2016).
909
910 52. Chikindas ML, Weeks R, Drider D, Chistyakov VA, Dicks LM. Functions and emerging 911 applications of bacteriocins. Curr Opin Biotechnol 49, 23-28 (2018).
912
31 bioRxiv preprint doi: https://doi.org/10.1101/2020.11.24.395384; this version posted November 24, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
913 53. Cotter PD, Ross RP, Hill C. Bacteriocins - a viable alternative to antibiotics? Nat Rev 914 Microbiol 11, 95-105 (2013).
915
916 54. Riley MA, Gordon DM. The ecological role of bacteriocins in bacterial competition. 917 Trends Microbiol 7, 129-133 (1999).
918
919 55. Algburi A, Zehm S, Netrebov V, Bren AB, Chistyakov V, Chikindas ML. Subtilosin 920 prevents biofilm formation by inhibiting bacterial quorum sensing. Probiotics 921 Antimicrob Proteins 9, 81-90 (2017).
922
923 56. Bates D, Mächler M, Bolker B, Walker S. Fitting Linear Mixed-Effects Models Using 924 lme4. Journal of Statistical Software; Vol 1, Issue 1 (2015), (2015).
925
926 57. Balouiri M, Sadiki M, Ibnsouda SK. Methods for in vitro evaluating antimicrobial 927 activity: A review. J Pharm Anal 6, 71-79 (2016).
928
929 58. Currie C, Bot A, Boomsma JJ. Experimental evidence of a tripartite mutualism: bacteria 930 protect ant fungus gardens from specialized parasites. Oikos 101, 91-102 (2003).
931
932 59. Andersen SB, Yek SH, Nash DR, Boomsma JJ. Interaction specificity between leaf- 933 cutting ants and vertically transmitted Pseudonocardia bacteria. BMC Evol Biol 15, 1 934 (2015).
935
936 60. Zivkovic AM, German JB, Lebrilla CB, Mills DA. Human milk glycobiome and its impact 937 on the infant gastrointestinal microbiota. Proc Natl Acad Sci U S A 108 Suppl 1, 4653- 938 4658 (2011).
939
940 61. Carlstrom CI, Field CM, Bortfeld-Miller M, Muller B, Sunagawa S, Vorholt JA. Synthetic 941 microbiota reveal priority effects and keystone strains in the Arabidopsis 942 phyllosphere. Nat Ecol Evol 3, 1445-1454 (2019).
943
944 62. Cai T, et al. Host legume-exuded antimetabolites optimize the symbiotic rhizosphere. 945 Mol Microbiol 73, 507-517 (2009).
946
947 63. Neal AL, Ahmad S, Gordon-Weeks R, Ton J. Benzoxazinoids in root exudates of maize 948 attract Pseudomonas putida to the rhizosphere. PLoS One 7, e35498 (2012).
949
32 bioRxiv preprint doi: https://doi.org/10.1101/2020.11.24.395384; this version posted November 24, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
950 64. Ryan PR, Dessaux Y, Thomashow LS, Weller DM. Rhizosphere engineering and 951 management for sustainable agriculture. Plant and Soil 321, 363-383 (2009).
952
953 65. Dowd SE, et al. Evaluation of the bacterial diversity in the feces of cattle using 16S 954 rDNA bacterial tag-encoded FLX amplicon pyrosequencing (bTEFAP). BMC Microbiol 8, 955 125 (2008).
956
957 66. Dowd SE, Wolcott RD, Sun Y, McKeehan T, Smith E, Rhoads D. Polymicrobial nature of 958 chronic diabetic foot ulcer biofilm infections determined using bacterial tag encoded 959 FLX amplicon pyrosequencing (bTEFAP). PLoS One 3, e3326 (2008).
960
961 67. DeSantis TZ, et al. Greengenes, a chimera-checked 16S rRNA gene database and 962 workbench compatible with ARB. Appl Environ Microbiol 72, 5069-5072 (2006).
963
964 68. R Core Team. R: A language and environment for statistical computing. In: R 965 Foundation for Statistical Computing, Vienna, Austria. https://www.R-project.org/) 966 (2017).
967
968 69. Kanehisa M, Sato Y, Morishima K. BlastKOALA and GhostKOALA: KEGG tools for 969 functional characterization of genome and metagenome sequences. J Mol Biol 428, 970 726-731 (2016).
971
972 70. Blin K, et al. AntiSMASH 4.0-improvements in chemistry prediction and gene cluster 973 boundary identification. Nucleic Acids Res 45, W36-W41 (2017).
974 975
33 bioRxiv preprint doi: https://doi.org/10.1101/2020.11.24.395384; this version posted November 24, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
1 Competition-based screening secures the
2 evolutionary stability of a defensive microbiome:
3 Supplementary Information
4 1 2,† 1,3,† 5 Sarah F. Worsley , Tabitha M. Innocent , Neil A. Holmes , Mahmoud M. Al- 1 3 4 2* 6 Bassam , Barrie Wilkinson , J. Colin Murrell , Jacobus J. Boomsma , Douglas 1,5,6,* 1,3,* 7 W. Yu , Matthew I. Hutchings
8
1 9 School of Biological Sciences, University of East Anglia, Norwich Research Park, Norwich, Norfolk,
10 UK NR47TJ 2 11 Centre for Social Evolution, Section for Ecology and Evolution, University of Copenhagen,
12 Copenhagen, Denmark. 3 13 Department of Molecular Microbiology, John Innes Centre, Norwich Research Park, Norwich,
14 Norfolk, UK NR47UH 4 15 School of Environmental Sciences, University of East Anglia, Norwich Research Park, Norwich,
16 Norfolk, UK NR47TJ 5 17 State Key Laboratory of Genetic Resources and Evolution, Kunming Institute of Zoology, Chinese
18 Academy of Sciences, Kunming, Yunnan, China 650223 6 19 Center for Excellence in Animal Evolution and Genetics, Chinese Academy of Sciences, Kunming
20 Yunnan, China 650223
21
22
23 †These authors contributed equally to the manuscript
24 *Correspondence: [email protected], [email protected], [email protected],
25
bioRxiv preprint doi: https://doi.org/10.1101/2020.11.24.395384; this version posted November 24, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
1. Bacteria 2. Density gradient ultracentrifugation of 3. metabolise 13C RNA label: incorporates into bacterial RNA 12C un-labelled “light” RNA 13C labelled glucose
13C labelled “heavy” RNA Increasing buoyant density buoyant Increasing
Fractionation of density gradient RNA extraction from ant laterocervical qPCR and 16S rRNA amplicon 4. Host secretions plates sequencing of fractions to identify taxa become 13C labelled involved in buoyant density shifts 26
27
28 Supplementary Figure 1. Overview of methodology used for RNA stable isotope probing of
29 the propleural plate microbiome of Acromyrmex echinatior leafcutter ants.
