The widespread use of pharmaceuticals in society and agriculture as well as their unintentional release from production sites has led to an alarming increase in their presence and accumulation in wastewater treatment plants1–3. Organic pollutants pose a significant environmental concern due to their multiple and still poorly understood impacts on ecosystems and human health 4,5. Consequently, there is a growing need to understand and mitigate the impact of these diverse pollutants on biological processes in wastewater. Traditionally, research efforts have focused on studying the removal efficiency of individual/single pharmaceutical pollutants by microbial communities, which led to a classification of easily biodegradable, such as paracetamol 6–8 , and recalcitrant micropollutants, such as ibuprofen and diclofenac 8–12. While these studies provided valuable insights into the degradation potential of specific compounds, they overlook the complexities of real-world scenarios, where several micropollutants co-occur . Here we assess whether 13,14 the complexity of the pollutome, the set of nefarious molecules present in a given environment, can determine the removal of the pollutants present.

A growing number of studies have highlighted the unpredictable effects of multiple environmental pressures on ecosystems` functioning, including microbes 15,16 . Such unpredictable effects are also likely to happen within a pollutome: When microbial catabolic pathways overlap for specific pollutants, the enrichment of organisms degrading a compound may also promote the degradation of other harboring similar chemical patterns 17–19 . In addition, induction of “promiscuous” enzymes with a large substrate spectrum may lead to 20,21 broader pollutant removal

. Further important mechanisms within a pollutome can probably be cross-feedings22,23, in which certain microorganisms form degradation metabolites to sustain other microorganisms’ growth, and co-metabolism 24,25 , the transformation of a non-growth substrate in the obligate presence of a growth compound 26. 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95

To gain first insights into the impact of the presence of several pharmaceuticals on their removal in wastewater, we exposed wastewater samples to a combinatorial mixture of one to five commonly detected pharmaceuticals. We measured pollutant removal and bacterial community compositions. We hypothesize that due to pollutant-pollutant and pollutantmicrobe interaction within the pollutome, recalcitrant pollutants can be more efficiently degraded.

We set up a total of 96 batch cultures with a volume of 20 mL each (100 mL Erlenmeyer flasks). Each culture contained as a basis synthetic wastewater, following the OECD standard procedures (0.08 g/L peptone, 0.05 g/L meat extract, 15 mg/L urea, 3.5 mg/L NaCl, 2 mg/L CaCl2 x 2

H2O, 0.1 mg/L

Incubation took place at 22°C for 11 days, under continuous shaking of the cultures (130 On days 0, 3, 4, 7 and 11, one 500 µ L sample was taken from each flask, centrifuged at 16,000 x g for 5 minutes and the filtered (0.45 µm pore filter) supernatant was used for HPLC analysis. In addition, on day 3 and day 11, 2 mL of culture was taken and centrifuged at 16,000 x g for 5 minutes, and the resulting pellet was taken for DNA extraction.

Pharmaceuticals were analyzed using high-performance liquid chromatography (HPLC) with a Hi-Plex Na column (Agilent Technologies). The separation was achieved by applying a 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 flow rate of 0.7 mL/min, using a mobile phase consisting of a mixture of water and methanol. Detection of the pharmaceuticals was performed using a UV/VIS DAD detector. The initial mobile phase ratio was set at 80:20 VV, comprising 0.1% formic acid in Millipore water (A) and methanol (B). The B gradient was programmed to transition from 20% to 95% over a span of 15 minutes, enabling simultaneous analysis of all five micropollutants in a single run. The retention times for each pharmaceutical were as follows: ibuprofen eluted at 14.23 minutes, enalapril at 11.08 minutes, caffeine at 7.97 minutes, atenolol at 2.11 minutes, and paracetamol at 2.89 minutes. paracetamol, ibuprofen, atenolol and caffeine were detected at a wavelength of 230 nm, while enalapril was detected at 205 nm. Standard curves were generated for each pollutant, ranging from 1 mg/L to 500 mg/L (1 mg/L, 10 mg/L, 50 mg/L, 100 mg/L, 500 mg/L).

