Changes in the genetic requirements for microbial interactions with increasing community complexity

Microbial community structure and function rely on complex interactions whose underlying molecular mechanisms are poorly understood. To investigate these interactions in a simple microbiome, we introduced E. coli into an experimental community based on a cheese rind and identified the differences in E. coli’s genetic requirements for growth in interactive and non-interactive contexts using Random Barcode Transposon Sequencing (RB-TnSeq) and RNASeq. Genetic requirements varied among pairwise growth conditions and between pairwise and community conditions. Our analysis points to mechanisms by which growth conditions change as a result of increasing community complexity and suggests that growth within a community relies on a combination of pairwise and higher-order interactions. Our work provides a framework for using the model organism E. coli as a readout to investigate microbial interactions regardless of the genetic tractability of members of the studied ecosystem.


22
Microorganisms rarely grow as single isolated species but rather as part of diverse 23 microbial communities. In these communities, bacteria, archaea, protists, viruses and fungi can score >= 3). We then removed genes with positive fitness (day 1 n= 10, day 2 n=14 and day 3 6 trxA and waaP) were associated with high fitness value variability across replicate experiments 156 and thus they were associated with a lower z-score. Altogether, this supports the reliability and 9 Figure 2: Analysis of E. coli essential genes during growth in pairwise conditions. We grew E. coli in pairwise 255 conditions with either with H. alvei, G. candidum or P. camemberti (A). Using the E. coli RB-TnSeq library we 256 identified the essential genes for growth on CCA in each pairwise condition. Requirements for growth with the 257 different partners overlapped between the conditions. Altogether, it constitutes 153 essential genes (B). Comparing 258 these genes to the genes essential for E. coli growth alone, we identified 82 genetic requirements that were no longer 259 required for growth in pairwise conditions as well as 75 genes that were added by pairwise growth (C). We mapped 260 the pairwise-specific and alone-specific genes to the KEGG BRITE database. 45 out of 82 genes and 33 out of 77 261 had hits (D). Most of the alone-specific requirements were mapped to metabolic pathways and especially amino-acid 262 metabolism (n=20 genes) while pairwise-specific requirements mapped to metabolic functions but also to functions associated with response to stress (biofilm formation n=5, signal transduction n=5 and drug resistance n=4).

265
Identification of E. coli genes essential for growth within the community and comparison 266 with genes essential for growth in pairwise conditions.

267
We next aimed to investigate the differences between genes essential for growth in a 268 community (complex interactive condition) and genes essential for growth in associated

274
We first compared the genes essential for growth in the community with the genes 275 essential for growth alone. We identified 89 genes that were essential for both conditions, 37 276 genes only essential with the community and 71 genes that were no longer essential to grow with the community (essential only for growth alone) ( Figure 3A). As for the presence of a single 278 partner, the presence of a complex community potentially introduces new essential genes for 279 growth while relieving some requirements.

280
By comparing the genes required for growth in interactive and non-interactive conditions, 281 we have identified the set of genes which are newly required during growth in a community 282 versus growth alone and the set of genes newly required for growth in pairwise conditions 283 versus growth alone. Comparing these two sets of additional genes can reveal if and how 284 community complexity modifies the genes that are essential in different interactives contexts 285 compared to growth alone. We identified 29 essential genes that were potentially newly  Meanwhile, we identified 46 genes that were essential only for growth in pairwise 291 conditions, but not with the community. These genes could be related to interactions that are 292 either alleviated or counteracted in a community, either by the presence of a specific species, or 293 the community as a whole. For example, some of the identified genes were associated with 294 antimicrobial resistance, and in a diverse community, other species could degrade the putative 295 antimicrobial molecule or prevent the producing species from secreting it. Consequently E. coli 296 would be exposed to a lesser level of antimicrobial, suppressing the necessity of a resistance 297 gene.

298
Finally, 8 essential genes appeared to be specifically required in the presence of the 299 community. These genes may be associated with specific interactions emerging from the 300 community context. 3 genes encode for uncharacterized proteins, and the others (purK, purE, 301 damX, ftsX, secB) are not associated with a single function or pathway. We therefore cannot 302 conclude whether specific interactions emerge from the community context or if these genes 303 appeared as specific because of the filtering process criteria.