30
31
1 mm
PP
DT
A BA BC DC 32
33 Supplementary Figure 2. The extent to which a 20% (w/v) glucose water diet containing
34 green fluorescent dye (A) was distributed over the ant body directly after taking a feed (B), 6
35 hours after feeding (C) and 24 hours after feeding (D). DT = fluorescence shining through from
36 the ant digestive tract, PP = propleural plates with growth of actinobacteria.
37
38
2 bioRxiv preprint doi: https://doi.org/10.1101/2020.11.24.395384; this version posted November 24, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
1.83 12C 1.82 13C 1.81 ) 1 - 1.8
1.79
1.78
1.77 Buoyant Density (g ml (g Density Buoyant
1.76
1.75
1.74 0 1 2 3 4 5 6 7 8 9 10 11 Fraction Number 39
40 Supplementary Figure 3. The relationship between RNA-SIP fraction number and buoyant
-1 41 density (in g ml ). Fractions were generated from density gradients containing RNA isolated 12 13 42 from the propleural plates of A. echinatior ants fed on either a C (blue) or a C (orange)
43 glucose water diet. Points represent averages (three samples each of 22 ants per dietary
44 treatment) ± standard error.
45
46
3 bioRxiv preprint doi: https://doi.org/10.1101/2020.11.24.395384; this version posted November 24, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
100 12C 90 13C 80
70
60
50
sample (%) 40
30
20
10
Percentage of total 16S rRNA gene copy number per copy gene rRNA 16S of total Percentage 0 1.76 1.77 1.78 1.79 1.8 1.81 1.82 Buoyant density (g ml-1) 47
48 Supplementary Figure 4. 16S rRNA gene copy number across different fractions of buoyant
49 density gradients, as determined via qPCR. Gene copy number in each fraction is displayed as
50 a percentage of total copy number in each sample. There were three replicate samples of 22 12 13 51 ants fed a C (light grey) or C (dark grey) glucose diet. Diamond symbols represent fractions
52 that were sent for 16S rRNA gene amplicon sequencing and yellow diamonds represent those
53 designated as “heavy” fractions under the different treatments.
54
4 bioRxiv preprint doi: https://doi.org/10.1101/2020.11.24.395384; this version posted November 24, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
100
80
60
40 Relative abundance (%) Relative 20
0 12CH 13CH Fraction
Pseudonocardia Streptomyces Wolbachia Microbacterium Gordonia Micrococcus Propioniferax Okibacterium Tsukamurella Actinotalea Rhodococcus Micromonospora Sporichthya Microlunatus Yonghaparkia Brevibacterium Agrococcus Actinomadura Jatrophihabitans Lechevalieria Blastococcus Pseudomonas Other 55
56 Supplementary Figure 5. The relative abundances (%) of genera in the “heavy” fractions of 12 13 57 buoyant density gradients for C and C samples, respectively. Heavy fractions were
58 determined via qPCR, which identified peaks in 16S rRNA gene copy number in both sets of
59 samples (see Supplementary Figure 4). Genera reported in Figure 1 (in unfractionated samples)
60 were also recovered in fractionated samples; Pseudonocardia was the most prevalent genus,
61 followed by Streptomyces, Wolbachia and Microbacterium. Additional actinomycete genera
62 were recovered at lower abundances. Genera labelled as “other” were recovered in trace
63 amounts.
64
5 bioRxiv preprint doi: https://doi.org/10.1101/2020.11.24.395384; this version posted November 24, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
4
3
2 Atom % 13C 1
0 12C 13C Glucose diet 65 13 12 13 66 Supplementary Figure 6. The atom percentage of C in ants fed either a C or C labeled
67 20% (w/v) glucose diet for 10 days, as determined by Isotope Ratio Mass Spectrometry (IRMS)
68 analysis.
69
70
71
6 bioRxiv preprint doi: https://doi.org/10.1101/2020.11.24.395384; this version posted November 24, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
Biosynthesis of antibiotics Biosynthesis of antibiotics Biosynthesis of other secondary metabolites Biosynthesis of other secondary metabolites Biosynthesis of secondary metabolites Biosynthesis of secondary metabolitesA B Metabolism of terpenoids and polyketides Metabolism of terpenoids and polyketides 2−Oxocarboxylic acid metabolism 2−Oxocarboxylic acid metabolism Amino acid metabolism Amino acid metabolism Biosynthesis of amino acids Biosynthesis of amino acids Carbohydrate metabolism Carbohydrate metabolism Carbon metabolism Carbon metabolism Degradation of aromatic compounds Degradation of aromatic compounds Energy metabolism Energy metabolism Environmental adaptation Environmental adaptation Kegg pathway category Kegg pathway category Fatty acid metabolism Fatty acid metabolism
(KO) classification (KO) Glycan biosynthesis and metabolism Glycan biosynthesis and metabolism cellular processes cellular processes Lipid metabolism Lipid metabolism environmental information processing environmental information processing Metabolic pathways Metabolic pathways genetic information processing Metabolism of cofactors and vitamins Metabolism of cofactors and vitamins genetic information processing Metabolism of other amino acids Metabolism of other amino acids primary metabolism primary metabolism Microbial metabolism in diverse environments Microbial metabolism in diverse environments Orthology secondary metabolism secondary metabolism Nucleotide metabolism Nucleotide metabolism Xenobiotics biodegradation and metabolism Xenobiotics biodegradation and metabolism Folding, sorting and degradation Folding, sorting and degradation Kegg orthology (KO) classification orthology Kegg (KO) Kegg orthology (KO) classification orthology Kegg (KO) Kegg Replication and repair Replication and repair Transcription Transcription Translation Translation Membrane transport Membrane transport Signal transduction Signal transduction Cell growth and death Cell growth and death Cell motility Cell motility Cellular community − prokaryotes Cellular community − prokaryotes Transport and catabolism Transport and catabolism 0 25000 50000 75000 100000 0 20000 40000 60000
RPKM RPKM Reads per kilobase of transcript per million mapped reads (RPKM)
Kegg Pathway Category Cellular processes Environmental information processing Genetic information processing
Primary metabolism Secondary metabolism 72
73 Supplementary Figure 7. Expression levels (in reads per kilobase of transcript per million
74 mapped reads, RPKM) of Kegg orthology pathway categories. (A) Pseudonocardia
75 octospinosus (colony Ae088) and (B) Pseudonocardia echinatior (colony Ae1083) on the
76 propleural plates of Acromyrmex echinatior ants. N= 1 sample of 80 pooled ants per colony.