DNA was isolated using the ZymoBIOMICS DNA Miniprep Kit (ZymoResearch, Irvine, USA) according to the manufacturer's instructions. We sequenced the V4 region of the 16S rRNA gene (primer sequences 515f “GTGYCAGCMGCCGCGGTAA” and 806r “GGACTACNVGGGTWTCTAAT”) 27 using the Quick-16S™ Plus NGS Library Prep Kit (V4) (ZymoResearch) to create a DNA library. The library, containing 4 pM DNA (spiked with 25% PhiX), was sequenced in-house on an Illumina MiSeq platform, following the manufacturer's instructions. Raw reads were processed using the R library dada2 (Callahan et al., 2016). This involved quality control steps included analyzing primer sequences, assessing error rates (maxN=0, maxEE=c( 2,2 ), truncQ=2), and identifying chimeras. The resulting sequence table (Min. number of reads = 187722, Max. number of reads = 2049193, Total number of reads = 160233034, Average number of reads = 88040) was aligned to the SILVA ribosomal RNA database (Quast et al., 2012) using version 138 (non-redundant dataset 99). A phyloseq object was then created using the phyloseq R library (McMurdie & Holmes, 2013). 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159

This object consisted of an amplicon sequence variant (ASV) table, a taxonomy table, and sample data. For further analysis, the R libraries phyloseq (McMurdie & Holmes, 2013) and vegan 28

were employed. Functions of microbial communities were determined using PICRUSt2 29. The phyloseq object, metadata, and detailed R code for analysis can be found on GitHub at https://github.com/Marcel2907. The raw sequencing data is available on the NCBI SRA (Sequence Read Archive) under the accession ID PRJNA1041291.

A statistical model was used to analyze the degradation of each compound on each day. In each model the response variable was the percentage remaining of the focal compound on a specific day. In each model there were four binary explanatory variables, each of these coding the presence of the four non-focal compounds. All two-way and three-way interaction terms among explanatory variables were included, as well as the one four-way interaction. In all cases the model was a linear model with Gaussian errors (model diagnostics were acceptable). So, for example, with paracetamol as the focal compound, the model in R would be lm(P ~ I*C*A*E).

For the heatmaps in figure 1 and figure 3 (third column, respectively), the effect sizes for each day were calculated based on the summary of statistical analyses of pollutant interactions. Rows show the estimated coefficients of the single, one-way, two-way, three-way, and fourway interaction terms on pollutant concentration. White cells indicate a response variable and coefficient pairs for which the coefficients were not significantly different from zero (t-test pvalue >.05), otherwise the diverging color palette illustrates the direction of the influence by the driver or interaction of drivers (based on the estimates of the f-test). 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 interactions.

A linear model was also used to analyze how microbial biomass, microbial diversity, and the relative abundance of three most abundant taxa depended on the five compounds and their All statistical analyses can be found in detail in supplementary table 1 and 2. Removal kinetics of single and multiple pharmaceuticals in batch cultures Degradation rates of single pharmaceutical substances in isolation clustered them in two groups: Category 1 (degradable) encompassed caffeine, paracetamol, and atenolol. Caffeine and paracetamol were completely removed within 3-4 days, while about 80% of atenolol was removed

within 11 days (Fig. 1a,b,c; Supplementary Fig. 2). The second category (recalcitrant) contained enalapril and ibuprofen, which were not degraded when individually present (Fig 1d, e; Supplementary Fig. 2).

The degradation rates in mixed pharmaceuticals strongly departed from this baseline. The presence of degradable pharmaceuticals increased the degradation of ibuprofen and enalapril. For example, ibuprofen concentration was reduced to 50% of the original concentration when present alongside all four other pharmaceuticals, and to 70-90% of the initial concentration in some combinations of two or three other compounds (Fig. 1e). Enalapril was reduced to approximately 30% of its initial concentration when paracetamol was also present and or when paracetamol was absent but both atenolol and caffein (“CAE”) were present (Fig. 1d). While pollutants of category 1 were to some extent degradable in all batch cultures, some inhibition effects were observed. For instance, atenolol degradation was hindered in the presence of ibuprofen or paracetamol, and by the presence of various other combinations of other compounds (Fig. 1c). In contrast, no striking additive negative effect was observed when atenolol was present alongside ibuprofen and paracetamol together (Fig. 1c “IPA”). 184 185 186 187 188 189 190 191 192 193 194 195 196

