**3. Results**

### *3.1. General Effects of NIMP Intoxication*

The severity of NIMP intoxication was evaluated based on our intoxication scale (Figure 1a). This scale includes 12 relevant behavioral changes, from no behavioral perturbation to death, the maximum on our scale. Almost all mice exposed to 0.5 LD50 of NIMP displayed fasciculation and face stereotypies (chewing, yawning), considered as light intoxication symptoms. On the other hand, the majority of the 0.9 LD50-exposed mice presented long-lasting convulsions. This dose was thus selected to be the high sublethal dose in our study. The mean intoxication scores were 2.5 ± 0.2 for 0.5 LD50 and 10.4 ± 0.2 for the 0.9 LD50-exposed mice (Figure 1b). Six animals (10.7%) exposed to the highest dose and none exposed to the lowest dose died in the first day following the NIMP challenge. No delayed death was observed.

As expected, measurements of cerebral ChE activities in mice exposed to 0.9 LD50 revealed a significant important total ChE inhibition 6 h after intoxication (5.4 ± 0.6% of ChE activity). Late measurements showed a persisting significant inhibition until 1-month post-intoxication (72.5 ± 6.7% of ChE activity), with recovery being complete 6 months post-exposure (Figure 1c). Exposure to 0.5 LD50 induced a large inhibition of ChE activity (36.9 ± 4.6%) at 6 h post-exposure. The recovery appeared incomplete at 1 month (81.7 ± 5%) and 6 months after exposure at this dose (77 ± 13%); however, it was not significant from CTL activity (Figure 1c). Interestingly, the profile of AChE activity was the same as the total ChE activity for both doses (Figure 1d), suggesting that in the CNS, the majority of ChE activity is due to AChE activity.

Finally, the evolution in weight gain during the 6 months post-exposure revealed a significant loss of weight for only the 0.9 LD50-exposed mice during the first week after intoxication. One month after exposure, the change in mean body weight of the exposed animals reached the change in body weight of the CTL group. Interestingly, the 0.5 LD50- exposed mice did not display any difference in weight gain at any time compared to the CTL group (Figure 1e).

### *3.2. Anxiety-like Behavior Modification*

Anxiety-related behaviors were evaluated every month after NIMP exposure. To avoid any test habituation, a new anxiety test was performed for each measurement. The initial test, which was performed before any intoxication, showed no difference in terms of the time passed exploring the center zone in the square area open-field and was thus considered as the anxiety level baseline (Figure 2a). One month after NIMP intoxication, the time passed in the open arms of the EPM significantly decreased in the NIMP-intoxicated animals compared to the CTL (Figure 2b). The number of rearings on the 5th step of the staircase measured on the second month post-exposure did not show any significant difference between groups, even though this number was slightly decreased in both NIMP intoxication groups (Figure 2c). In contrast, the recovery was complete 3 months post-exposure, since the time passed in the light box during the dark–light box test was equivalent between groups (Figure 2d). Four months post-intoxication, the time passed exploring the center zone in the circular area open-field decreased drastically in the 0.9 LD50-exposed mice group, while the 0.5 LD50-exposed mice did not show any difference with the CTL group (Figure 2e). Finally, the two NIMP-intoxicated animal groups at 5 and 6 months post-NIMP exposure showed a significant increase in the anxiety index, as measured by the decreased frequency of food zone exploration in the neophobia test and the time passed in the open arms in the O-maze (Figure 2f,g).