304
We next investigated if the interactions related to relieved requirements in community 305 growth versus alone and relieved requirements in pairwise conditions versus growth alone were 306 similar. Thus, we compared the alone specific requirements versus growth with the community 307 and the alone specific requirements versus growth in pairwise conditions ( Figure 3B and 308 supplementary file 4). 68 genes were no longer required for both growth in pairwise conditions 309 and with the community compared to growth alone. These genes can also represent pairwise 310 interactions maintained in the community context. Amino acid biosynthesis was highly We also identified 14 genes that were no longer required in pairwise conditions 315 compared to growth alone yet remained essential for growth with the community. Finally, 3 316 genes appeared to be specifically no longer required in the presence of the community. In both 317 cases, too few genes are involved for us to infer any hypothesis on the existence of specific  To summarize, the genes that would be newly essential in the community compared to 321 growth alone were mostly maintained from newly essential genes identified in pairwise 322 conditions compared to growth alone. Similarly, the genes that were no longer essential for the 323 growth in the community were highly similar to the genes that were no longer essential in 324 pairwise condition. However, many newly essential genes for growth in pairwise condition 325 compared to growth alone were not found as essential for growth within the community.

326
Altogether, this highlights that part of the pairwise interactions are conserved within a 327 community while underlining the presence of higher order interactions related to the higher 328 complexity of the growth condition. identified genes required to grow with the community (H. alvei + G. candidum + P. camemberti). Then, we compared the essential genes for growth alone and with 333 the community (A) to identify the newly essential and relieved genes during growth with the community. Next, we compared the newly essential genes in 334 community growth versus growth alone with the newly essential genes in pairwise growth (PW) versus growth alone. We did the same for relieved requirements 335 (B). Within the requirements that were relieved both in pairwise and community growth, genes associated with numerous amino acid biosynthetic pathways were

482
Therefore, consistent with our RB-TnSeq data, this suggests that cross-feeding from a single 483 partner is maintained in a more complex context. Finally, 28 genes were specifically down-484 regulated when E. coli was grown with the community. Individual analysis of these genes 485 highlighted 9 genes that code for uncharacterized proteins but did not highlight over-      screening (absolute t-value >=3) and fitness screening (fitness value < 0) were used to identify 530 essential genes. We first analyzed the 179 genes essential for P. psychrophila growth alone 531 ( Figure 6B and supplementary file 7). Most of the essential genes for P. psychrophila growth 532 alone on CCA appeared to be associated with metabolism. More specifically, they are predicted 533 to be involved in amino acid biosynthesis (all amino acids except for lysine), biosynthesis of 534 cofactors and vitamins as well as energy metabolism (sulfate assimilation). Also, pathways 535 associated with membrane transport were related to iron-siderophore uptake and glycine 536 betaine transport. Essential functions for P. psychrophila growth alone were consistent with the 537 previous functions identified for E. coli growth alone.

538
Next, we identified 152 genes essential for P. psychrophila JB418 growth with H. alvei, 539 176 genes essential for growth with G. candidum, and 164 genes essential for growth with P.

546
We identified 74 newly essential genes for P. psychrophila growth in pairwise conditions 547 compared to growth alone ( Figure 6Ci) and 34 of them were mapped to the KEGG Brite 548 DataBase (Supplementary figure 5). These genes included 11 associated with aromatic amino 549 acid biosynthesis. Since amino acid biosynthesis is also required for growth alone, these data 550 suggest that introduction of a partner leads to an amplified depletion of free amino acids, likely 551 due to competition for these nutrients. Additionally, 4 genes associated with DNA repair were 552 identified as essential in the presence of another species. This could suggest presence of a growth alone with the relieved requirements in pairwise conditions versus growth alone ( Figure   558 6Ciii), we concluded that only 8 genes were no longer required uniquely in the community. 4 of 559 these 8 genes were associated with valine and isoleucine biosynthesis, suggesting that addition 560 of more species might lead to cross-feeding of these amino acids similar to what was observed 561 with E. coli.

562
We identified 46 genes that were newly essential for P. psychrophila growth with the 563 community versus growth alone (Figure6Cii). Comparing these genes to the newly essential 564 genes identified in pairwise growth versus growth alone, we identified 8 genes uniquely 565 essential during community growth and not during any pairwise growth condition (Figure 6Civ).

566
Of these genes, 4 were annotated as either transporters or putative transporters. Here, as for E.

567
coli, we observed that newly essential and relieved requirements in the community context were 568 mostly maintained from pairwise conditions, while not all pairwise newly essential or relieved 569 requirements were maintained in the community.

570
To summarize, essential functions for P. psychrophila growth alone on CCA were 571 consistent with the ones identified for E. coli. While demonstrating important environmental 572 parameters to be considered for growth on CCA such as low available iron, low available free 573 amino acid and high osmolarity, it also underlines the robustness of the RB-TnSeq approach in 574 accurately identifying essential genes. Analysis of the functions of newly essential or relieved 575 genes in interactive contexts allowed us to examine potential P. psychrophila interactions with 576 the members of our model community. Interestingly, the possible interactions probed with P.