77
78
79
A B
80
81 Supplementary Figure 8. The bioactivity of Pseudonocardia isolates (A) P. echinatior PS088
82 and (B) P. octospinosus PS1083 against the specialized fungus-garden pathogen Escovopsis
83 weberi (Supplementary Table 1).
84
7 bioRxiv preprint doi: https://doi.org/10.1101/2020.11.24.395384; this version posted November 24, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
85
86
87 Supplementary Figure 9. Individual growth-rate experiments of Acromyrmex-resident, non-
88 producer strains, assessed as bacterial colony sizes after 5 days at 30 ºC, with the boxplots
89 indicating medians ± one quartile. The white section shows growth rates on control media, and
90 the grey section shows growth rates on the Pseudonocardia-infused media. Red boxes
91 represent non-producer strains, and blue boxes represent producer strains. For analysis, a linear
92 mixed-effects model, including Pseudonocardia strain (n=17) and inoculated bacterial species
93 (n=20) as random factors, was used to test for interactions and main effects of growth media
94 (Control vs. Ps1-infused vs. Ps2-infused) and antibiotic production (Non producers vs.
2 95 Streptomyces). There was no significant interaction effect (c = 2.64, df = 2, p = 0.27), but both
96 main effects were highly significant. The resident non-producers isolated from cuticular
2 97 microbiomes had significantly slower growth rates on all media (c = 20.96, df = 1, p < 0.0001)
98 including the control media without antibiotics. This suggests that they are unable to
99 outcompete producer strains on the cuticle of Acromyrmex ants and raises the question why
100 these non-producer species can persist at all. Bacterial growth was also generally slower on
8 bioRxiv preprint doi: https://doi.org/10.1101/2020.11.24.395384; this version posted November 24, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
2 101 Ps2-infused media than on Ps1-infused media (c = 21.43, df = 1, p < 0.000 when analyzed in
102 a balanced design without the control-media.
103
104 Supplementary Figure 10. Antibiotic resistance profiles for producer, non-producer, and
105 resident non-producer strains (Supplementary Table 1). Boxplots indicate medians (notches)
106 ± one quartile. For analysis, we calculated each strain’s mean growth score across the eight
107 tested antibiotics (reducing from n = 155 to n = 20; details in S3). Producers showed higher
108 levels of resistance than did non-producers for both measurements: Wilcoxon two-sided test
109 (wilcox.test), W = 94.5, p = 0.0017 for LEC (A) and W = 80, p = 0.0253 for MIC (B), after
110 correction for multiple testing. Producers and Resident Non-producers showed no difference
111 in resistance levels (p = 0.44 and 0.25). The ‘rabbit ears’ in B indicate that the medians are also
112 the highest values. Data and details of the analysis are included in the code for Figure 5.
113
114 Supplementary Table 1. Details of ant colonies, bacterial strains and fungal strains, reference
115 genomes and primers used in experiments. *Pseudonocardia strains that have been genome-
116 sequenced. †Pseudonocardia strains that were only used in the growth-rate experiment with
117 non-producers.
9 bioRxiv preprint doi: https://doi.org/10.1101/2020.11.24.395384; this version posted November 24, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
Ant colony name Description Origin Acromyrmex echinatior ant colony Ae1083 harboring Pseudonocardia Gamboa, Panama. octospinosus.
Acromyrmex echinatior ant colony Ae088 harboring Pseudonocardia Gamboa, Panama. echinatior.
Microbial strain description origin
Pseudonocardia octospinosus PS1083 isolated from Acromyrmex Isolated in this study echinatior ants in colony Ae1083.
Pseudonocardia echinatior isolated PS088 from Acromyrmex echinatior ants in Isolated in this study the colony Ae088.
Strain of the parasitic fungus Westerdijk Fungal Biodiversity CBS 810.71 Escovopsis weberi. Institute
8 Pseudonocardia Ps1 strains
Ae356* Derived from colony Ae356 Holmes, Innocent 1
Ae263* Derived from colony Ae263 Holmes, Innocent 1
Ae322 Derived from colony Ae322 This Study
Ae150A* Derived from colony Ae150A Holmes, Innocent 1
Ae168* Derived from colony Ae168 Holmes, Innocent 1
Ae707-CP-A2* Derived from colony Ae707-CP-A2 Holmes, Innocent 1 (Ae707_Ps1)
Ae712† Derived from colony Ae712 This Study
Ae280† Derived from colony Ae280 This Study
10 bioRxiv preprint doi: https://doi.org/10.1101/2020.11.24.395384; this version posted November 24, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
11 Pseudonocardia Ps2 strains
Ae406* Derived from colony Ae406 Holmes, Innocent 1
Ae160 Derived from colony Ae160 This Study
Ae505* Derived from colony Ae505 Holmes, Innocent 1
Ae717* Derived from colony Ae717 Holmes, Innocent 1
Ae703 Derived from colony Ae703 This Study
Ae702 Derived from colony Ae702 This Study
Ae707 Derived from colony Ae707 This Study
Ae331* Derived from colony Ae331 Holmes, Innocent 1
Ae706* Derived from colony Ae706 Holmes, Innocent 1
Ae704 Derived from colony Ae704 This Study
Ae715 Derived from colony Ae715 This Study
10 environmental antibiotic-producing strains (all Streptomyces)
S1. Streptomyces Streptomyces coelicolor M145 ∆act John Innes Centre, Norwich, M1146 ∆red ∆cpk ∆cda NR4 7UH, UK 2
John Innes Centre, Norwich, S2. S. lividans 66 Soil derived Streptomyces species NR4 7UH, UK 3
Soil derived Streptomyces, SCP1- John Innes Centre, Norwich, S3. S. coelicolor SCP2- Pgl+ NR4 7UH, UK 4 M145
S4. S. scabies 87- Soil derived Streptomyces species Bignell et al. 5 22
USDA ARS Culture Collection. S5. S. venezuelae Soil derived Streptomyces species https://nrrl.ncaur.usda.gov/cgi- NRRL B-65442 bin/usda Strain no. B-65442.