**Figure 1.** General effects of NIMP intoxication. (**a**) Intoxication scale based on the most relevant behavioral signs. (**b**) Average intoxication score induced by the two doses of interest: 2.55 ± 0.17 for 0.5 LD50 and 10.38 ± 0.16 for 0.9 LD50. (**c**) ChE activity at different timepoints post-NIMP exposure: 6 h (100 ± 11.5 for 0 LD50; 37.0 ± 4.6 for 0.5 LD50 and 5.4 ± 0.6 for 0.9 LD50); 24 h (100 ± 13.6 for 0 LD50; 43.0 ± 4.5 for 0.5 LD50 and 11.3 ± 1.3 for 0.9 LD50); 3 days (100 ± 9.6 for 0 LD50; 63.5 ± 8.8 for 0.5 LD50 and 29.0 ± 4.4 for 0.9 LD50); 7 days (100 ± 6.3 for 0 LD50; 64.8 ± 3.2 for 0.5 LD50 and 39.7 ± 3.3 for 0.9 LD50); 1 month (100 ± 6.7 for 0 LD50; 81.7 ± 5.0 for 0.5 LD50 and 72.5 ± 6.7 for 0.9 LD50) and 6 months (100 ± 17.2 for 0 LD50; 77.0 ± 13.7 for 0.5 LD50 and 98.9 ± 14.6 for 0.9 LD50); (*n* = 8 per group). Significant differences were determined by one-way ANOVA (F = 51.45, *p* < 0.0001 for 6H; F = 26.45, *p* < 0.0001 for 24H; F = 19.82, *p* < 0.0001 for 3D; F = 42.6, *p* < 0.0001 for 7D; F = 4.99, *p* = 0.02 for 1M and F = 0.76, *p* = 0.48 for 6M) with Dunnett's post hoc test compared to the CTL group

(\*\*\*\* *p* < 0.0001; \*\*\* *p* < 0.001; \*\* *p* < 0.01; \* *p* < 0.05). (**d**) AChE activity at different timepoints post-NIMP exposure: 6 h (100 ± 11.9 for 0 LD50; 37.1 ± 4.5 for 0.5 LD50 and 5.3 ± 0.7 for 0.9 LD50); 24 h (100 ± 14.6 for 0 LD50; 44.3 ± 5.4 for 0.5 LD50 and 12.6 ± 1.3 for 0.9 LD50); 3 days (100 ± 8.7 for 0 LD50; 68.0 ± 12.0 for 0.5 LD50 and 37.9 ± 6.0 for 0.9 LD50); 7 days (100 ± 5.6 for 0 LD50; 68.9 ± 4.2 for 0.5 LD50 and 40.9 ± 2.3 for 0.9 LD50); 1 month (100 ± 6.8 for 0 LD50; 82.2 ± 5.2 for 0.5 LD50 and 75.3 ± 7.2 for 0.9 LD50) and 6 months (100 ± 19.3 for 0 LD50; 74.8 ± 14.1 for 0.5 LD50 and 93.5 ± 15.1 for 0.9 LD50); (*n* = 8 per group). Significant differences were determined by one-way ANOVA (F = 48.94, *p* < 0.0001 for 6H; F = 24.4, *p* < 0.0001 for 24H; F = 10.86, *p* = 0.0006 for 3D; F = 45.38, *p* < 0.0001 for 7D; F = 3.7, *p* = 0.04 for 1M and F = 0.67, *p* = 0.52 for 6M) with Dunnett's post hoc test compared to the CTL group (\*\*\*\* *p* < 0.0001; \*\*\* *p* < 0.001; \*\* *p* < 0.01; \* *p* < 0.05). (**e**) Body weight change relative to baseline (*n* = 14 for 0 LD50; *n* = 15 for 0.5 LD50 and *n* = 16 for 0.9 LD50 group). Significant differences were determined by two-way repeated measures ANOVA (FTime = 282.8, *p* < 0.0001; FDose = 2.3, *p* = 0.11; FTimexDose = 1.67, *p* = 0.0009) with Tukey's multiple comparisons test (\*\*\* *p* < 0.001; \*\* *p* < 0.01; \* *p* < 0.05 CTL vs. 0.9 LD50; ΦΦ *p* < 0.01 0.5 LD50 vs. 0.9 LD50).

