Evolved bacterial resistance to the chemotherapy gemcitabine modulates its efficacy in co-cultured cancer cells

  1. Serkan Sayin
  2. Brittany Rosener
  3. Carmen G Li
  4. Bao Ho
  5. Olga Ponomarova
  6. Doyle V Ward
  7. Albertha JM Walhout
  8. Amir Mitchell

  • Curated by eLife

    eLife logo

    eLife assessment

    This fundamental work advances our understanding of how bacteria evolve to resist drugs used for cancer treatment and how this could potentially affect drug efficacy and treatment outcome. The data were collected and analyzed using a solid methodology and can be used as a starting point for functional studies of the interaction between microbiome interactions and cancer drug treatment. The findings will be of broad interest to microbiologists and organismal biologists interested in the role of microbiomes in drug responses.

Reviewed by

Read the full article See related articles

Listed in

Log in to save this article

Abstract

Drug metabolism by the microbiome can influence anticancer treatment success. We previously suggested that chemotherapies with antimicrobial activity can select for adaptations in bacterial drug metabolism that can inadvertently influence the host’s chemoresistance. We demonstrated that evolved resistance against fluoropyrimidine chemotherapy lowered its efficacy in worms feeding on drug-evolved bacteria (Rosener et al., 2020). Here, we examine a model system that captures local interactions that can occur in the tumor microenvironment. Gammaproteobacteria-colonizing pancreatic tumors can degrade the nucleoside-analog chemotherapy gemcitabine and, in doing so, can increase the tumor’s chemoresistance. Using a genetic screen in Escherichia coli, we mapped all loss-of-function mutations conferring gemcitabine resistance. Surprisingly, we infer that one third of top resistance mutations increase or decrease bacterial drug breakdown and therefore can either lower or raise the gemcitabine load in the local environment. Experiments in three E. coli strains revealed that evolved adaptation converged to inactivation of the nucleoside permease NupC, an adaptation that increased the drug burden on co-cultured cancer cells. The two studies provide complementary insights on the potential impact of microbiome adaptation to chemotherapy by showing that bacteria–drug interactions can have local and systemic influence on drug activity.

Article activity feed

  1. Version published to 10.7554/elife.83140 on eLife
  2. eLife

    Author Response

    Reviewer #1 (Public Review):

    Sayin et al. sought to determine if bacterial drug resistance has impact on drug efficacy. They focused on gemcitabine, a drug used for pancreatic cancer that is metabolized by E. coli. Using an innovative combination of genetic screens, experimental evolution, and cancer cell co-cultures to reveal that E. coli can evolve resistance to gemcitabine through loss-of-function mutations in nupC, with potential downstream consequences for drug efficacy.

    Major strengths include:

    • Paired use of genetic screens and experimental evolution

    • The spheroid model is a creative approach to modeling the tumor microbiome that I hadn't seen before

    • Rigorous microbiology, including accounting for mutation rate in both selective and non-selective conditions

    • Timely research question

    Major weaknesses of the methods and results include the following:

    1. Limited scope of the current work. Just a single drug-bacterial pair is evaluated and there are no experiments with microbial communities, animal models, or attempts to test the translational relevance of these findings using human microbiome datasets.

    We agree with the reviewer that uncovering evidence from human microbiome datasets will be very exciting and complementary to our study. However, since gemcitabine is administered intravenously it’s unclear whether it will impose a considerable selective pressure on the gut microbiome. Therefore, it also remains unclear if adaptive mutations, as those we identified, are expected to be found in datasets for the gut microbiome. While metagenomics datasets that are bacterial-centric of infected pancreatic tumors will be ideal for addressing the reviewer’s suggestion, they do not exist to the best of our knowledge. It should be noted however, that our work generated hypotheses that can be tested in pancreatic tumor tissues infected with gammaproteobacteria and can be tested in the future by targeted sequencing for the specific genes of interest (e.g, nupC and cytR).

    1. No direct validation of the primary genetic screen. The authors use a very strict cutoff (16-fold-change) without any rationale for why this was necessary. More importantly, a secondary screen is necessary to evaluate the reproducibility of the results, either by testing each KO in isolation or by testing a subset of the library again.

    We used a strict cutoff to allow the reader to focus on a manageable list of gene names in the main figure (2E). To partly address this limitation in scope, we also included results from pathway enrichment analysis in the same figure (2F). This analysis utilizes all enrichment values and is therefore independent from any choice of cutoff value. We also now refer the reader to explore more of the hit genes in the supplementary information (line 152).

    As the reviewer suggested we evaluated the reproducibility of the results by performing two validation screens. The first validation screen was performed as a biological replicate of the original screen and relied on the original collection of knockouts strains. The second validation screen was performed with a knockout strain collection that was cloned independently from the strains used in our original screen. The results from these two completely independent biological replicates are presented on supp. figure 1D. The results (resistance/sensitivity) from the two screens are highly correlated. We refer to this comparison in the main text (lines 142-147).

    1. Some methodological concerns about the spheroid system. As I understood it, these cells are growing aerobically, which may not be the best model for the microbiome. Furthermore, bacterial auxotrophs are used and only added for 4 hours, which will really limit their impact. It also was unclear if the spheroids are truly sterile. Finally, the data lacks statistical analysis, making it unclear which KOs are meaningful. Delta-cdd looks clearly distinct by eye, but the other two genes are more subtle.

    The 4 hour time interval chosen to address two opposing requirements of the co-culture system – mitigate overgrowth of the bacterial cultures (which hinders spheroid growth irrespective of the drug) while still allowing enough incubation time to allow for drug degradation. As the reviewer notes, removal after 4 hours may limit the bacteria impact. However, such a limitation will only result in underestimation the bacterial impact (but will have no impact on how we evaluate how strains compare to one-another). We now comment on this in the methods section (lines 699-705).