S6. S. Ae150A- Streptomyces derived from lab This study B1 workers of captive colony Ae150A
11 bioRxiv preprint doi: https://doi.org/10.1101/2020.11.24.395384; this version posted November 24, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
Streptomyces derived from lab This study S7. S. Ae356-S1 workers of captive colony Ae356
S8. S. formicae Tetraponera penzigi derived Seipke et al.6, Holmes et al 7 KY5 Streptomyces species
Acromyrmex octospinosus derived Barke et al.8; Seipke et al.9 S9. S. S4 Streptomyces albidoflavus S4
S10. S. S4 Streptomyces albidoflavus S4 Seipke et al. 10 ∆antA::apr ∆antA::apr
10 environmental non-producer strains
St1. Escherichia Non-pathogenic ESKAPE ATCC® 11775TM coli laboratory screening strain
Handelsman Lab, Small World St2. Lysobacter Non-pathogenic ESKAPE Initiative, University of antibioticus laboratory screening strain Wisconsin-Madison, USA
Handelsman Lab, Small World St3. Bacillus Non-pathogenic ESKAPE Initiative, University of subtilis laboratory screening strain Wisconsin-Madison, USA
St4. Handelsman Lab, Small World Non-pathogenic ESKAPE Pseudomonas Initiative, University of laboratory screening strain putida Wisconsin-Madison, USA
Handelsman Lab, Small World St5. Erwinia Non-pathogenic ESKAPE Initiative, University of caratova laboratory screening strain Wisconsin-Madison, USA
St6. Enterobacter Non-pathogenic ESKAPE ATCC® 51697TM aerogenes laboratory screening strain
St7. Non-pathogenic ESKAPE Acinetobacter ATCC® 33305TM laboratory screening strain baylyi
12 bioRxiv preprint doi: https://doi.org/10.1101/2020.11.24.395384; this version posted November 24, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
St8. Non-pathogenic ESKAPE Staphylococcus ATCC® 14990TM laboratory screening strain epidermidis
St9. Micrococcus Micrococcus luteus (NCTC2665, ATCC® 4698TM luteus “Fleming strain”)
St10. Serratia Tetraponera penzigi derived Seipke et al.6 KY15 Serratia
10 Acromyrmex-resident non-producer strains
Sr1. This study Acromyrmex derived Ochrobactrum sp
Isolated from large A. echinatior This study Sr2. Erwinia sp. worker ants in lab colonies
Sr3. Isolated from large A. echinatior Acinetobacter sp. worker ants in lab colonies This study
Sr4. Isolated from large A. echinatior Sphingobacterium worker ants in lab colonies This study sp.
Sr5. Isolated from large A. echinatior Acinetobacter sp. worker ants in lab colonies This study
Sr6. Luteibacter Isolated from large A. echinatior This study sp. worker ants in lab colonies
Sr7. Isolated from large A. echinatior Flavobacterium worker ants in lab colonies This study sp.
Sr8. Isolated from large A. echinatior Brevundimonas worker ants in lab colonies This study sp.
Sr9. Isolated from large A. echinatior Acinetobacter sp. worker ants in lab colonies
13 bioRxiv preprint doi: https://doi.org/10.1101/2020.11.24.395384; this version posted November 24, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
This study
Sr10. Isolated from large A. echinatior Brachybacterium worker ants in lab colonies This study sp.
Genome Description Accession number/ reference
Acromyrmex Whole genome shotgun sequencing AEVX00000000, Nygaard, echinatior project of the A. echinatior genome. Zhang 11
Whole genome shotgun sequencing of a wild-type isolate of MCIR00000000, Holmes, Ae707 Pseudonocardia octospinosus, Innocent 1 isolated from the cuticle of a large worker.
Whole genome shotgun sequencing of a wild-type isolate of MCIQ00000000, Holmes, Ae706 Pseudonocardia echinatior, isolated Innocent 1 from the cuticle of a large worker.
Primer Sequence Reference
5’-CCTACGGG 341F Amplifies the V3 region of the AGGCAGCAG-3’ 16S rRNA gene 12. 518R 5’- ATTACCGCGGCTGCTGG -3’
118
119
120
121
122
123
124
125
126
14 bioRxiv preprint doi: https://doi.org/10.1101/2020.11.24.395384; this version posted November 24, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
127 Supplementary Table 2. The total number of RNA-sequencing reads (and percentage of total
128 reads in brackets) originating from propleural plate samples taken from the ant colonies
129 Ae1083 or Ae088, respectively, that successfully aligned to the A. echinatior genome
130 (Supplementary Table 1) and to the genomes of the Pseudonocardia species associated with
131 the ant colony of origin (P. octospinosus or P. echinatior, respectively).
132
Alignment
Sample Source A. echinatior P. octospinosus P. echinatior genome genome Ae707 genome Ae706
80 pooled sets of 8,548,640 103,820 Ae1083 propleural plates from - (78.7 %) (1.0 %) large workers
80 pooled sets of 7,058,678 189,989 Ae088 propleural plates from - (73.5 %) (2.0 %) large workers
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
15 bioRxiv preprint doi: https://doi.org/10.1101/2020.11.24.395384; this version posted November 24, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
150 Supplementary Table 3. Secondary metabolite BGCs in the Pseudonocardia mutualist 1 151 genomes (table adapted from Holmes et al. ) and their associated expression values (in reads
152 per kilobase of transcript per million mapped reads, RPKM) in RNA-sequencing experiments.
153 Yellow rows are BGCs shared between strains. Green and blue represent BGCs that are unique
154 to P. octospinosus and P. echinatior, respectively.
155
Cluster number Cluster RPKM expression values Classification P. octospinosus P. echinatior code P. octospinosus P. echinatior
1 7 Oligosaccharide A 57.05 45.47 5 5 Terpene B 81.19 37.58 7 4 Nystatin C 1.76 24.28
8 2 Terpene D 341.82 424.23 11 9 Bacteriocin E 118.95 80.24 14 8 Ectoine F 453.11 192.08 2 - Other G 34.34 - 3 - Other H 110.18 - 4 - NRPS I 10.97 - 6 - Bacteriocin J 111.07 - 9 - NRPS K 41.92 - 12 - Other L 52.55 - 13 - NRPS M 25.64 - - 1 Other N - 16.75
- 3 T1PKS-NRPS O - 16.87 - 6 Bacteriocin P - 91.71 - 10 Terpene Q - 23.75 - 11 Bacteriocin R - 30.50
156
157
158
159
16 bioRxiv preprint doi: https://doi.org/10.1101/2020.11.24.395384; this version posted November 24, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