**Figure 2.** Development of anxiety-like behavior caused by NIMP intoxication. Anxiety-like behaviors were evaluated in NIMP-exposed and CTL animals using a different test every month (*n* = 14 for 0 LD50; *n* = 15 for 0.5 LD50 and *n* = 16 for 0.9 LD50 group). (**a**) Open-field, square area, at the baseline with the percentage of time spent in the center zone (9.6 ± 1.3 for 0 LD50; 9.9 ± 1.2 for 0.5 LD50 and 9.4 ± 1.4 for 0.9 LD50). (**b**) EPM at 1 month with the percentage of time spent in the open arms (28.1 ± 3.1 for 0 LD50; 18.9 ± 3.0 for 0.5 LD50 and 13.8 ± 1.9 for 0.9 LD50). (**c**) Staircase test, 2 months post-intoxication with the number of rearings on the top step (11.5 ± 0.8 for 0 LD50; 9.1 ± 1.1 for 0.5 LD50 and 9.4 ± 0.9 for 0.9 LD50). (**d**) Dark-light box at 3 months with the percentage of time spent in the light box (48.6 ± 2.8 for 0 LD50; 46.5 ± 3.2 for 0.5 LD50 and 47.5 ± 3.2 for 0.9 LD50). (**e**) Open-field, circular area, at 4 months with the percentage of time spent in the center zone (21.1 ± 3.8 for 0 LD50; 20.2 ± 2.7 for 0.5 LD50 and 11.9 ± 2.9 for 0.9 LD50). (**f**) Neophobia test at 5 months showing the frequency of visits to the center platform (11.8 ± 2.3 for 0 LD50; 6.5 ± 1.2 for 0.5 LD50 and 6.9 ± 1.2 for 0.9 LD50). (**g**) O-maze at 6 months post-exposure with the percentage of time spent in the open arms (29.4 ± 2.7 for 0 LD50; 19.2 ± 2.8 for 0.5 LD50 and 17.2 ± 2.6 for 0.9 LD50). Significant differences were determined by one-way ANOVA (F = 0.04, *p* = 0.9 for BL; F = 7.1, *p* = 0.002 for 1M; F = 1.6, *p* = 0.2 for 2M; F = 0.11, *p* = 0.89 for 3M; F = 2.8, *p* = 0.08 for 4M; F = 3.2, *p* = 0.04 for 5M and F = 5.4, *p* = 0.008 for 6M) with Dunnett's post hoc test compared to the CTL group (\*\* *p* < 0.01; \* *p* < 0.05).

### *3.3. Longitudinal Imaging Examination and Long-Term Induced Inflammation*

To explain these behavioral long-term modifications, we conducted MRI in parallel on 0.5 and 0.9 LD50 NIMP-exposed mice for 6 months to evaluate morphological modifications and brain edema after exposure. The image analysis did not show significant neuroanatomical changes using T2-weighted images or edema formation using ADC mapping on DWI images (Figure 3).

**Figure 3.** NIMP exposure does not induce morphological modifications or brain edema in the amygdala/piriform cortex. (**a**) T2 images with the apparent diffusion coefficient (ADC) map of one representative NIMP-exposed mouse (0.9 LD50) at different timepoints: before intoxication (baseline), and 3 days, 7 days, 1 month and 6 months after NIMP exposure. Regions of interest used for analyses are shown with red circles. (**b**) The percentage of ADC normalized to the baseline as a function of different timepoints for NIMP-exposed animal groups [(0.5 LD50: −5.1 ± 2.7 at 3D; −6.0 ± 3.5 at 7D; 1.7 ± 3.9 at 1M and −2.9 ± 3.1 at 6M) and (0.9 LD50: −0.8 ± 3.1 at 3D; 2.1 ± 4.3 at 7D; 2.9 ± 5.2 at 1M and −0.5 ± 2.7 at 6M)]. Statistical analyses were conducted by mixed-effects model (REML) analysis (FTime = 0.89, *p* = 0.44; FDose = 1.5, *p* = 0.24; FTimexDose = 0.36, *p* = 0.78) (*n* = 7 to 9 per group).

Next, cerebral cytokine levels were evaluated in the amygdala/pyriform cortex extracts using Luminex technology. A slight increase in several cytokine levels was observed in the NIMP-exposed mice group 6 months post-exposure (Figure 4a). Thus, sublethal doses of NIMP did not induce any robust long-term neuroinflammation. In addition, histological exploration showed no persistent microglia reactivity 6 months post-intoxication in the amygdala of exposed animals (Figure 4b,c), confirming the previous cytokine results.

