Microbiota-derived 3-IAA influences chemotherapy efficacy in pancreatic cancer

Pancreatic ductal adenocarcinoma (PDAC) is expected to be the second most deadly cancer by 2040, owing to the high incidence of metastatic disease and limited responses to treatment1,2. Less than half of all patients respond to the primary treatment for PDAC, chemotherapy3,4, and genetic alterations alone cannot explain this5. Diet is an environmental factor that can influence the response to therapies, but its role in PDAC is unclear. Here, using shotgun metagenomic sequencing and metabolomic screening, we show that the microbiota-derived tryptophan metabolite indole-3-acetic acid (3-IAA) is enriched in patients who respond to treatment. Faecal microbiota transplantation, short-term dietary manipulation of tryptophan and oral 3-IAA administration increase the efficacy of chemotherapy in humanized gnotobiotic mouse models of PDAC. Using a combination of loss- and gain-of-function experiments, we show that the efficacy of 3-IAA and chemotherapy is licensed by neutrophil-derived myeloperoxidase. Myeloperoxidase oxidizes 3-IAA, which in combination with chemotherapy induces a downregulation of the reactive oxygen species (ROS)-degrading enzymes glutathione peroxidase 3 and glutathione peroxidase 7. All of this results in the accumulation of ROS and the downregulation of autophagy in cancer cells, which compromises their metabolic fitness and, ultimately, their proliferation. In humans, we observed a significant correlation between the levels of 3-IAA and the efficacy of therapy in two independent PDAC cohorts. In summary, we identify a microbiota-derived metabolite that has clinical implications in the treatment of PDAC, and provide a motivation for considering nutritional interventions during the treatment of patients with cancer.

Pancreatic ductal adenocarcinoma (PDAC) is expected to be the second most deadly cancer by 2040, owing to the high incidence of metastatic disease and limited responses to treatment 1,2 . Less than half of all patients respond to the primary treatment for PDAC, chemotherapy 3,4 , and genetic alterations alone cannot explain this 5 . Diet is an environmental factor that can influence the response to therapies, but its role in PDAC is unclear. Here, using shotgun metagenomic sequencing and metabolomic screening, we show that the microbiota-derived tryptophan metabolite indole-3-acetic acid (3-IAA) is enriched in patients who respond to treatment. Faecal microbiota transplantation, short-term dietary manipulation of tryptophan and oral 3-IAA administration increase the efficacy of chemotherapy in humanized gnotobiotic mouse models of PDAC. Using a combination of loss-and gain-of-function experiments, we show that the efficacy of 3-IAA and chemotherapy is licensed by neutrophil-derived myeloperoxidase. Myeloperoxidase oxidizes 3-IAA, which in combination with chemotherapy induces a downregulation of the reactive oxygen species (ROS)-degrading enzymes glutathione peroxidase 3 and glutathione peroxidase 7. All of this results in the accumulation of ROS and the downregulation of autophagy in cancer cells, which compromises their metabolic fitness and, ultimately, their proliferation. In humans, we observed a significant correlation between the levels of 3-IAA and the efficacy of therapy in two independent PDAC cohorts. In summary, we identify a microbiota-derived metabolite that has clinical implications in the treatment of PDAC, and provide a motivation for considering nutritional interventions during the treatment of patients with cancer.
Polychemotherapy, either with 5-fluorouracil (5-FU), irinotecan and oxaliplatin in combination with folinic acid (FOLFIRINOX), or with gemcitabine and nab-paclitaxel (GnP), is considered the standard of care for patients suffering from metastatic PDAC (mPDAC) 3,4 . However, less than half of all patients are responsive to the therapy, and patients who do not respond (NR patients) suffer from pain and eventually die within weeks 3 . Genetic alterations in PDAC poorly explain the differences between patients who respond to therapy (responder (R) patients) and NR patients 5,6 , which leaves environmental factors-including the intestinal microbiota-as the potential mediators of chemotherapy efficacy. There is, therefore, an urgent need to identify environmental factors that might explain the differences between R and NR patients so that new concepts can be developed for future therapies.
The intestinal microbiota has been shown to induce a response to immunotherapy in patients with melanoma, and can be modulated by dietary habits [7][8][9][10] . In rare patients with localized PDAC who are long-term survivors, bacteria can translocate from the intestine into the tumour and control anti-tumour immune activation 11,12 . However, most patients suffering from aggressive immunotherapy-resistant mPDAC are treated with polychemotherapy, and it is at present unclear whether and how the microbiota or dietary habits affect its efficacy 1,13 . for detailed criteria). R patients (n = 11) had a mean PFS of 40.9 weeks, which was significantly higher than the PFS of 12.8 weeks for NR patients (n = 12), and the overall survival was 51.9 weeks for R and 26.4 weeks for NR patients (Extended Data Fig. 1b,c). The microbiota of R patients was distinct from the microbiota of NR patients ( Fig. 1b and Extended Data Fig. 1d-f). To study a potential cause-effect relationship between the microbiota and the response to chemotherapy, we transferred microbiota from the first ten recruited R and NR patients into gnotobiotic mice, followed by orthotopic injection of Pdx1-Cre, LSL-KRAS G12D , LSL-Trp53 R172H/+ (KPC) pancreatic cancer cells (Extended Data Fig. 2a-d).
Notably, four patients were treated with FOLFIRINOX and one patient was treated with GnP in both groups (R and NR). Irrespective of the original donor treatment, we observed smaller tumours in mice that were colonized with the microbiota of R patients, but not in mice colonized with the microbiota of NR patients, after chemotherapy treatment (5-FU, irinotecan and oxaliplatin; FIRINOX) ( Fig. 1c and Extended Data Fig. 2e). Given the potential translocation of intestinal bacteria into PDAC tumours 11 , we next analysed the intratumoural bacteria by 16S rRNA sequencing. We could only detect intratumoural bacteria in 2 out of 12 tumours (Extended Data Fig. 2f), and thus hypothesized that the response to chemotherapy is indirectly controlled through circulating microbiota-derived metabolites.
To address this, we analysed serum from R and NR patients and matching colonized gnotobiotic mice and performed a targeted metabolomic screen using liquid chromatography coupled to mass spectrometry. We found that the tryptophan metabolite 3-IAA was the most significantly enriched metabolite in R compared to NR patients (Fig. 1d). In line with this, 3-IAA was also enriched in the serum of gnotobiotic mice that were colonized with R compared to NR microbiota ( Fig. 1e and Extended Data Fig. 2g).
To characterize which gut bacteria in R patients could contribute to increased production of 3-IAA, we analysed the abundance of common 3-IAA-producing bacterial strains 14 in the microbiota of R compared to NR patients. Out of the fifteen analysed 3-IAA producers, we found that   Fig. 1 | 3-IAA induces a response to FIRINOX in mouse models of PDAC. a, The intestinal microbiota of 23 patients with mPDAC was sequenced before the start of chemotherapy (chemo). GF, germ-free. b, Principal coordinate analysis (PCoA), with Bray-Curtis dissimilarity matrix, of pre-treatment NR (n = 12) and R (n = 10; one patient was excluded after the quality control) microbiota. c, Gnotobiotic mice were colonized with five different R or NR microbiota, KPC cancer cells were orthotopically injected and mice were treated with FIRINOX or left untreated. Tumour weight is shown relative to the mean tumour weight of the untreated group of each experiment at day 20 after tumour cell injection (n = 9 NR1-3; n = 8 R2; n = 7 R1 and R3; n = 5 NR5, R4 and R5; n = 4 NR4; pooled from nine independent experiments). d,e, Volcano plots showing differentially abundant metabolites in the serum of three R and NR patients (d) and in the serum of gnotobiotic mice colonized with R or NR microbiota (n = 3 each, biological replicates) (e). NS, not significant. f, R-microbiota-colonized gnotobiotic mice were fed with the indicated concentration of tryptophan, and the 3-IAA serum concentration at the fourth day of dietary intervention is shown (n = 5 each). g, Left, tumour weight of orthotopic KPC tumours (n = 8, 5 or 9) after FIRINOX treatment. Right, correlation between 3-IAA serum concentration and tumour weight of five randomly selected mice per dietary group from f (n = 15). h, SPF mice were orthotopically injected with KPC cells and treated with or without (+/−) 3-IAA and/or with or without FIRINOX, and tumour weight was assessed at day 20 of the experiment (n = 5 or 6). Each symbol represents one mouse. One out of two (f,g) or three (h) independent experiments is shown. Error bars indicate s.e.m. Significant P values are indicated and were determined by MANOVA (b), twotailed nested t-test (c), fold change analysis and two-tailed t-test (d,e), simple linear regression and Pearson's r (g) and one-way ANOVA with Tukey's post-hoc test (f-h).