160 Supplementary Table 4. Media recipes and antibiotics used in this study.
161
-1 Medium Name Component g L dH2O
Soy flour 20 Soya Flour Mannitol Mannitol 20 (SFM) Agar Agar 20
Potato Glucose Agar PGA (Sigma Aldrich) 39 (PGA)
Glucose 4
Yeast extract 4 Glucose, Yeast, Malt Malt extract 10 (GYM) Agar CaCO3 2
Agar 15
Tryptone 10
Lennox Broth (LB) NaCl 10
Yeast extract 5
Antibiotic Concentrations Used Target Compound Type
Chloramphenicol 0, 2.5, 5, 10, 25, 50 Translation Synthetic µg/ml inhibitor
Rifampicin 0, 0.5, 1, 2, 5, 10 µg/ml Translation Polyketide inhibitor
Streptomycin 0, 5, 10, 20, 50, 100 Translation Aminoglycoside µg/ml inhibitor
Vancomycin 0, 1, 2, 4, 10, 20 µg/ml Cell wall Glycopeptide synthesis inhibitor
17 bioRxiv preprint doi: https://doi.org/10.1101/2020.11.24.395384; this version posted November 24, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
Phosphomycin 0, 5, 10, 20, 50, 100 Cell wall Small molecule µg/ml synthesis inhibitor
Nalidixic Acid 0, 2.5, 5, 10, 25, 50 DNA gyrase Synthetic µg/ml inhibitor quinolone
Apramycin 0, 5, 10, 20, 50, 100 Translation Aminoglycoside µg/ml inhibitor
Ampicillin 0, 10, 20, 40, 100, 200 Cell wall ß lactam µg/ml synthesis inhibitor
162
163
164 Supplementary Note 1: Supplementary methods
165
13 13 166 Isotope ratio mass spectrometry analysis. The C composition of ants fed on a C-labelled
167 diet was determined by using a coupled Delta plus XP Isotope Ratio Mass Spectrometer/Flash
168 HT Plus Elemental Analyser (Thermo Finnigan) in the University of East Anglia Analytical
13 169 Facility. Ants were fed on a 5% glucose solution (w/v) for 10 days; five ants were fed a C
12 170 glucose solution and five were fed on a C glucose solution. After 10 days, ants were washed
171 once in 70% ETOH, then sequentially in sterile dH2O before drying on filter paper. The ants
172 were then flash frozen and stored at -80°C until being placed in a ScanVac Coolsafe freeze
173 dryer for 5 days. Each ant was then put into an individual 75 µl tin capsule (Elemental
174 Microanalysis); capsules were loaded into an automatic sampler and completely converted to
175 CO2, N2 and H2O through combustion in an excess of oxygen (oxidation was carried out at
176 1020°C, followed by reduction at 650°C). Nitrous oxides formed during combustion where
177 reduced using Cu. Helium was used as a carrier gas. After passing through a water trap
178 (MgClO4), the gases were separated chromatographically on an isothermal GC column
18 bioRxiv preprint doi: https://doi.org/10.1101/2020.11.24.395384; this version posted November 24, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
179 (Thermo PTFE, 0.8m, 50°C); the resulting peaks sequentially entered the ion source of the
180 Isotope Ratio Mass Spectrometer. Gas species were then measured using a Faraday cup
181 universal collector array, with masses of 44, 45 and 46 being monitored for the analysis of CO2.
182 Casein and collagen were used to calibrate the system and normalize the data post run; these
183 standards have been calibrated against international certified standards and have an assigned
13 184 δ C value. Empty tin capsules were used as blanks. Each sample was analysed in triplicate.
13 13 185 The C content of samples was reported as the C atom percent, which was calculated using
186 the following formula:
13 12 13 187 ( C/ C+ C)*100
188
189 Fluorescent microscopy of ant feeding habits. To confirm that the glucose water diet did
190 not spread over the ant cuticle, a 20% glucose solution labelled with non-toxic fluorescent
-1 191 green drain tracing dye (Hydra) was fed to ants. Five mg ml of dye was added to a 20%
192 glucose solution, of which 300 µl was supplied to ants in the cap of an Eppendorf. Ants were
193 sampled just after taking a feed, and following 6 and 24 hours of being exposed to the dye, to
194 trace the spread of the solution over time. After sampling, ants were carefully fixed on their
195 backs and brightfield and fluorescent images were acquired using a Zeiss M2 Bio Quad SV11
196 stereomicroscope. The samples were illuminated either with a halogen lamp (brightfield) or a
197 100W Hg arc lamp (fluorescence) and reflected-light images were captured with an AxioCam
198 HRc CCD camera and AxioVision software (Carl Zeiss, Cambridge, UK). Green
199 fluorescence was excited with light passed through a 470 nm filter (40 nm bandpass) and the
200 emission was collected through a 525 nm filter (50 nm bandpass).
201
19 bioRxiv preprint doi: https://doi.org/10.1101/2020.11.24.395384; this version posted November 24, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
202 RNA extraction protocol. A modified version of the Qiagen RNeasy Micro Kit protocol was
203 used for all RNA extractions. Briefly, 700 µl of RLT buffer (with 1% beta mercaptoethanol)
204 was added to each lysis matrix E tube before the samples could thaw. Tubes were then placed
205 in a FastPrep-24™ 5G benchtop homogenizer (MP Biomedicals) and disrupted for 40
206 seconds at 6 m/s. Samples were then centrifuged for 2 minutes at 13’000 rpm and the
207 supernatant was collected into a QIAshredder tube. This was centrifuged for 2 minutes at
208 13’000 rpm to homogenize the lysate. The resulting flow-through was mixed vigorously with
209 700 µl acidic phenol chloroform, then allowed to rest for 3 minutes at room temperature
210 before centrifugation for 20 minutes at 13’000 rpm. The upper phase was then collected and a
211 50% volume of 96% ethanol was added. The mixture was then placed into a minelute column
212 supplied with the Qiagen RNeasy Micro Kit. The kit protocol (including the on-column
213 DNase I treatment) was then followed through to elution of the RNA, at which point 50 µl of
214 RNase free water (heated to 37°C) was added to the column membrane and incubated at
215 37°C for 5 minutes, before centrifuging for one minute at 13’000 rpm to elute the RNA. To
216 remove any remaining DNA, RNA was treated with the turbo DNase kit (Invitrogen): 5 µl of
217 10x buffer and 2 µl of Turbo DNase was added to 50 µl of RNA and incubated at 37°C for 25
218 minutes. RNA was then purified using the Qiagen Micro RNA Kit clean-up protocol.