Also, the removal of caffeine was slower in the presence of other pharmaceuticals, especially when paracetamol was present a long lag-phase occurred (Fig. 1b). In contrast, paracetamol degradation was only marginally inhibited by the presence of caffeine (Fig. 1a). These effects of the presence of combinations of other pharmaceuticals on the degradation of a pharmaceutical were not only visually clear but were also clearly revealed by statistical analyses (Fig. 1 third column, Supplementary Table 1). These analyses revealed strong evidence of 2-way, 3-way, and 4-way interaction effects and dependencies among pollutants in the removal processes for all tested pharmaceuticals and were statistically confirmed. The findings suggest that introducing one/more specific pollutant/s significantly affects the degradation of another compound, with the outcome being influenced by the presence of other pollutants. Particularly in the case of ibuprofen removal, identified as the most recalcitrant substrate in this study, notable variations were observed depending on the presence or absence of enalapril. This made up to 30% difference (Fig. 2). dynamics 3 day11 [Ibuprofen]

I:C:A:E C:A:E I:A:E I:C:E I:C:A A:E C:E I:E C:A I:A I:C E A C I I:P:A:E P:A:E I:A:E I:P:E I:P:A A:E P:E I:E P:A I:A I:P E A P I I:P:E:C P:E:C I:E:C I:P:E I:P:C E:C P:E I:E P:C I:C I:P E C P I I:P:A:C I:A:C P:A:C I:P:A I:P:C A:C P:A I:A P:C I:C I:P A C P I P:C:A:E P:C:E C:A:E P:A:E P:C:A A:E C:E P:E C:A P:A P:C E A C P estimates 100 0 −100 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229

Microbial community composition dependencies on pharmaceutical combinations Biomass increased in all batch cultivations, indicating an active microbial community. On day 3, biomass was significantly higher in the presence of single degradable pollutants (atenolol, caffeine, paracetamol) compared to pharmaceutical-free controls or single recalcitrant pollutants (ibuprofen, enalapril) (Figure 3a, main effects in Supplementary Table 2). There was no clear relationship between the number of pollutants and microbial biomass (f-test pvalue 0.57 (number of stressors as a factor)). There is perhaps a pattern of higher biomass when certain combinations of pollutants are present (i.e., PA, PCA, IPCA, ICAE, PECA, and IPCAE). Furthermore, while ibuprofen had uniquely high degradation in the IPCAE treatment, that treatment did not have uniquely high microbial biomass (e.g., the background community ICAE without or with

Efficient pollutant removal from wastewater is essential for environmental safety, yet current water treatment facilities fail to remove organic pollutants such as pharmaceuticals 32 .

Steering microbial communities within these unique ecosystems may be key to designing better removal strategies. Microbial community dynamics can rapidly change in composition and function depending on the incoming water composition33. Therefore, these wastewater treatment plants can be seen as a model system for studying multiple drivers on microbial communities and their degradation capacity of pollutants. Pollutant removal has been extensively studied in isolation, providing detailed insights into the molecular mechanisms underlying biodegradation. However, these findings only marginally translate to real-world scenario, where multiple drivers co-occur 34,35

.

In this work we shed light on the interactive effects of pollutants within the pollutome, using mixtures of pharmaceutical varying in biodegradability as a model. We demonstrate that the complexity of the pollutome is a major driver of biodegradation and that the presence of multichemical background pollution was essential for the removal of recalcitrant molecules. We found in particular the degradation of recalcitrant pollutants to be strongly modulated by the presence of other pollutants. Easily degradable pharmaceuticals such as paracetamol, atenolol, and caffeine enable the degradation of the more recalcitrant ibuprofen and enalapril. In addition, some pollutants may hinder the biodegradation of other. This was particularly striking for atenolol, which degradation was notably inhibited in the presence of ibuprofen. These two chemicals show strong structural similarities (e.g. benzene ring and alkyl chain), which might inhibit enzymatic activity. Ibuprofen deserves special attention, as it proved to be only degradable when incubated alongside other pharmaceutical compounds. This observation underscores the significance of studying the environmental fate of pharmaceuticals as a collective group rather than at single compound level. 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341