Article
Bacteroides fragilis and Bacteroides thetaiotaomicron were increased in R patients and confirmed their capacity to produce 3-IAA in vitro (Extended Data Fig. 2h,i).
We next wondered whether we could alter the serum levels of 3-IAA in mice and thereby affect the efficacy of chemotherapy by modulating the dietary concentration of tryptophan, the precursor of 3-IAA (ref. 15 ). Because tryptophan can impair the development of anti-tumour immune responses and thus promote tumour growth 16,17 , we first tested different lengths of dietary intervention. Fourteen days of tryptophan exposure promoted tumour growth, whereas a treatment period of only four days was not sufficient (Extended Data Fig. 2j,k). We therefore decided to choose a four-to-five-day intervention for the rest of our experiments to avoid the pro-tumorigenic effect 16,17 .
Second, we observed that four days of a high-tryptophan diet was already sufficient to increase the concentration of 3-IAA in the serum of gnotobiotic mice that were colonized with R microbiota (Fig. 1f). Moreover, when this dietary intervention was combined with FIRINOX, we observed a decreased tumour weight (Fig. 1g). In addition, 3-IAA serum levels correlated inversely with tumour weight (Fig. 1g). Notably, these tryptophan-mediated effects-that is, increased concentration of 3-IAA and decreased tumour weight-were lost in NR-microbiota-colonized mice (Extended Data Fig. 2l,m). These data suggest that the effect of the tryptophan-high diet is mediated by 3-IAA, but we cannot rule out additional mechanisms contributing to the observed increased response in this set of experiments. Therefore, we directly supplemented 3-IAA in specific-pathogen-free (SPF) mice and found that our intervention was sufficient to reach the concentrations measured in mice that were colonized with R microbiota, and to increase the efficacy of chemotherapy not only in SPF but also in NR-colonized gnotobiotic mice ( Fig. 1h and Extended Data Fig. 3a-c). We also tested other microbiota-modulated metabolites that have implications in different gastrointestinal cancers, such as the secondary bile acid deoxycholic acid (DCA) and the primary bile acid glycocholic acid (GCA) 18,19 . In addition, we chose to test two other metabolites that were increased in R patients or R-microbiota-colonized gnotobiotic mice-namely, the indole derivate indole-3-propionic acid (IPA) and hippuric acid-in combination with FIRINOX. None of these metabolites led to a similar efficacy to that of 3-IAA (Extended Data Fig. 3d,e). Together, these data indicate that the microbiota-derived tryptophan metabolite 3-IAA is increased in the serum of humans and mice that respond to chemotherapy, and that the levels of 3-IAA can be modulated by dietary interventions with tryptophan. Moreover, 3-IAA rendered even chemotherapy-resistant PDAC susceptible to the treatment.

The effect of 3-IAA is licensed by myeloperoxidase
Microbiota-derived metabolites, especially tryptophan metabolites, have a crucial role in shaping innate and adaptive immunity 17 , which in turn have an important role in determining the efficacy of chemotherapy and prognosis in PDAC 11,20 . Therefore, we decided to analyse tumour-infiltrating immune cells in chemotherapy-naive or treated R-or NR-microbiota-colonized gnotobiotic mice. We consistently observed increased frequencies of CD8 + T cells and decreased neutrophils in R-compared to NR-microbiota-colonized mice after chemotherapy. We did not observe any changes in tumour-infiltrating immune cells when comparing untreated mice (Extended Data Fig. 4a,b). CD8 + T cells can induce the regression of PDAC when activated by the intestinal microbiota 11,21 . However, depleting CD8 + or both CD4 + and CD8 + T cells through antibody treatments did not decrease the efficacy of chemotherapy in R-microbiota-colonized mice that received either a standard or a high-tryptophan diet (Extended Data Fig. 5a-d). Similarly, depletion of CD4 + and CD8 + T cells did not lower the efficacy of 3-IAA and FIRINOX in SPF mice (Extended Data Fig. 5e). Thus, the effect of 3-IAA does not seem to depend on the presence of T cells.
Neutrophils are highly abundant in PDAC, and a low neutrophil count (neutropenia) is associated with a good prognosis in mPDAC 22 . 3-IAA is specifically toxic to cells with a high concentration of myeloperoxidase (MPO), which is a hallmark of neutrophils 23 . Mechanistically, MPO can  oxidize 3-IAA, inducing toxic products such as 3-methylene-2-oxindole  (MOI) 24,25 . In line with this, we found that culturing bone-marrow-derived neutrophils, but not T cells (neutrophils have higher levels of MPO than T cells), with 3-IAA and FIRINOX led to reduced cell survival (Extended Data Fig. 6a). Moreover, adding MPO to bone-marrow-derived neutrophil cultures or using immature neutrophils (pre-neutrophils) that have higher levels of endogenous MPO (ref. 26 and Extended Data Fig. 6b) increased the toxicity of 3-IAA and FIRINOX (Extended Data Fig. 6c,d). By contrast, IPA did not increase the efficacy of FIRINOX in inducing neutrophil cell death in vitro (Extended Data Fig. 6e). Further characterization of the response of neutrophils to 3-IAA and FIRINOX revealed that the treatment induced neutrophil degranulation, which was measured as the release of MPO, and necrosis, but did not lead to the formation of neutrophil extracellular traps (NETs) or apoptosis (Extended Data Fig. 6f-h). Next, we wondered whether these results were also reproducible in vivo. The combination of 3-IAA and FIRINOX reduced the frequency and number of neutrophils in the tumour and spleen of SPF mice ( Fig. 2a and Extended Data Fig. 6i). Notably, this was not the case for 3-IAA treatment alone (Extended Data Fig. 6j).
Given the crucial role of neutrophil-derived MPO in the release of toxic 3-IAA products, we wondered whether this could explain the increased efficacy of 3-IAA and FIRINOX on tumour growth. To address this question, we established bone marrow chimeras from either wild-type or Mpo -/mice. As expected, treatment with FIRINOX and 3-IAA resulted in the depletion of neutrophils and showed synergistic efficacy in wild-type reconstituted mice (Fig. 2b). However, in mice that were reconstituted with Mpo -/bone marrow, the addition of 3-IAA did not result in smaller tumours (Fig. 2b), suggesting that MPO is essential for the efficacy of 3-IAA and FIRINOX. Further supporting the role of MPO, the 3-IAA oxidation product MOI was increased in tumours from wild-type SPF mice after treatment with 3-IAA and FIRINOX compared to that with FIRINOX alone (Extended Data Fig. 7a). In addition, the levels of MOI were strongly reduced in tumours from Mpo -/bone-marrow-reconstituted mice (Extended Data Fig. 7b).
Finally, considering the key role of the aryl hydrocarbon receptor (AhR) as an intracellular receptor for 3-IAA and other indols 27 , we wondered whether Ahr -/bone-marrow-reconstituted mice are resistant to 3-IAA and FIRINOX. The size of the tumours in Ahr -/bone-marrow-reconstituted mice was reduced to the same extent as that of tumours from wild-type reconstituted mice after treatment with 3-IAA and FIRINOX (Fig. 2c), arguing against a role of the AhR in mediating the efficacy of 3-IAA and FIRINOX treatment. Moreover, the AhR was also dispensable in cancer cells as Ahr-knockdown KPC cells responded to 3-IAA and FIRINOX treatment in vivo (Extended Data Fig. 7c).
These data show that the efficacy of 3-IAA and FIRINOX is licensed by immune cell-derived MPO, but not by AhR signalling.