219
220 Density gradient ultracentrifugation and fractionation. Density gradient ultracentrifugation
13 221 was carried out to separate C labelled (“heavy”) from un-labelled (“light”) RNA within the
222 same RNA sample. To make one complete gradient solution for the ultracentrifugation of one
223 RNA sample, 4.5 ml of caesium trifluoroacetate (CsTFA, ~2 g ml-1, GE Healthcare, Munich,
224 Germany) was added to 850 µl of gradient buffer and 197.5 µl formamide. Gradient buffer was
13 225 made using an established protocol , after which 270 ng of RNA from one replicate sample
226 was added to the gradient solution and the refractive index (R.I) of a 60 µl aliquot measured
20 bioRxiv preprint doi: https://doi.org/10.1101/2020.11.24.395384; this version posted November 24, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
227 using a refractometer (Reichert Analytical Instruments, NY, USA). R.I was normalized to
228 1.3725 (approximately 1.79 g ml-1 CsTFA). Samples underwent ultracentrifugation in a
229 Beckman Optima XL-100K ultracentrifuge for 50 hours at 20°C, 38’000 rpm with a vacuum
230 applied, using a Vti 65.2 rotar (Beckman Coulter, CA, USA). Deceleration occurred without
231 brakes. Following centrifugation, samples were divided into 12 fractions using a peristaltic
14 232 pump to gradually displace the gradient, according to an established protocol . The R.I of
233 fractions was measured to confirm the formation of a linear density gradient. RNA was
234 precipitated from fractions by adding 1 volume of DEPC-treated sodium acetate (1M, pH 5.2),
235 1 µl (20 µg) glycogen (from mussels, Sigma Aldrich) and 2 volumes of ice cold 96% ETOH.
236 Fractions were incubated over night at -20°C then centrifuged for 30 minutes at 4°C at 13’000
237 x g, before washing with 150 µl of ice cold 70% ETOH and centrifuging for a further 15
238 minutes. Pellets were then air-dried for 5 minutes and re-suspended in 15 µl of nuclease free
239 water.
240
241 QPCR experiments. QPCR experiments were carried out to enable the quantification of 16S
242 rRNA gene copy number across the cDNA fractions resulting from density gradient
243 ultracentrifugation of RNA-SIP samples. 1 µl of either template cDNA, standard DNA, or
244 dH2O as a control, was added to 24 µl of reaction mix containing 12.5 µl of 2x Sybr Green
245 Jumpstart Taq Ready-mix (Sigma Aldrich), 0.125 µl each of the primers PRM341F and 518R
-1 246 (Supplementary Table 1), 4 µl of 25 mM MgCl2, 0.25 µl of 20 µg µl Bovine Serum Albumin
247 (Sigma Aldrich), and 7 µl dH2O. Sample cDNA, standards (a dilution series of the target 16S
248 rRNA gene at known quantities), and negative controls were quantified in duplicate. Reactions
249 were run under the following conditions: 95°C for 10 mins; 40 cycles of 95°C for 15 sec, 55°C
250 for 30 sec, and 72°C for 30 sec; plate read step at 83.5°C for 10 seconds (to avoid primer
21 bioRxiv preprint doi: https://doi.org/10.1101/2020.11.24.395384; this version posted November 24, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
251 dimers); 96°C for 15 sec; 100 cycles at 55°C-95°C for 10 secs, ramping 0.5°C per cycle,
252 followed by a plate read. Reactions were performed in 96-well plates (Bio-Rad). The threshold
253 cycle (CT) for each sample was then converted to target molecule number by comparing to CT
254 values of a dilution series of target DNA standards.
255
256 Processing of reads generated from RNA sequencing experiments. The quality of Illumina
257 sequences (returned from Vertis Biotechnologie AG) was assessed using the program FastQC
258 (Babraham Institute, Cambridge, UK), before using TrimGalore version 0.4.5 (Babraham
259 Institute, Cambridge, UK) to trim Illumina adaptors and low quality base calls from the 3’ end
260 of reads (an average quality phred score of 20 was used as cut-off). After trimming, sequences
261 shorter than 20 base pairs were discarded. Trimmed files were then aligned to the reference
11 262 genome for Acromyrmex echinatior , Supplementary Table 1, and the appropriate
263 Pseudonocardia genome (either Ae707 for the sample from colony Ae1083, or Ae706 for the
1 264 sample from colony Ae088 , see Supplementary Table 1 for genome information). All
15 265 alignments were done using the splice-aware alignment program HiSat2 with the default
266 settings. For each cuticular sample, reads that had mapped successfully to their respective
267 Pseudonocardia genomes (Supplementary Table 3) were then mapped back to the ant genome
268 (and vice versa) to check that reads did not cross-map between the two genomes (i.e that they
269 were either uniquely ant or bacterial reads) - reads that did not cross-map were retained for
16 270 downstream analysis. Following alignment, the program HTSeq was used to count mapped
271 reads per annotated coding sequence (CDS) using the General Feature Format (GFF) file
272 containing the annotated gene coordinates for each reference genome. Reads that mapped to
273 multiple locations within a genome were discarded at this point and only uniquely mapped
274 reads were used in the counting process. Read counts per CDS were then converted to reads
275 per kilobase of exon model per million reads (RPKM) by extracting gene lengths from the GFF
22 bioRxiv preprint doi: https://doi.org/10.1101/2020.11.24.395384; this version posted November 24, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
276 file. Converting reads to RPKM values normalizes counts for RNA length and for differences
277 in sequencing depth, which enables more accurate comparisons both within and between
17 278 samples .
279
280 Isolation of Pseudonocardia and Escovopsis bioassays. The Pseudonocardia strains PS1083
281 and PS088 (Supplementary Table 1) were isolated from the propleural plates of individual
282 Acromyrmex echinatior leafcutter ants taken from colonies Ae1083 and Ae088
283 (Supplementary Table 1), respectively. Similarly, Pseudonocardia strains Ae322, Ae712,
284 Ae280, Ae160, Ae703, Ae702, Ae707, Ae704 and Ae715 were isolated from large worker
285 ants from colonies with the same labels maintained at the University of Copenhagen (these
286 strains were only used for the growth-rate experiments). A sterile needle was used to scrape
287 bacterial material off the propleural plates on the ventral part of the thorax; this was then
288 streaked over Soya Flour Mannitol (SFM, Supplementary Table 4) agar plates and incubated
289 at 30°C. Resulting colonies resembling Pseudonocardia were purified by repeatedly
290 streaking single colonies onto SFM agar plates. Spore stocks were created using an
18 291 established protocol . The taxonomic identity of each Pseudonocardia isolate was confirmed
1 292 via colony PCR and 16S rRNA sequencing, as described in Holmes et al. . Each resulting
293 sequence was also aligned to both the Pseudonocardia octospinosus and Pseudonocardia
1 294 echinatior 16S rRNA gene sequences to reveal their percentage identities to each of the two
295 species. Antifungal bioassays were carried out using the Escovopsis weberi strain CBS
296 810.71, acquired from the Westerdijk Fungal Biodiversity Institute (Supplementary Table 1).