Interactions between pollutants are likely due to shifts in microbial community composition and function. We identified a range of potential key players (based on high relative abundances) for pharmaceuticals, namely Achromobacter, Pseudomonas, Acinetobacter, Comamonas and Trichococcus. All of them have already been shown to be associated with 6,7,36–39 pollutant degradation

. These genera strongly respond to the composition of the pollutant mix, potentially explaining previous observations of their fluctuations in wastewater treatment systems 40,41

. In particular, paracetamol had a strong effect on the microbial community composition. When degraded, paracetamol is broken down into aminophenol, which is toxic for many microorganisms 42. Paracetamol-treated communities indeed showed an increased abundance of the aminophenol pathway. This pathway may have served as detoxification mechanism 7 but may also be involved in the degradation of other recalcitrant pharmaceuticals. It's also possible that the pathways involved in paracetamol degradation could lead to the breakdown of another substance, utilizing enzymes such as monooxygenase and dioxygenase, for instance. One caveat to mention here is the fact that due to the comparatively high concentration used in this study, it is possible that the microbial community was not able to degrade fast enough the formed aminophenol, which might have accumulated transiently.

The major question which needs to be tackled in future studies is the reason for this observation. In general, underlying reason could be rooted in mechanisms like cross-feedings, elevated enzyme activity, increased energy levels, and the induced expression of genes encoding promiscuous enzymes20–23,26. However, our study can probably rule out increased biomass as a contributing factor since the ibuprofen-degrading cultures did not yield more biomass than other treatments. Co-metabolic effect related to enzyme that fortuitously accept various chemically related substrate could play a role, since the chemical structures of the used pharmaceuticals show some chemical similarities, such as aromatic rings 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 (Supplementary Figure 1). Therefore, it could be possible that e.g. specific dioxygenases that play a role in paracetamol degradation 7,43 could potentially also show (low) activity against ibuprofen and enalapril.

All tested pharmaceuticals are globally detectable in wastewater influents and effluents and occur worldwide in ng/L - µ g/L scale 6,44–47

, which is significantly lower compared to the high concentration we used in this study. However, the CODs used in this study can occur in 48 wastewater of industrial production sites of pharmaceuticals .

This study indicates that the presence of easily degradable micropollutants, such as caffeine, atenolol, and paracetamol, promoted the degradation of recalcitrant substrates like ibuprofen and enalapril. In contrast, these latter compounds were not degraded when present as the sole pollutant. The significance of these discoveries is noteworthy, as they can serve as potential starting points for the development of future applications aimed at the effective removal of pharmaceuticals: The study demonstrated that the addition of specific compounds at specific time points can enhance degradation of a target pollutant. Addition of non-toxic functional mimics of existing pollutants may thus improve the microbial removal of persistent pollutants, contributing to safe water, ecosystems, and food supply. We conclude that pollutants should be treated as part of a complex system, with emerging pollutants potentially showing cascading effects and offering leverage to promote bioremediation. 392 393 394 395

Declarations

This work was funded through the European Union’s H orizon 2020 project NMYPHE under grant ID 10106625.

Marcel Suleiman and Philippe Corvini planned the experimental setup. Natalie La Ley performed all experiments in the lab. Francesca Demaria and Boris Kolvenbach supported analytical lab work and Boris Kolvenbach supported the analysis of metabolic pathways. Marcel Suleiman performed the up- and downstream bioinformatics. Marcel Suleiman, Owen L. Petchey and Alexandre Jousset performed the data analysis. Owen Petchey and Alexandre Jousset supported the use of statistical models. Silvia Cretoiu performed the Picrust2 analysis and gave bioinformatic support. Owen L. Petchey, Marcel Suleiman, Philippe Corvini and Alexandre Jousset drafted the manuscript. All authors confirmed the final version of the manuscript.

Not applicable 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522