The effect of 3-IAA depends on ROS and autophagy
Oxidation of 3-IAA via MPO induces ROS in cultured neutrophils 23 and ROS are major mediators of chemotherapy-induced cell death 28 . Therefore, we hypothesized that the efficacy of 3-IAA and FIRINOX in the presence of MPO is mediated by ROS. To address this, we cultured mouse and human PDAC cells with increasing dosages of 3-IAA, 3-IAA and neutrophils or 3-IAA and MPO with or without chemotherapy. 3-IAA directly increased ROS in a dose-dependent manner. The addition of neutrophils or MPO further enhanced the accumulation of ROS in cancer cells, and the addition of MPO reduced the viability of mouse and human PDAC cell lines (Extended Data Fig. 7d-k). Notably, direct addition of the 3-IAA oxidation product MOI also increased the efficacy of FIRINOX in mouse and human PDAC cells (Extended Data Fig. 7l,m). Furthermore, to test the effects of 3-IAA and FIRINOX treatment on ROS induction in vivo, we treated SPF mice with either 3-IAA and FIRINOX or FIRINOX alone. In line with our in vitro data, treatment with 3-IAA and FIRINOX induced high oxidative stress, measured as ROS in cancer cells by flow cytometry or as nitrotyrosine in whole-tumour histology stains (Extended Data Fig. 7n,o). Further strengthening the link between ROS production and the oxidation of 3-IAA, we observed much lower levels of ROS after treatment with 3-IAA and FIRINOX in cancer cells from Mpo -/compared to wild-type bone-marrow-reconstituted mice, by flow cytometry (Extended Data Fig. 7p).
We then wondered which oxidative-stress-related pathways are involved in the accumulation of ROS in 3-IAA-and FIRINOX-treated tumours. To this end, we analysed the expression of known ROS-producing or ROS-degrading enzymes in mRNA sequencing data, comparing tumours from mice that were treated with 3-IAA and FIRINOX to tumours from mice treated with FIRINOX alone (Extended Data Fig. 8a). We found that the ROS-degrading enzymes glutathione peroxidase 3 (GPX3) and glutathione peroxidase 7 (GPX7) were downregulated in vivo, and treatment with 3-IAA, FIRINOX and MPO led to a similar downregulation of GPX3 and GPX7 in KPC cells in vitro (Extended Data Fig. 8b). Knockdown of Gpx3 and Gpx7 in cancer cells was sufficient to increase the accumulation of ROS after treatment with FIRINOX, and increased the susceptibility of the cancer cells to FIRINOX to a similar extent as 3-IAA and FIRINOX (Extended Data Fig. 8c,d). Of note, knockdown of Gpx3 was sufficient to establish susceptibility to FIRINOX in vivo (Extended Data Fig. 8e). Finally, we could show that ROS accumulation is essential for the efficacy of 3-IAA and FIRINOX, because treatment with the ROS scavenger N-acetylcysteine (NAC) abolished the efficacy of FIRINOX in R-microbiota-colonized (that is, high-3-IAA) mice (Fig. 3a).
Next, we wondered what the molecular response to high levels of ROS is. To address this, we performed a targeted protein screen and mRNA sequencing on tumours isolated from SPF mice that were treated with 3-IAA and FIRINOX or FIRINOX alone (Fig. 3b). Autophagy was one of the main pathways that was downregulated in tumours isolated from mice treated with 3-IAA and FIRINOX (Extended Data Fig. 8f,g and Supplementary Tables 1-3). Notably, autophagy is an essential metabolic program for PDAC to proliferate and thrive 29 . Next, we found that the autophagy substrate LC3B was reduced and p62/SQSTM was increased in the tumours isolated from SPF mice that were treated with 3-IAA and FIRINOX ( Fig. 3c and Extended Data Fig. 8h). In line with this, tumours from R-microbiota-colonized mice (high 3-IAA) also showed a reduced abundance of LC3B and an increased abundance of p62/ SQSTM1 compared to tumours from NR-colonized mice (low 3-IAA) (Extended Data Fig. 8i,j). Changes in LC3B and p62/SQSTM were accompanied by a decline in tumour cell proliferation, measured as the expression of Ki67, in mice treated with 3-IAA and FIRINOX (Fig. 3c). However, we did not observe a change in the number of apoptotic cells, as measured by cleaved caspase-3 (Extended Data Fig. 8k), as late as three days after treatment. To test whether downregulation of autophagy is an essential downstream mechanism that explains the synergy of 3-IAA and FIRINOX, or whether it is merely associated with the observed reduction in cancer cell proliferation, we set up gain-and loss-of-function experiments in vivo. First, we observed that treatment with the disaccharide trehalose 30 normalizes autophagic activity, as measured by a decreasing GFP/RFP ratio in autophagy reporter cells 31 ( Fig. 3d and Extended Data Fig. 8l). Second, trehalose was sufficient to completely reverse the treatment efficacy of 3-IAA and FIRINOX, similar to the ROS scavenger NAC (Fig. 3e,f). Third, treatment with the autophagy blocker hydroxychloroquine sensitized tumours from NR-microbiota-colonized mice (low 3-IAA) to FIRINOX treatment (Fig. 3g). Finally, inhibiting autophagy in cancer cells through a doxycycline-inducible dominant-negative ATG4B protein (mSt-ATG4B) 29 increased the susceptibility to FIRINOX treatment in SPF mice in comparison to control cells (mSt) or mSt-ATG4B cells without doxycycline treatment (Extended Data Fig. 8m).
Overall, these data suggest that when 3-IAA and MPO are present during FIRINOX treatment, the accumulation of ROS increases and the stress adaptation of cancer cells is impaired (that is, autophagy is downregulated), ultimately inducing the reduced proliferation of tumour cells.

3-IAA has therapeutic potential
To further investigate potential therapeutic implications of 3-IAA, we tested the effect of repetitive applications of 3-IAA and FIRINOX on mouse survival, treatment of other cancer entities and the efficacy of 3-IAA with a different chemotherapy combination, namely GnP. Up to three cycles of 3-IAA and FIRINOX together-but neither alone-significantly increased the duration of survival with orthotopic PDAC (Fig. 4a). Notably, 3-IAA was also synergistic with FIRINOX in the treatment of subcutaneous colorectal (MC38) or lung (LLC) tumours (Extended Data Fig. 9a,b) and also synergized with GnP in orthotopic PDAC (Extended Data Fig. 9c). These findings highlight the potential general role of 3-IAA in cancer treatment.
Finally, we wanted to assess the relevance of our findings in humans. We observed higher rates of neutropenia in patients who responded to chemotherapy, consistent with our mouse data and the known

Fig. 2 | The efficacy of 3-IAA and FIRINOX is licensed by MPO. a, SPF mice
were orthotopically injected with KPC cells and treated with FIRINOX with or without 3-IAA, and tumours were analysed at day three after FIRINOX treatment (n = 8 each). Immune subsets of orthotopic tumours or respective spleens were determined by flow cytometry. CD8 + T cells (CD3 + CD8 + ); CD4 + T cells (CD3 + CD4); macrophages (CD11b + F4/80 + ); neutrophils (CD11b + Ly6G + ) are shown as relative to total living immune cells (CD45 + ). b, Tumour weight and counts of immune subsets of orthotopic KPC tumours or respective spleens of irradiated and wild-type (WT; n = 4) or Mpo -/-(n = 5 or 6) bone-marrow (BM)-reconstituted mice treated and analysed as in a. c, Irradiated and wild-type, Mpo -/or Ahr -/bone-marrow-reconstituted mice received KPC cells orthotopically and were treated with FIRINOX or FIRINOX + 3-IAA for five days (n = 5 each). All mice received a four-day dietary intervention with a high-tryptophan diet. Tumour weight was assessed at day seven after FIRINOX treatment. Each symbol represents one mouse. Two independent experiments were pooled (a) or one out of two independent experiments (b,c) is shown. Error bars indicate s.e.m. Significant P values are indicated and were determined by two-tailed Mann-Whitney test (a) or two-tailed t-test (b,c). Article association between neutropenia and PDAC prognosis 22 (Extended Data Fig. 9d). Notably, the levels of neutrophil decrease after chemotherapy, but not lymphocyte or monocyte decrease, correlated with 3-IAA serum concentrations, as expected from their different levels of MPO ( Fig. 4b-d). Furthermore, we observed a significant correlation of the 3-IAA serum concentration and PFS or overall survival in our observation cohort (Hamburg cohort; Extended Data Table 1), which was absent for the secondary bile acid DCA (Fig. 4f,g and Extended Data Fig. 9e). We validated these findings in a second cohort of patients with mPDAC from the Ludwig-Maximilians-University (LMU) Hospital (Munich cohort; Extended Data Table 1) (Fig. 4h,i). Reflecting the missing value of other clinically assessable markers to predict a chemotherapy response a priori, neither patient-specific (age, gender and weight) nor tumour-specific (tumour size and tumour marker) variables were significantly associated with PFS ( Fig. 4e and Extended Data Fig. 9f).