297 E. weberi was actively maintained on potato glucose agar (PGA, Supplementary Table 4) at
298 room temperature. Fungal mycelia were transferred to a fresh plate every month. For
299 bioassays, a plug of actively growing mycelium was transferred from PGA plates to the edge
300 of a Glucose Yeast Malt (GYM, Supplementary Table 4) agar plate with a growing
23 bioRxiv preprint doi: https://doi.org/10.1101/2020.11.24.395384; this version posted November 24, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
301 Pseudonocardia colony, using the end of a sterile glass Pasteur pipette. Plates were then left
302 at room temperature for 2 weeks. A zone of clearing around the Pseudonocardia colony
303 indicated the presence of antifungal activity. Three replicate experiments (with three replicate
304 bioassay plates per Pseudonocardia strain) were carried out, whereby different E. weberi
305 starter plates were used as an inoculum.
306
307 Individual growth-rate experiments. To create the Pseudonocardia-infused media, lawns of
308 each of the 19 isolates were grown (Supplementary Table 1), plating 30 μl of spores (in 20%
309 glycerol) onto 90 mm SFM agar plates (Supplementary Table 4). The control plates were
310 inoculated with 20% glycerol only. We incubated these plates at 30 °C for 6 weeks, which
311 ultimately produced confluent lawns from 17 strains that could be included in the experiments
312 (6 Ps1, 11 Ps2). Once each plate was fully covered, we flipped the agar to reveal a surface open
313 for colonisation. The 10 environmental producer strains and the 10 environmental non-
314 producer strains (Supplementary Table 1) were inoculated onto the plates. Each of the plates
315 received 10 evenly spaced colonies, with 3 replicates, generating 2 invader types x 17 Ps-
316 media-types x 3 replicates = 102 Ps-infused plates and 2 invader types x 3 replicates = 6 control
317 plates, for 1020 treatment and 60 control inoculations. Each strain inoculation used 5 μl of
318 solution (approx. 1 x 106 cells per ml in 20% glycerol), spotted at evenly spaced positions and
319 without coming into direct contact. All plates were incubated at 30 °C for five days, after which
320 photographs were taken.
19, 20 321 Images were processed in Fiji software , creating binary negatives (black & white)
322 so automated tools could identify discrete areas of growth (black) and measure growth areas
323 for each invading strain; in the few cases where binary image resolution was insufficient,
324 outlines were added manually before area calculation. 48 producer-inoculated and 57 non-
325 producer-inoculated treatment measurements were excluded because plate condition had
24 bioRxiv preprint doi: https://doi.org/10.1101/2020.11.24.395384; this version posted November 24, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
326 deteriorated to become unscorable or they were contaminated, leaving final sample sizes of
327 1020-48-57= 915 treatment inoculations and 60 controls.
328 The second growth-rate experiment compared the 10 Acromyrmex-resident, non-
329 producer strains with 9 of the environmental producer strains (1 of the 10 inoculations failed
330 to grow). All 19 Pseudonocardia strains grew sufficiently to be included in this experiment,
331 and each plate was again inoculated with 10 or 9 evenly spaced colonies. Starting sample sizes
332 were therefore 2 invader types x 19 Ps-media-types x 3 replicates = 114 Ps-infused plates and
333 2 x 3 = 6 control plates, for 1083 treatment and 57 control inoculations. Fifty producer and 20
334 non-producer treatment measurements were excluded for the same reasons as above, leaving
335 final sample sizes of 1083-50-20=1013 treatment and 57 control inoculations, scored as above.
336
337 Pairwise competition experiment. Experiments were set up to test whether antibiotic-
338 producer strains could win in direct competition against non-producing strains, both on
339 normal media and on media infused with Pseudonocardia secondary metabolites. To create
340 the Pseudonocardia-infused media, we plated 30 μl of spores (in 20% glycerol) onto 50 mm
341 SFM agar plates (Supplementary Table 4). The control plates were inoculated with 20% (v/v)
342 glycerol only. We incubated these plates at 30 °C for 6 weeks, which ultimately produced
343 confluent lawns. As above, the agar was flipped to reveal a surface open for colonisation.
344 Environmental producers and non-producers were then coinoculated onto these media (as
345 well as on control media with no Pseudonocardia present), and we measured the outcome of
346 competition as a win, loss or draw. To keep the number of tests manageable, we used two
347 combinations of Pseudonocardia-infused media and Streptomyces: Pseudonocardia
348 octospinosus (strain Ae707-CP-A2) + Streptomyces S8 and Pseudonocardia echinatior
349 (strain Ae717) + Streptomyces S2 (Supplementary Table S1). We competed the two
350 Streptomyces strains (S2, S8) against the 10 environmental non-producer strains (20
25 bioRxiv preprint doi: https://doi.org/10.1101/2020.11.24.395384; this version posted November 24, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
6 351 pairings). Each Strepromyces strain was prepared as 10 spores per ml in 20% glycerol. Each
352 non-producer strains was grown overnight in 10 ml of Lennox broth (Supplementary Table
353 4), before subculturing (1:100 dilution) into 10 ml of fresh Lennox, and incubating at 37 °C
9 354 for 3-4 hours. The OD600 was then measured, assuming that OD600 = 1 represented 8 × 10
6 355 cells. Similar dilutions of 10 cells per ml were made for each non-producer strain in 20%
356 (v/v) glycerol, after which producer and non-producer preparations were mixed at a ratio of
4 357 1:1 (v/v) and co-inoculated as a mixture of 20 µl (10 spore-cells of each) on the designated
358 Pseudonocardia-infused media with 5 replicates per pairing. We used 150 plates for the S8
359 experiment (including 100 control plates; 10 replicates per pairing) and 100 plates for the S2
360 experiment (including 50 control plates; 5 replicates per pairing). Plates were incubated at 30
361 °C for 5 days before imaging, after which images were scored with respect to the producer
362 as: win (dominant growth), draw (both strains growing with no clear dominant) or lose (little
363 or no visible growth), always with reference to images of each strain grown alone on control
364 medium to minimise observer bias. One plate’s outcome was too ambiguous to score and was
365 discarded. All plates were independently scored by two observers, one using photos of the
366 original images, which produced datasets giving the same statistical results. We report the
367 direct observer’s scores (Supplementary Methods). For analysis, draw outcomes were
368 omitted, and a general linear mixed-effects model, including non-antibiotic-producer strain as
369 a random intercept (10 groups), was used to test for an effect of the medium term (Control vs.