    We do not expect the spheroid to remain infected after bacterial removal since we treat spheroids with antibiotics. We didn’t not detect any bacterial growth in the 7 days post infection in the microscope and we did not observe influence on spheroid growth when compared to spheroid that were not infected. Growth of spheroid before infection was performed w/o antibiotics and we did not detect any evidence of bacterial growth prior to introducing the bacteria intentionally (the cell-line itself was also tested for animal pathogens and bacterial contamination prior to the experiments).

    We repeated the spheroid experiments and observed similar shifts in the EC50 fronts. We now include these replicates as supplementary figure 7. We comment on these replicates in the main text (lines 273-274).

  3. eLife

    eLife assessment

    This fundamental work advances our understanding of how bacteria evolve to resist drugs used for cancer treatment and how this could potentially affect drug efficacy and treatment outcome. The data were collected and analyzed using a solid methodology and can be used as a starting point for functional studies of the interaction between microbiome interactions and cancer drug treatment. The findings will be of broad interest to microbiologists and organismal biologists interested in the role of microbiomes in drug responses.

  4. eLife

    Reviewer #1 (Public Review):

    Sayin et al. sought to determine if bacterial drug resistance has impact on drug efficacy. They focused on gemcitabine, a drug used for pancreatic cancer that is metabolized by E. coli. Using an innovative combination of genetic screens, experimental evolution, and cancer cell co-cultures to reveal that E. coli can evolve resistance to gemcitabine through loss-of-function mutations in nupC, with potential downstream consequences for drug efficacy.

    Major strengths include:
    • Paired use of genetic screens and experimental evolution
    • The spheroid model is a creative approach to modeling the tumor microbiome that I hadn't seen before
    • Rigorous microbiology, including accounting for mutation rate in both selective and non-selective conditions
    • Timely research question

    Major weaknesses of the methods and results include the following:

    1. Limited scope of the current work. Just a single drug-bacterial pair is evaluated and there are no experiments with microbial communities, animal models, or attempts to test the translational relevance of these findings using human microbiome datasets.

    2. No direct validation of the primary genetic screen. The authors use a very strict cutoff (16-fold-change) without any rationale for why this was necessary. More importantly, a secondary screen is necessary to evaluate the reproducibility of the results, either by testing each KO in isolation or by testing a subset of the library again.

    3. Some methodological concerns about the spheroid system. As I understood it, these cells are growing aerobically, which may not be the best model for the microbiome. Furthermore, bacterial auxotrophs are used and only added for 4 hours, which will really limit their impact. It also was unclear if the spheroids are truly sterile. Finally, the data lacks statistical analysis, making it unclear which KOs are meaningful. Delta-cdd looks clearly distinct by eye, but the other two genes are more subtle.

    Despite these concerns, this paper is a valuable addition to the growing literature on interactions between cancer chemotherapy and the microbiome, which will definitely inspire follow-up work in complex microbial communities, animal models, and human cohorts.

  5. eLife

    Reviewer #2 (Public Review):

    Many cancers, including pancreatic tumors, host microbes that have the ability to metabolize anti-cancer drugs, thus altering cancer response to these treatments. However, many anti-cancer drugs also are quite toxic to bacteria. Thus, the authors first investigate how a model bacterium that could live pancreatic tumors can become resistant to the pancreatic chemotherapy gemcitabine. Second, they investigate how bacteria that are resistant to gemcitabine impact cancer cell response to this therapy compared to bacteria that are not resistant. By answering these two questions, the authors hope to determine how bacterial evolution to chemotherapy can impact how well chemotherapy works in pancreatic cancer.

    To answer the first question, the authors perform both genetic screens and laboratory evolution experiments of E. coli bacteria exposed to gemcitabine. Both the genetic screen and laboratory evolution experiments identified mutation of the bacterial protein nupC as mediating bacterial resistance to gemcitabine. NupC is the transporter protein that bacteria use to take up gemcitabine. Thus, the authors conclude that loss of ability to take up gemcitabine would likely underlay bacterial evolution to gemcitabine in pancreatic tumors.

    To answer the second question, the authors take either control of nupC mutant bacteria and expose these to gemcitabine. They then take the bacterial media with its residual gemcitabine and treat mouse colorectal cancer cells with these media. They find the amount of gemcitabine is higher in nupC mutant media and media from these mutants cause correspondingly higher killing of cancer cells.

    Thus, the authors conclude that bacteria become resistant to gemcitabine by not taking it up, leaving more gemcitabine around in tumors to kill the cancer cells. The findings of the first question are a major strength of the manuscript - the complementary genetic screen and laboratory evolution experiment convincingly show that loss of nupC is likely a major genetic route for bacteria to become resistant to gemcitabine. Excellent biochemical studies delineate mechanistically how the different mutations including nupC contribute to gemcitabine resistance in the bacteria.

    However, a major weakness of the manuscript is the extension to how this laboratory evolved nupC resistance to gemcitabine influences tumor response to gemcitabine. The only experiments done to assess this are performed in colorectal cell culture models in vitro. Importantly, these in vitro models do not recapitulate chemotherapy resistance observed in pancreas cancer and utilize levels of bacteria and gemcitabine that are likely not relevant to tumor physiology. Thus, additional experiments assessing in vivo if nupC mutations become prevalent in the pancreatic tumor microbiome and how much mutations affect tumor gemcitabine levels and response will be necessary to fully answer the authors second question of how bacterial evolution to gemcitabine affects tumor response to this agent.

  6. Version published to 10.1101/2022.09.07.506952v1 on bioRxiv