Discussion
In summary, our results identify the microbiota-derived metabolite 3-IAA as a key amplifier of the response to chemotherapy in PDAC. When 3-IAA and MPO are present in high concentrations during FIRINOX treatment, the accumulation of ROS increases and the stress adaptation of cancer cells is impaired, which ultimately results in the reduced proliferation of PDAC cells. Despite the limited number of patients analysed (Hamburg cohort n = 23, of whom 21 had available serum samples; Munich cohort n = 24), we observed a robust correlation between 3-IAA serum concentrations measured during chemotherapy (Hamburg cohort) or even before the start of chemotherapy (Munich cohort) and PFS or overall survival.
These data set the early premise to initiate clinical trials that aim to raise the serum concentration of 3-IAA during chemotherapy treatment through direct treatment or dietary intervention, and ultimately even increase the survival of NR patients with PDAC. Increasing the 3-IAA substrate tryptophan through a specific diet could be easily achieved; however, this approach is affected by the composition of the microbiota, as suggested by the finding that 3-IAA serum concentrations only increased in R-microbiota-colonized mice after dietary intervention. Therefore, when it becomes possible to produce 3-IAA under good manufacturing practice standard conditions, direct treatment with 3-IAA would be ideal, especially considering its capacity to bypass the presence of an undesired microbiota, as shown by our experiments in mice colonized with a NR microbiota. d,e, The GFP-LC3B-RFP reporter cell line Hy19636_GLRM was injected into SPF mice and mice were treated as indicated (n = 5 each). The graphs show the GFP/ RFP ratio at day one (d) and tumour weight at day three (e) after FIRINOX treatment. Representative merged immunofluorescence images or indicated areas with a magnification of 3× are shown; scale bars, 10 μm. f, KPC cancer cells were orthotopically injected into R-microbiota-colonized mice and the indicated treatment was applied as shown in the scheme (n = 4 or 5). Tumour weight is shown at day 18 of the experiment. g, Tumour weight of orthotopic KPC tumours from gnotobiotic mice colonized with NR microbiota is shown nine days after the indicated treatment (n = 4 or 5). CQ, hydroxychloroquine. One experiment (c) or one out of two independent experiments (a,d,e-g) is shown. Each symbol represents one mouse. Error bars indicate s.e.m. Significant P values are indicated and were determined by one-way ANOVA followed by Dunnett's (a,c,f) or Tukey's (d,e) post-hoc test or Kruskal-Wallis test followed by Dunn's post-hoc test (g).
Before planning future clinical trials, it is important to consider that, in the absence of chemotherapy, indoles or other tryptophan metabolites can impair the development of anti-tumour immune responses through the AhR in PDAC 16 . However, our data show that four to five days of 3-IAA treatment without chemotherapy is not sufficient to affect the growth of the tumour, and only reduces the growth of the tumours via an AhR-independent mechanism when it is combined with chemotherapy. Nevertheless, more observational studies are needed before proceeding with any interventional ones.
Alternatively, faecal microbiota transplantation, which is being tested for the treatment of patients with cancer 8,9 , may be sufficient to increase the serum concentrations of 3-IAA and, accordingly, the response to chemotherapy, as shown in our gnotobiotic mice experiments using the stool of ten different human donors. Our study suggests that B. fragilis and B. thetaiotaomicron are enriched in the microbiota of R patients and are able to produce 3-IAA. However, larger studies are needed to reveal whether these bacteria are indeed responsible for the production of 3-IAA production in R patients, considering the vast number of taxonomically distinct bacterial species that are able to produce 3-IAA (refs. 14,15 ). Such studies are of interest to narrow the selection of donors for potential future studies of faecal microbiota transplantation.
In parallel to the effect of tumour size through treatment with 3-IAA and FIRINOX, we also observed a depletion of neutrophils in mouse models of PDAC. In patients with PDAC, we observed a positive correlation between the depletion of neutrophils during chemotherapy and the serum concentration of 3-IAA in the Hamburg cohort. These data suggest that, as in mice, 3-IAA can also contribute to the development of neutropenia in humans. However, given that other factors can lead to neutropenia, such as an altered chemotherapeutic metabolism, blocked excretion of chemotherapeutics or differing chemotherapeutic dosages 32 , studies that focus on this particular aspect are needed to firmly conclude whether there is a causative link between 3-IAA and neutropenia in humans. Furthermore, as neutrophil levels are a key predictor of survival in many cancer types, investigations of the described mechanism in other cancers may be warranted 33,34 .
Our findings have the potential to be transformative for the treatment of PDAC and other cancer types such as colorectal cancer, especially for patients who are treated with FIRINOX-based regimens. In addition, our findings will trigger the development of studies that address the effects of microbiota-derived metabolites on the ROSautophagy axis in response to chemotherapeutic treatments.

Online content
Any methods, additional references, Nature Portfolio reporting summaries, source data, extended data, supplementary information, acknowledgements, peer review information; details of author contributions and competing interests; and statements of data and code availability are available at https://doi.org/10.1038/s41586-023-05728-y.

Patients and human material
Patients who had been diagnosed with mPDAC and scheduled for treatment with GnP or FOLFIRINOX were recruited to the study (n = 30). Informed consent was obtained from all patients as approved by the ethics commission Hamburg (Ethikkommission der Ärztekammer Hamburg, Germany). Patients who received antibiotics during the first three months of treatment, did not deliver pre-treatment stool, did not start chemotherapy or suffered from COVID-19 were excluded from the analyses (n = 7). Stool was collected before the start of chemotherapy using a home sampling kit (OMNIgene, Gut OMR-200) for shotgun metagenomic sequencing. One sample was excluded after sequencing owing to overloading of the sample tube that led to unsuccessful fixation and limited representation of the microbiota. Therefore, the final cohort for microbiota analysis contained 22 patients. The classification of R and NR patients was primarily based on tumour shrinkage of at least 25%, which is calculated as a decrease of the primary tumour and largest metastasis in CT scans comparing the time point before treatment initiation to the time point that revealed the best response during first-line treatment as assessed by the local radiologist (20 to 30% of changes are usually considered significant in CT response evaluation 35 ). In cases of disease stabilization or missing CT scans, the criteria of a PFS of over 140 days (median PFS of real-world FOLFIRINOX and GnP cohorts 36 ) or a decline of at least 40% in serum tumour markers (prognostic cut-off in palliative PDAC treatment 37 ) were considered. Stool for functional experiments was collected in tubes without any additive, transported directly to the laboratory, subsequently diluted in 20% glycerol (Teknova, G1723), aliquoted and frozen at −80 °C. To preserve most of the donor microbiota for functional experiments, the maximum allowed turnaround time from toilet to freezer was 4 h. Blood was drawn at cycle three or four of chemotherapy treatment, mixed 1 to 1 with phosphate-buffered saline (PBS) and plasma was isolated using gradient centrifugation. Serum material of the Munich cohort was taken before the start of chemotherapy and processed according to local standards. Informed consent was obtained from all patients as approved by the ethics commission Munich (project number: 284-10).