**Figure 4.** NIMP exposure does not induce long-term neuroinflammation in the amygdala/piriform cortex. (**a**) Expression levels of the cytokines KC, IL-1<sup>α</sup>, IL-10, IL-9 and IL-17 normalized to CTL values at 6 months post-intoxication [(KC: 1.0 ± 0.2 for 0 LD50; 1.3 ± 0.2 for 0.5 LD50 and 1.5 ± 0.2 for 0.9 LD50); (IL-1α: 1.0 ± 0.2 for 0 LD50; 1.1 ± 0.3 for 0.5 LD50 and 1.3 ± 0.2 for 0.9 LD50); (IL-10: 1.0 ± 0.2 for 0 LD50; 1.1 ± 0.5 for 0.5 LD50 and 1.4 ± 0.2 for 0.9 LD50); (IL-9: 1.0 ± 0.1 for 0 LD50; 1.3 ± 0.3 for 0.5 LD50 and 1.3 ± 0.1 for 0.9 LD50) and (IL-17: 1.0 ± 0.1 for 0 LD50; 1.2 ± 0.3 for 0.5 LD50 and 1.3 ± 0.1 for 0.9 LD50)]. Statistical analyses were conducted by one-way ANOVA (*n* = 8 per group). (**b**) IBA-1 labeling in the amygdala region of CTL (**left**) and 0.9 LD50 (**right**) mice. Scale bar = 50 μm. (**c**) Fold change of IBA1 labeling the optical density percentage normalized to CTL at 6 months post-intoxication: 1.0 ± 0.1 for 0 LD50; 0.98 ± 0.2 for 0.5 LD50 and 1.0 ± 0.2 for 0.9 LD50. Statistical analyses were conducted by Kruskal–Wallis test (*p* = 0.83; *n* = 7 per group).

Because the impact of a long-term systemic inflammation could alter the nervous system, we evaluated the evolution of WBC after NIMP exposure. A first phase could be observed, with significant WBC modifications in the 0.9 LD50-exposed mice group. Two timepoints, one day and 7 days after NIMP exposure, showed significant decreases in lymphocytes, monocytes and eosinophils in the highest dose-exposed group, whereas the lowest dose induced little or no modifications in most cell types (Figure 5). One-month post-intoxication, no difference was observed between the NIMP-exposed groups or in comparison to the CTL group. Surprisingly, a second phase was also observed between 3 and 6 months post-NIMP exposure, with significant modifications to WBC quantities in the 0.5 LD50-exposed mice group. With the exception of basophils and eosinophils, which were significantly decreased in the two exposed groups, the 0.5 LD50-exposed mice group presented significant long-term modification in the proportion of lymphocytes, monocytes and neutrophils compared to the 0.9 LD50-exposed and CTL groups (Figure 5).

Although serum cytokine levels were not significantly increased 6 months postexposure, the expression of some cytokines was slightly enhanced in the 0.9 LD50-exposed mice group (+183% ± 146% for G-CSF; +46% ± 26% for KC; +66% ± 35% for IL-1 α). Interestingly, the expression of granulocyte-macrophage colony-stimulating factor (GM-CSF or CSF2) as well as IL-17 was slightly elevated in the 0.5 LD50-exposed mice group (+184% ± 131% and 86% ± 58%, respectively) (Figure 6).

### *3.4. Gut Modification*

The consequences of chronic systemic low-level inflammation could have a significant impact on multiple physiological systems. Among the possible targets, we investigated the impact on intestinal function. A gu<sup>t</sup> morphological evaluation was thus conducted 6 months post-NIMP exposure. Histological modifications were observed for both NIMP sublethal doses (Figure 7a). The quantification of goblet cell numbers and the ratio of intestinal villi size revealed a significant decrease in the 0.5 and 0.9 LD50-exposed mice groups (goblet cell numbers compared to CTL: 77.2% ± 8.3% and 78.7% ± 3.5%, respectively; intestinal villi size compared to CTL: 75.8% ± 3% and 69.6% ± 2.7%, respectively) (Figure 7b,c). Moreover, lymphoid foci numbers were significantly increased in both exposed mice groups (×3 ± 1.1 and ×5.5 ± 1.3 for 0.5 LD50 and 0.9 LD50, respectively) (Figure 7d). Finally, we conducted an evaluation of the Gram+/Gram − ratio. A dysbiosis of Gram − microbiota was observed in both NIMP-treated groups (Gram+/Gram − ratio: 0.55 ± 0.09 and 0.47 ± 0.11 for 0.5 LD50 and 0.9 LD50, respectively) (Figure 7e). This level of gu<sup>t</sup> dysbiosis can trigger the innate immune response and chronic low-grade inflammation, leading to many age-related degenerative pathologies and unhealthy aging, which could in turn influence the gu<sup>t</sup> microbiota composition [30].