370 Ps1/2-infused) on competitive outcome (Win vs. Loss) ((lme4::glmer(outcome ~ medium +
371 (1 | non.producer.strain), family = binomial)). Significance was estimated using term
372 deletion.
373
374 Antibiotic resistance assays. The key assumption of screening theory is that antibiotic-
375 producers are better at resisting antibiotics, as measured by growth rates in the presence of
26 bioRxiv preprint doi: https://doi.org/10.1101/2020.11.24.395384; this version posted November 24, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
376 antibiotics, because this correlation is what allows producer strains to better endure the
377 demanding environment produced by Pseudonocardia. We tested this assumption by growing
378 the 10 environmental producer strains, the 10 environmental non-producer strains, and the 10
379 resident non-producers strains (Supplementary Table 1) in the presence of 8 different
380 antibiotics (Supplementary Table 4), representing a range of chemical classes and modes of
381 action. Antibiotics were added to 1 ml of LB-Lennox/SFM medium (Supplementary Table 4)
382 in a 24-well microtitre plate at 6 different concentrations. The relative concentration range
383 was the same for each antibiotic, although actual concentrations reflected activity
384 (Supplementary Table 4). Producers and non-producers were inoculated onto plates and
385 incubated at 30 °C for 7 days, then photographed. Lowest Effective Concentration (LEC,
386 lowest concentration with inhibitory effect) and Minimum Inhibitory Concentration (MIC,
387 lowest concentration with no growth) scores were assigned on a Likert scale of 1–6, where 1
388 was no resistance and 6 was resistance above the concentrations tested (adapted from
21 389 generalized MIC methods; reviewed by Balouiri et al. ).
390
391 Data Analysis. R markdown-format scripts, input datafiles, and html output files for the
392 analyses in Main Text Figures 4, 5, and 6, and in Supplementary Figure 9 are provided as a
393 single R project folder at github.com/dougwyu/Worsley_et_al_screening_test_R_code
394
395 References
396 1. Holmes NA, et al. Genome analysis of two Pseudonocardia phylotypes associated with 397 Acromyrmex leafcutter ants reveals their biosynthetic potential. Front Microbiol 7, 398 2073 (2016).
399
400 2. Gomez-Escribano JP, Bibb MJ. Engineering Streptomyces coelicolor for heterologous 401 expression of secondary metabolite gene clusters. Microb Biotechnol 4, 207-215 402 (2011).
27 bioRxiv preprint doi: https://doi.org/10.1101/2020.11.24.395384; this version posted November 24, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
403
404 3. Hopwood DA, Kieser T, Wright HM, Bibb MJ. Plasmids, Recombination and 405 Chromosome Mapping in Streptomyces lividans 66. Microbiology 129, 2257-2269 406 (1983).
407
408 4. Bentley SD, et al. Complete genome sequence of the model actinomycete 409 Streptomyces coelicolor A3(2). Nature 417, 141-147 (2002).
410
411 5. Bignell DR, Seipke RF, Huguet-Tapia JC, Chambers AH, Parry RJ, Loria R. Streptomyces 412 scabies 87-22 contains a coronafacic acid-like biosynthetic cluster that contributes to 413 plant-microbe interactions. Mol Plant Microbe Interact 23, 161-175 (2010).
414
415 6. Seipke RF, Barke J, Heavens D, Yu DW, Hutchings MI. Analysis of the bacterial 416 communities associated with two ant-plant symbioses. Microbiologyopen 2, 276-283 417 (2013).
418
419 7. Holmes NA, Devine R, Qin Z, Seipke RF, Wilkinson B, Hutchings MI. Complete genome 420 sequence of Streptomyces formicae KY5, the formicamycin producer. J Biotechnol 265, 421 116-118 (2018).
422
423 8. Barke J, et al. A mixed community of actinomycetes produce multiple antibiotics for 424 the fungus farming ant Acromyrmex octospinosus. BMC Biol 8, (2010).
425
426 9. Seipke RF, et al. A single Streptomyces symbiont makes multiple antifungals to support 427 the fungus farming ant Acromyrmex octospinosus. PLoS One 6, e22028 (2011).
428
429 10. Seipke RF, Patrick E, Hutchings MI. Regulation of antimycin biosynthesis by the orphan 430 ECF RNA polymerase sigma factor sigma (AntA.). PeerJ 2, e253 (2014).
431
432 11. Nygaard S, et al. The genome of the leaf-cutting ant Acromyrmex echinatior suggests 433 key adaptations to advanced social life and fungus farming. Genome Res 21, 1339- 434 1348 (2011).
435
436 12. Muyzer G, de Waal EC, Uitterlinden AG. Profiling of complex microbial populations by 437 denaturing gradient gel electrophoresis analysis of polymerase chain reaction- 438 amplified genes coding for 16S rRNA. Appl Environ Microbiol 59, 695-700 (1993).
439
28 bioRxiv preprint doi: https://doi.org/10.1101/2020.11.24.395384; this version posted November 24, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
440 13. Lueders T, Manefield M, Friedrich MW. Enhanced sensitivity of DNA- and rRNA-based 441 stable isotope probing by fractionation and quantitative analysis of isopycnic 442 centrifugation gradients. Environ Microbiol 6, 73-78 (2004).
443
444 14. Whiteley AS, Thomson B, Lueders T, Manefield M. RNA stable-isotope probing. Nat 445 Protoc 2, 838-844 (2007).
446
447 15. Kim D, Langmead B, Salzberg SL. HISAT: a fast spliced aligner with low memory 448 requirements. Nat Methods 12, 357-360 (2015).
449
450 16. Anders S, Pyl PT, Huber W. HTSeq- a Python framework to work with high-throughput 451 sequencing data. Bioinformatics 31, 166-169 (2015).
452
453 17. Mortazavi A, Williams BA, McCue K, Schaeffer L, Wold B. Mapping and quantifying 454 mammalian transcriptomes by RNA-Seq. Nat Methods 5, 621-628 (2008).
455
456 18. Kieser T, Bibb MJ, Buttner MJ, Chater KF, Hopwood DA. Practical Streptomyces 457 Genetics. John Innes Foundation (2000).
458
459 19. Rueden CT, et al. ImageJ2: ImageJ for the next generation of scientific image data. 460 BMC Bioinformatics 18, 529 (2017).
461
462 20. Schindelin J, et al. Fiji: an open-source platform for biological-image analysis. Nat 463 Methods 9, 676-682 (2012).
464
465 21. Balouiri M, Sadiki M, Ibnsouda SK. Methods for in vitro evaluating antimicrobial 466 activity: A review. J Pharm Anal 6, 71-79 (2016).
467
468
469
29