Animal models
All mice used in this study were of a C57BL/6 background. All mice were used in accordance with the institutional review board 'Behörde für Soziales, Familie, Gesundheit und Verbraucherschutz' (Hamburg, Germany). Mice were kept under SPF or germ-free conditions, with an ambient temperature of 20 ± 2 °C, humidity of 55 ± 10% and a dark-light cycle of 12 h. Age-and sex-matched littermates between 4 and 16 weeks old were used for the most part. Mpo -/bone marrow used to establish bone marrow chimeras was provided by S. Baldus and M. Mollenhauer. Ahr -/bone marrow used to establish bone marrow chimeras was provided by C. Esser.
For colonization of human microbiota, stool was thawed, washed with brain heart infusion broth (BHI) and diluted in BHI. Two hundred microlitres of the suspension was gavaged once orally to gnotobiotic mice housed in isocages. Two to four weeks later, tumour experiments were initiated.
Sample sizes were calculated on the basis of small pilot experiments and mice were randomized before the beginning of treatment. The person treating the mice was not blinded because of the complex treatment schedules. For models of orthotopic PDAC, 5 × 10 4 -10 × 10 4 KPC, 2 × 10 5 Hy19636, 1 × 10 5 mSt-ATG4B/mSt or 1 × 10 5 knockdown KPC (AhR, GPX, scramble control) cells were injected orthotopically in a 1:1 mixture of PBS and Matrigel (Corning, 356231). Tumour weight was assessed at day 20 after tumour cell injection unless otherwise indicated. For subcutaneous tumour growth, 2.5 × 10 5 LLC-GFP or 1 × 10 6 MC38 cells were subcutaneously injected into the flank. Subcutaneous tumour growth was measured every other day using a caliper. Tumour weight was assessed at day 17 after tumour cell injection. The maximum allowed diameter of subcutaneous tumours for active experimental mice was 1.5 cm and limits were not exceeded.
For bone marrow chimeras, recipient mice were irradiated with 9.6 Gy (BioBEAM 2000) 24 h before bone marrow transfer. One day later, 1 × 10 6 -4 × 10 6 bone marrow cells isolated from wild-type, Ahr -/or Mpo -/mice were transferred by intravenous injection. Three to four weeks after transfer, cancer cells were orthotopically injected and treated as described below. Successful engraftment was validated by quantitative PCR (qPCR) on tumour-infiltrating immune cells.
Treatment with chemotherapy was initiated intraperitoneally (i.p.) at the indicated time point (usually day 11 after cancer cell injection). Oxaliplatin (Accord) 5 mg kg −1 , irinotecan (Accord) 20 mg kg −1 and 5-FU (Medac) 50 mg kg −1 were used for FIRINOX treatment as described in other studies 38,39 . Folinic acid was not used and gemcitabine (Hexal) 120 mg kg −1 and nab-paclitaxel (Celgene) 30 mg kg −1 were used for GnP treatment. Depletion of T cells was achieved by treatment with 200 μg anti-CD8 antibody clone 53-6.7 (BioXcell, BE0004) alone or in combination with 200 μg anti-CD4 antibody clone GK1.5 (BioXcell, BE0003) every third day i.p., beginning one day before chemotherapy treatment. Control mice were treated with the respective isotype control clone 2A3 (BioXcell, BE0089) at similar concentrations and intervals. Autophagy induction was achieved using two i.p. injections of trehalose (Sigma, T9449) at a concentration of 3 g kg −1 one day before and on the day of chemotherapy treatment. Similarly, 60 mg kg −1 chloroquine (Sigma, C6628) dissolved in PBS was injected i.p. one day before and on the day of chemotherapy treatment.
Treatment with 3-IAA (500 mg kg −1 ) was applied by oral gavage using indole-acetic-acid sodium salt (Sigma, I5148) dissolved in PBS every day for five consecutive days (two days before chemotherapy until two days after chemotherapy), unless otherwise indicated. Indole-3-propionic acid (3-IPA; Sigma, 57400), GCA (Sigma, G2878), hippuric acid (Sigma, 112003) and DCA (Sigma, 30960) were dissolved in 1 M NaOH in PBS and pH-adjusted to 7.4 using 1 M HCl. The solution was gavaged for five consecutive days orally at a concentration of 500 mg kg −1 (3-IPA, hippuric acid) or 250 mg kg −1 each (GCA and DCA). NAC (Sigma, a7250) was applied for five days in the drinking water ad libitum at a concentration of 1 g l −1 .
Dietary tryptophan modulation was initiated three days before chemotherapy until one day after treatment. In one experiment, tryptophan modulation was applied for a total of 14 days. Standard diet (2.3 g kg −1 tryptophan; Altromin, 1320) was changed to either synthetic crystalline AA tryptophan-free (0 g kg −1 ; SSNIFF, S9336-E701) diet or crystalline AA tryptophan-high (12 g kg −1 ; SSNIFF, S5714-E711) diet. Subsequently, the diet was changed back to the standard diet.
Dietary doxycycline (Sigma, D9891) was administered at a dosage of 625 mg kg −1 through the diet starting at day five or eight after cancer cell injection for a total of up to seven consecutive days. Subsequently, the diet was changed back to the standard diet.
For measurement of serum metabolites, blood was drawn at the end of the experiment or at the indicated time point. Blood was allowed to clot for 30 min and was centrifugated (1,000g) for 10 min thereafter. Serum was diluted and used as described in the specific section.

DNA extraction and shotgun metagenomics
For DNA extraction, samples were isolated with the ZymoBIOMICS 96 MagBead DNA kit (Zymo Research, D4302) and purified with Zymo DNA Clean and Concentrator-5 (Zymo Research, D4004). Libraries were prepared with the NEBnext Ultra II DNA Library Prep Kit for Illumina (New England Biolabs, E7645) with 150 ng of total DNA, size selection of 400-500 bp and 4× PCR cycles.
For shotgun metagenomics, Illumina library preparation was performed using the NEBNext Ultra II FS DNA Library Prep Kit (New England Biolabs, E7805). The library preparation was performed according to the manufacturer's instructions. The size selection Article was performed using AMPure XP beads (Beckman Coulter, A63882) and adaptor enrichment was performed using seven cycles of PCR using the NEBNext Multiplex oligos (New England Biolabs, E7335) from Illumina and then subjected to Illumina NovaSeq (2 × 150 bp) sequencing.
For bioinformatic analysis, raw reads were trimmed for low quality and filtered against the phix174 and human hg19 genome with bbduk (ref. 40 ). For taxonomic species profiling, all libraries were mapped against the Unified Human Gastrointestinal Genome collection (n = 4,644) using BBMap (refs. 41,42 ). Mapping rates were normalized into transcripts per million (TPM) and genomes with less than 10% genome coverage (genome breadth) were considered not prevalent in the sample. Data were summarized as metagenomics operational taxonomic units (OTUs) into biom format and analysed with phyloseq and LEfSe (refs. 43,44 ). Species-level functional profiling was performed with HUMAnN3 and also using the 9.9 million gene integrated reference catalogue of the human microbiome 41 .

Bacterial strains and isolation
Bacteroides thetaiotaomicron (DSM 2079), B. fragilis (DSM 2151) and Prevotella copri (DSM 18205) were obtained from the German Collection of Microorganisms and Cell Cultures (DSMZ). For bacteria isolation, faecal samples frozen in glycerol were thawed and streaked out anaerobically in serial dilutions on BHI blood agar plates (5% defibrinated sheep's blood & vitamin K 3 ) supplemented with vancomycin. After growing the agar plates inside an incubator at 37 °C for 2 days, single colonies were picked into BHI-S medium in a 96-well plate and were further incubated for 24 h. The resulting bacterial cultures were screened by PCR using specific primers for B. thetaiotaomicron (B. theta F 5′-GAGGGTGTCGTATTTCCGAAGG-3′ R 5′-GTTCCCTGATCCAGTGTGTTGG-3′) or B. fragilis (B. frag F 5′-AATGATTCCGCATGGTTTCA-3′ R 5′-ATTTTGGGATTAGCATACGG-3′). After passaging PCR-positive wells on agar plates to obtain pure cultures and additional confirmation of identity by Sanger sequencing, resulting strains were maintained in BHI-S until further use. All bacterial strains were cultured and maintained anaerobically in BHI broth supplemented with 10% fetal bovine serum (FBS) and vitamin K 3 (BHI-S).

Extraction of bacterial supernatant
BHI-S medium supplemented with 1% tryptophan was inoculated from a fully grown overnight bacterial culture (1:50 ratio) and was incubated anaerobically until the early exponential phase. Cultures were taken out of the anaerobic chamber and were centrifuged at 4,700 rpm (4,816g) for 5 min at room temperature. The supernatant was removed and immediately frozen at −20 °C.
16S rRNA sequencing DNA from tumours was extracted using the DNeasy Blood & Tissue Kit (Qiagen, 69504). Approximately 10 mg tissue was digested with Proteinase K in ATL buffer at 56 °C for 1 h. Afterwards, samples were processed according to the manufacturer's protocol. Blank extraction controls were included during extraction of samples.
Variable regions V1 and V2 of the 16S rRNA gene were amplified using the primer pair 27F-338R in a dual-barcoding approach according to a previous report 45 . For tumours, 3.5 μl DNA was used for amplification and PCR products were verified using agarose gel electrophoresis. Final PCR products were normalized using the SequalPrep Normalization Plate Kit (Thermo Fisher Scientific, A1051001), pooled equimolarly and sequenced on the Illumina MiSeq v3 2×300bp (Illumina). Demultiplexing after sequencing was based on 0 mismatches in the barcode sequences. Data processing was performed using the DADA2 46 workflow for big datasets (https://benjjneb.github.io/dada2/bigdata.html; the workflow adjusted for the V1-V2 region can be found here: https:// github.com/mruehlemann/ikmb_amplicon_processing/blob/master/ dada2_16S_workflow.R), resulting in abundance tables of amplicon sequence variants (ASVs). ASVs underwent taxonomic annotation using the Bayesian classifier provided in DADA2 and using the Ribosomal Database Project (RDP) version 16 release. Sequences that were not assignable to genus level were binned into the finest possible taxonomic classification.

Metabolomic screen
We used ultra-high-pressure liquid chromatography-tandem mass spectrometry (UPLC-MS/MS) to acquire data in both positive and negative ionization modes that allowed for identification and quantification of 630 metabolites. Plasma samples were processed using the MxP Quant 500 Kit (Biocrates) according to the manufacturer's instructions. In brief, 10 μl of plasma sample, calibration standard and control sample were transferred onto a filter containing internal standards for internal standard calibration. Filters were dried under a stream of nitrogen using a pressure manifold (Waters). Samples were incubated with derivatization reagent phenyl isocyanate for 60 min. After drying under nitrogen, analytes were extracted with 5 mmol l −1 ammonium acetate in methanol and the eluate was further diluted for the UPLC-MS/MS analysis. The targeted analysis covered 630 metabolites (https://biocrates.com/mxp-quant-500-kit/) detected by MS/MS after UPLC separation and flow injection analysis (FIA). Each measurement required two UPLC runs and three FIA runs to cover all metabolites. All analyses were performed on an ACQUITY UPLC I-Class system (Waters) coupled to a Xevo TQ-S mass spectrometer (Waters). Reversed-phase chromatographic separation was accomplished using a C18 LC-column (Biocrates) with 0.2% formic acid in water with 0.2% formic acid in acetonitrile as the eluent system. The FIA solvent was methanol, with a modifier provided by the kit manufacturer. Data analysis of the UPLC-MS/MS results was based on a seven-point curve or one-point calibration and internal standard normalization. Values below the lower threshold were set to zero. Concentration data were analysed using MetaboAnalyst v.5. Concentrations were log-transformed before analysis and raw P values and log 2 -transformed fold change values are shown in the graphics of Fig. 1.

3-IAA and DCA CLIA
To quantify 3-IAA or DCA serum concentrations, mouse or human serum was diluted 1 to 10 with PBS and the chemiluminescence immune assay (CLIA) (Abbexa, 3-IAA abx190011; DCA 258844) was performed according to the manufacturer's protocol. For detection of 3-IAA in cultures, supernatant was processed as described above and directly used for the assay. Chemiluminescence was detected using FLUOstar Omega (BMG Labtech) with 1-s sampling per well, and a gain of 3,400 to 4,000 was individually adjusted for each assay. The concentration was determined using supplied standards and interpolated using Prism 8.4.0.

3-IAA and MOI measurement in tumour tissue
Frozen tumour tissue was mixed with ice-cold methanol 1:1-3 and homogenized using 0.4-mm beads at 5,500 rpm for 2 × 30 s using a homogenizer (Precyllys 24 touch). Extracts were centrifuged (10,000g) at 4 °C and the supernatant was used for LC-MS/MS analysis. Reversed-phase chromatography was conducted using a biphenyl stationary phase (Raptor Biphenyl (Restek), dimensions: 50 mm × 2.1 mm ID; particle size: 2.7 μm) with eluent A: water + 0.1% formic acid + 5 mM ammonium acetate and eluent B: methanol + 0.1% formic acid + 5 mM ammonium acetate. The flow rate was set to 0.5 ml per min. Elution starts with 95% eluent A, which linearly decreases to 75% over 0.5 min. This composition was held for 4 min before returning to initial conditions. The injected volume was 2 μl and the column temperature was set to 55 °C. 3

Isolation of immune cells and flow cytometry
Tumours were taken and cut into similar-sized pieces. Tumours were rinsed with cold PBS and digested in RPMI (Sigma, 61870044) supplemented with 10% FBS (Gibco, 10500064), 2.5 mg ml −1 collagenase D (Roche, 11088866001) and 0.2 mg ml −1 DNase l (Roche, 11284932001) for 35 min at 37 °C with continuous shaking. Afterwards, the suspension was strained through a 40-μm cell strainer and quenched with cold PBS. Subsequently, immune or tumour cells were stained with Fc block and live/dead staining (Thermo Fisher Scientific L34957 and L10119) for 30 min in the dark. Afterwards cells were washed, stained with the indicated flow cytometry antibodies and incubated for 30 min in the dark again. Flow cytometry was performed on a Fortessa flow cytometer (BD). To assess the cytokine profile of immune cells, restimulation of T cells with 50 ng ml −1 PMA, 500 ng ml −1 ionomycin and 1 μg ml −1 brefeldin A was performed for 3 h at 37 °C. After surface staining, cells were fixed and permeabilized using the eBioscience Foxp3 intracellular staining kit (00-5523-00). The following intracellular antibodies were used: IFNγ (1:200 dilution) and TNF (1:400 dilution). The following surface antibodies were used to classify lymphocytes (CD3 (1:300 600 dilution)). Epithelial cancer cells were stained with EPCAM (1:200 dilution) antibody. For Ly6B staining, the goat anti-rat IgG antibody (1:400 dilution) was used. Cells were counted by flow cytometry using 2,500 to 5,000 beads (Spherotec, ACBP 100-10). Viability of neutrophils in vitro was assessed using PI/ Annexin V staining according to the manufacturer's protocol. ROS expression was determined in immune cells in vivo after tissue digestion as described above. Afterwards, the ROS dye CellROX (Thermo Fisher Scientific, C10422) was stained in parallel with flow cytometry antibodies at a dilution of 1:1,000. Software analysis and histogram generation was carried out using FlowJo v.10.
Cell culture KPC cells were obtained from Ximbio under catalogue number 153474; Hy19636_GLRM reporter cells and mSt-ATG4B/mSt cells were provided by A. Kimmelman 31 ; MC38 and LLC-GFP (ATCC) cells were provided by A. Giannou; and MIA PaCa-2, BxPC-3 and T3M-4 cells (all ATCC) were provided by C. Güngör. All cells tested negative for mycoplasma contamination. Cells were maintained under standard conditions at 37 °C and 5% CO 2 and regularly visually inspected. Cells were grown in DMEM GlutaMAX (Thermo Fisher Scientific, 10566016) supplemented with penicillin and streptomycin (Gibco, 15140122) and 10% FBS (Gibco, 10500064). For in vitro experiments, cancer cells were plated at 5,000-10,000 cells per well in a 96-well plate. Cells were allowed to seed overnight, except in experiments with neutrophil co-culture, treatment of GPX-knockdown cells or co-treatment with MPO. Subsequently, treatment with indicated compounds and treatment duration was initiated. 3-IAA or 3-IPA (3-IPA, Sigma, 57400; 3-IAA, Sigma, I3750) were dissolved in 1 M NaOH in PBS and PH-adjusted to 7.4 using 1 M HCl or DMSO when indicated. NAC was diluted in PBS and used at a concentration of 1 mM. H 2 O 2 (Sigma, H1009) was used at a concentration of 400 μM.
Tumour cell viability and proliferation was assessed using a MTT or MTS assay (Abcam, ab201191 and ab197010) according to the manufacturer's protocol. Absorbance was assessed using the FLUOstar Omega (BMG Labtech). In other experiments, viability was assessed by flow cytometry using SYTOX (Thermo Fisher Scientific, S34857), PI (Biolegend, 421301) and 2,500-5,000 counting beads (Spherotec, ACBP 100-10) as a reference. In parallel, intracellular ROS was measured using CellROX (Thermo Fisher Scientific, C10422) according to the manufacturer's protocol and assessed by flow cytometry.
Neutrophils or T cells were isolated from bone marrow or spleens of healthy, untreated mice and sorted using fluorescence-activated cell sorting (FACS) as LY6G + CD11b + and TCRβ + CD4 + or CD8 + , respectively. Cells were seeded at a concentration of 20,000-50,000 per well in a 96-well plate. Viability of neutrophils or T cells was assessed by flow cytometry using SYTOX (Thermo Fisher Scientific, S34857) or PI/Annexin V staining (Biolegend,640914) at the indicated time of the experiment. In some experiments, 200 mU ml −1 MPO or 400 mU ml −1 MPO (Merck, 475911) or indicated concentrations of MOI (Sigma, 493397) were added. Degranulation of neutrophils was assessed by the release of MPO. A total of 1 × 10 6 neutrophils were sorted using FACS as described above. Neutrophils were incubated in HBSS (Gibco, 14065-56) and the indicated treatment, or N-formylmethionyl-leucyl-phenylalanine (fMLP; Merck, F3506), as a positive control, was added. After 30 min of incubation, MPO activity was measured using an MPO activity assay kit (Abcam, ab105136) according to the manufacturer's protocol. NET formation of neutrophils was determined using SYTOX stain after 3 h of incubation with the indicated compounds or 100 nM phorbol-12-myristate-13-acetate (PMA; Merck, 524400) as a positive control. NETs were measured as SYTOX-positive cells using flow cytometry.

Protein screen
Tumour tissue from three individual mice per group was pooled and proteins were extracted using scioExtract buffer (Sciomics). The samples were labelled for 2 h with scioDye 2 (Sciomics) at an adjusted protein concentration. The reference sample was labelled with scioDye 1 (Sciomics). After 2 h, the reaction was stopped and the buffer was exchanged with PBS. All labelled protein samples were stored at −20 °C until use. The samples were analysed in a dual-colour approach using a reference-based design on scioDiscover antibody microarrays (Sciomics). The arrays were blocked with scioBlock (Sciomics) on a Hybstation 4800 (Tecan) and afterwards the samples were incubated competitively with the reference sample using a dual-colour approach. After incubation for 3 h, the slides were thoroughly washed with 1× PBSTT, rinsed with 0.1× PBS as well as with water and subsequently dried with nitrogen. Slide scanning was conducted using a Powerscanner (Tecan) with constant instrument laser power and PMT settings. Spot segmentation was performed with GenePix Pro 6.0 (Molecular Devices). Acquired raw data were analysed using the (LIMMA) package of R-Bioconductor after uploading the median signal intensities. For normalization, a specialized invariant Lowess method was applied. Downregulated proteins (M-value of < −0.35) or upregulated proteins (>0.35) in the 3-IAA and FIRINOX sample were uploaded to the STING database (https://string-db.org/cgi/input.pl). The standard pipeline was used to perform KEGG enrichment analysis. Only pathways with a false discovery rate (FDR) < 0.05 were considered statistically significant.

RNA extraction and mRNA sequencing
Fresh frozen tumour tissue was lysed in TRIzol reagent (Thermo Fisher Scientific, 15596018) and RNA was extracted using the chloroform-isopropanol method. mRNA was purified from total RNA using poly-T oligo-attached magnetic beads. After fragmentation, the first-strand cDNA was synthesized using random hexamer primers followed by second-strand cDNA synthesis. The library was ready after end repair, A-tailing, adapter ligation, size selection, amplification and purification. The library was checked with Qubit and qPCR for quantification, and Bioanalyzer for size distribution detection. Quantified libraries were sequenced on the Illumina platform with at least 60 million reads per sample.
Sequence reads were processed with fastp (v0.20.1) to remove sequences of sequencing adapters and low-quality (Phred quality score below 15) sequences from the 3′ end of the sequence reads. Afterwards, reads were aligned to the mouse reference assembly (GRCm39.104) using STAR (v.2.7.9a) 47 . Differential expression was assessed with DESeq2 (ref. 48 ). A gene was considered significantly differentially expressed if the corresponding absolute log 2 -transformed fold change (log2FC) was no less than 0.6 and, in addition, the FDR did not exceed a value of 0.1. The gene set enrichment analysis (GSEA) of the Reactome pathway Autophagy (R-HSA-9612973) was performed using fgsea (v.4.1) 49 .

qPCR
Total RNA was extracted from cell lines using Trizol Reagent (Invitrogen,15596018) and the total RNA extraction kit (Qiagen, 74004/74104) according to the manufacturer's protocol. The High-Capacity cDNA Synthesis Kit (Thermo Fisher Scientific, 4368813) was used for cDNA synthesis. Primers and probes were purchased from Applied Biosystems. Mouse primers and probes are: Gpx3 (Mm00492427_m1), Gpx7 (Mm00481133), Ahr (Mm00478932_m1). qPCR was performed using the TaqMan Master Mix (Thermo Fisher Scientific, 4369016) on the Ste-pOne Plus system (Applied Biosystems). Forty to 44 cycles were applied for every assay and technical doublets or triplicates were used. If at least two out of three values of the technical replicates were undetectable, the expression was considered non-detectable. Relative expression was normalized to Gapdh (Mm99999915_g1).

Lentiviral transfer of shRNAs
Lentiviral vectors expressing short hairpin RNAs (shRNAs) under control of the human U6 promoter (MISSION pLKO.1-puro) directed against mouse Gpx3 (TRCN000076539), mouse Gpx7 (TRCN0000076563) and mouse Ahr (TRCN0000218025) as well as a nontargeted control shRNA (SHC002, scrambled) were obtained from Sigma-Aldrich. The production of lentiviral particles has been described in detail elsewhere 50 , and protocols are available online (http://www.LentiGO-Vectors.de). For the transduction of KPC cells with the HIV-1-derived lentiviral vectors, 2.5 × 10 4 cells were plated in 0.5 ml medium with 8 μg ml −1 polybrene per well of a 24-well plate. After plating, the addition of 10 μl VSV-G pseudotyped, non-concentrated lentiviral particles led to the stable integration of shRNAs and the puromycin resistance gene into the cell's genome. To increase the transduction rate by spin-inoculation, the plate was centrifuged at 1,000g and 25 °C for 1 h. The selection of successfully transduced cells with 1 μg ml −1 puromycin in the culture medium was started four days after transduction.
Quantification for LC3B, Ki67, nitrotyrosine and CC3 was done in a blinded manner. Positive cells were determined using ImageJ v.2.1.0/1.53c and the threshold was adjusted according to the staining intensity of the respective antibody and maintained for all tumours analysed with the same staining. The number of positive cells was counted per 250-500 × 500-μm field (10× to 40× magnification) in three to five fields per sample.

Widefield microscopy
To visualize the GFP-LC3-RFP reporter signal expression of Hy19636_ GLRM PDAC cells, tissues were fixed in 2% PFA solution at 4 °C overnight, incubated in PBS containing 30% sucrose and embedded in Tissue-Tek OCT compound (Sakura Finetek) on dry ice. For further analysis, 7-μm sections were used. Widefield imaging was performed using the THUN-DER Imager 3D Live Cell and 3D Cell Culture (Leica Microsystems) equipped with a 40× 1.10 NA water immersion objective. LED power and exposure time and other system-specific settings were first optimized using positive control tissue and not changed between image acquisitions of the different groups to provide optimal comparability. For each tissue section, an average of five regions of interest were randomly selected and imaged for subsequent quantitative analysis. ImageJ imaging software was used for file navigation, adjustment of colour balance and image analysis. GFP and RFP quantification was determined using ImageJ v.2.1.0/1.53c. The mean fluorescence intensity for each respective signal was determined per slide in at least five areas per tumour.

Statistical analyses
All statistical analyses, unless otherwise indicated, were performed using GraphPad Prism 9.3.1. Normality and log-normality were tested using Shapiro-Wilk or Kolmogorov-Smirnov tests. If normality was not given, non-parametric testing was performed. Tests were performed two-sided if not otherwise indicated and resulting significant (P < 0.05) P values are shown.

Reporting summary
Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Data availability
RNA-seq data have been submitted to the European Nucleotide Archive (ENA). They are publicly available under accession number PRJEB58222. Shotgun metagenomic sequencing data were filtered for human reads and are available under accession number PRJEB58222. Source data are provided with this paper. . Strains were all cultured in BHI containing 1% tryptophan, the strain BFi2 was also cultured in BHI alone (n = 3 biological replicates). j, SPF mice were orthotopically injected with KPC cells. Mice were either supplemented with tryptophan-free, standard diet (2.3 g/kg TRP) or tryptophan-high diet (12 g/kg TRP) starting 7 days after tumour cell injection for a total of 14 days (n = 3). k, As in j, except dietary intervention was applied for a total of 4 days before standard diet was reintroduced. Tumour weight was assessed 9 days after dietary intervention (n = 3 or 4). l, NR microbiota-colonized mice (red) were fed either standard diet (SD) or tryptophan-high diet for 4 days (TRP high). Blood was drawn at day 4 of intervention and 3-IAA was measured in the serum using 3-IAA CLIA (n = 4). m, Tumour sizes of mice from l at day nine after treatment with FIRINOX are shown (n = 3 or 4). Each symbol represents one mouse, human or in vitro replicate. One experiment was performed (f,i-m) or four (e,g) independent experiments were pooled. Error bars indicate SEM, whiskers indicate 10% and 90% (h), significant p-values are indicated and were determined by Gehan-Breslow-Wilcoxon test (b,c), one-way ANOVA followed by Tukey's (e,m) or Dunnett's (i-k) post-hoc test, two-way ANOVA (g), twotailed Wilcoxon paired-ranks test (h) or two-tailed t-test (l). Fig. 3 | 3-IAA induces a response to FIRINOX in mouse models of PDAC. a, 3-IAA serum concentration of SPF mice gavaged with 500 mg/kg 3-IAA was measured using CLIA at 2, 6 and 24 h after 3-IAA application (n = 3 mice). Unrelated serum concentration of R-microbiota-colonized gnotobiotic mice is shown as a control. b, SPF mice were orthotopically injected with KPC cells and received two treatments of FIRINOX or FIRINOX + 3-IAA (n = 5). Tumour weight at day 20 of the experiment is depicted in the statistic. c, NR-colonized gnotobiotic mice were injected with KPC tumour cells orthotopically. 9 days later, mice were substituted +/− 3-IAA for five days (d9-13) and treated +/− FIRINOX as depicted in the experimental scheme (n = 3 or 4). Tumour weight is shown at day 20 of the experiment. d, SPF mice were orthotopically injected with KPC tumour cells. At day 9 after tumour cell injection, mice were treated with either 500 mg/kg 3-IAA, 3-IPA or 250 mg/kg GCA and 250 mg/kg DCA for five consecutive days (n = 4). FIRINOX treatment was applied at day 11. Tumour weight is shown at day 20 of the experiment. e, As in d, except mice were treated with FIRINOX, FIRINOX + 3-IAA (500 mg/kg) or FIRINOX + hippuric acid (500 mg/kg) (n = 4 or 5). Each symbol represents one mouse. One experiment (e) or one out of two independent experiments is shown (a-d). Error bars indicate SEM, significant p-values are indicated and were determined by Kruskal-Wallis test followed by Dunn's post-hoc test (a,c,e), or one-way ANOVA followed by Tukey's (b) or Dunnett's (d) post-hoc test.

Extended Data Fig. 4 | Tumour-infiltrating immune cells of KPC tumours
from gnotobiotic mice. a, Flow cytometric gating strategy to classify immune subsets (left panel) or intracellular cytokines and coinhibitory receptors (right panel). Immune subsets were determined using the following marker combinations on live CD45 + immune cells: CD8 + T cells (CD11b − CD3 + CD8 + ); CD4 + T cells (CD11b − CD3 + CD4 + ); neutrophils (CD11b + Ly6G + ); and macrophages (CD11b + Ly6G − F4/80 + ). b, Immune subsets or cytokine profiles and PD-1 expression of immune cells from orthotopic tumours obtained from gnotobiotic mice colonized with R or NR microbiota was analysed at day 20 of the experiment. Each symbol represents one mouse. Three to five independent experiments were pooled (n = 11 to 23 in total). Error bars indicate SEM, significant p-values are indicated and were determined by two-tailed t-test.

Extended Data Fig. 5 | CD4 + and CD8 + T cells are dispensable for the efficacy of 3-IAA and FIRINOX.
R-microbiota-colonized gnotobiotic mice were injected with KPC tumour cells orthotopically. Before treatment with FIRINOX, mice were injected with either isotype control antibody or CD8-depleting antibody every third day, as indicated. The tumour weight (a) and the depletion efficacy of intratumoral CD8 + T cells (b) as relative to total immune cells (CD45 + ), determined by flow cytometry, is shown at day 20 of the experiment (n = 3). c, As in a, except that CD4 + and CD8 + T cells were depleted simultaneously (n = 3 or 4). d, R-microbiota-colonized gnotobiotic mice with orthotopic KPC tumours were treated with FIRINOX, FIRINOX + tryptophan-high diet (d8-12) + isotype control antibody or FIRINOX + tryptophan-high diet (d8-12) + CD4/CD8depleting antibody (n = 4 or 5). Tumour weight at day 20 of the experiment is depicted. e, As in c, except that SPF mice were treated with 3-IAA +/− FIRINOX and CD4/CD8-depleting or isotype control antibody (n = 4). Tumour weight is depicted at day 20 of the experiment. Each symbol represents one mouse. One experiment each was performed. Error bars indicate SEM, significant p-values are indicated and were determined by two-tailed t-test (a,b), one-way ANOVA followed by Dunnett's (c,d) post-hoc test or Kruskal-Wallis test followed by Dunn's post-hoc test (e).
Corresponding author(s): Joseph Tintelnot, Nicola Gagliani Last updated by author(s): Dec 14, 2022 Reporting Summary Nature Portfolio wishes to improve the reproducibility of the work that we publish. This form provides structure for consistency and transparency in reporting. For further information on Nature Portfolio policies, see our Editorial Policies and the Editorial Policy Checklist.

Statistics
For all statistical analyses, confirm that the following items are present in the figure legend, A description of all covariates tested A description of any assumptions or corrections, such as tests of normality and adjustment for multiple comparisons A full description of the statistical parameters including central tendency (e.g. means) or other basic estimates (e.g. regression coefficient) AND variation (e.g. standard deviation) or associated estimates of uncertainty (e.g. confidence intervals) For null hypothesis testing, the test statistic (e.g. F, t, r) with confidence intervals, effect sizes, degrees of freedom and P value noted

Software and code
Policy information about availability of computer code Data collection Flow cytometry data were collected using FACSDiva version 8 on a Fortessa flow cytometer; widefield images were obtained using Leica Application Suite X software v.3.6.0; MTT and CLIA were analyzed using FLUOstar Omega; Protein screen was acquired using GenePix Pro 6.0., Illumina MiSeq v3 was used for 16S rRNA sequencing and Illumina NovaSeq for shotgun metagenomic sequencing; Xevo TQ-S mass spectrometer was used for metabolomic screening; StepOne Plus system for rt-PCR and NanoZoomer 2.0-HT for slide scanning

Data analysis
The following software were used for data analysis: MetaboAnalyst v. 5

Data
Policy information about availability of data All manuscripts must include a data availability statement. This statement should provide the following information, where applicable: -Accession codes, unique identifiers, or web links for publicly available datasets -A description of any restrictions on data availability -For clinical datasets or third party data, please ensure that the statement adheres to our policy Field-specific reporting Please select the one below that is the best fit for your research. If you are not sure, read the appropriate sections before making your selection.

Life sciences Behavioural & social sciences Ecological, evolutionary & environmental sciences
For a reference copy of the document with all sections, see nature.com/documents/nr-reporting-summary-flat.pdf

Life sciences study design
All studies must disclose on these points even when the disclosure is negative.

Sample size
Sample sizes were based on experience with the described models as well as small pilot experiments with up to 5 mice per group for new models. Regarding human data, no sample size was calculated; rather metabolomic screening was performed on a small pilot group. In the following time of recruitment, all eligible patients were allowed to enroll into the study.
Data exclusions No data or animals were excluded from analysis, except for clear technical failure.

Replication
Experiments were repeated multiple times with similar results as indicated in the figure legends. All critical in vivo experiments were reproduced using the same system (e.g. Chemotherapy treatment in R colonized mice using the same donor and mice) and/ or using orthogonal approaches (e.g. endogenous 3-IAA production by R microbiota or 3-IAA gavage in SPF or NR microbiota colonized mice).
Randomization Prior to treatment initiation mice bearing tumors were randomized. Tumor sizes were not measured at the time of randomization.
Human study participants were not randomized, because the caring physician decided about the treatment and the individual response to therapy defined the allocation to either R or NR group.

Blinding
Blinding was not performed in mouse experiments. The investigator needed to know the treatment groups in order to perform the study.
Tumor weights (an objective measurement) and tumor sizes were carried out only at the study endpoints after mice were euthanized and tumors were harvested, if possible together with a blinded rater. No active blinding was performed for human materials since patients were treated as defined by the caring physician and the investigator needed to know the response of the specific patient in order to define the allocation to the R or NR group. Still, at the time of stool and serum sample collection, the response of the individual patient was not defined.

Reporting for specific materials, systems and methods
We require information from authors about some types of materials, experimental systems and methods used in many studies. Here, indicate whether each material, system or method listed is relevant to your study. If you are not sure if a list item applies to your research, read the appropriate section before selecting a response. The following antibodies were used for IHC: anti-LC3B antibody (Thermo Fischer Scientific, PA1-46286), anti-Nitrotyrosine antibody