### *3.5. Altered Microbiota Composition*

A total of 788,900 16S rDNA reads was analyzed to establish the gu<sup>t</sup> microbiota composition of the mice included in the present study. Our analyses led to the detection of 873 OTUs, evenly distributed among the mice groups (CTL: 551 ± 41, 0.5 LD50: 562 ± 54 and 0.9 LD50: 561 ± 44). No differences were detected regarding gu<sup>t</sup> bacterial diversity and richness (Figure 8a) for mice challenged with NIMP 6 months post-exposure in comparison to the CTL group. As displayed in Figure 8b, nine predominant bacterial genera (Alistipes, Barnesiella, Bacteroides, Prevotella, Odoribacter, Alloprevotella, Clostridium\_XIVa, Bilophila and Oscillibacter) shaped the gu<sup>t</sup> microbiota composition of the mice, with highly similar abundances between groups. Furthermore, we observed a moderate shift in the gu<sup>t</sup> microbiota composition of NIMP-exposed mice in comparison to the CTL group (Figure 8c). Two clusters consisting of CTL and NIMP-exposed mice (both 0.5 and 0.9 LD50 doses) were identified in a principal coordinate analysis (PCoA). This moderate shift can mostly be explained by higher abundances of Turicibacter (in both 0.5 and 0.9 LD50 groups) and Parabacteroides (0.9 LD50 group only) and a lower abundance of Coprococcus in NIMP-exposed mice compared to the CTL group (Figure 8d).

**Figure 5.** Evolution in white blood cell count after NIMP exposure. (**a**) Lymphocyte count at different timepoints after NIMP exposure relative to CTL: 24 h (1.0 ± 0.01 for 0 LD50; 0.93 ± 0.02 for 0.5 LD50 and 0.64 ± 0.06 for 0.9 LD50); 7 days (1.0 ± 0.02 for 0 LD50; 0.85 ± 0.03 for 0.5 LD50 and 0.72 ± 0.04 for 0.9 LD50); 1 month (1.0 ± 0.004 for 0 LD50; 0.85 ± 0.07 for 0.5 LD50 and 0.87 ± 0.05 for 0.9 LD50); 3 months (1.0 ± 0.003 for 0 LD50; 0.98 ± 0.01 for 0.5 LD50 and 0.99 ± 0.02 for 0.9 LD50) and 6 months (1.0 ± 0.12 for 0 LD50; 0.81 ± 0.05 for 0.5 LD50 and 0.94 ± 0.02 for 0.9 LD50); (*n* = 8 to 10 per group). Significant differences were determined by one-way ANOVA (F = 22.52, *p* < 0.0001 for 24H; F = 13.33, *p* = 0.0002 for 7D; F = 5.21, *p* = 0.02 for 1M; F = 0.32, *p* = 0.73 for 3M and F = 8.51, *p* = 0.0023 for 6M with Tukey's multiple comparisons test. (**b**) Monocyte count at different timepoints after NIMP exposure relative to CTL: 24 h (1.0 ± 0.04 for 0 LD50; 0.69 ± 0.08 for 0.5 LD50 and 0.60 ± 0.13 for 0.9 LD50); 7 days (1.0 ± 0.02 for 0 LD50; 1.41 ± 0.12 for 0.5 LD50 and 1.68 ± 0.12 for 0.9 LD50); 1 month (1.0 ± 0.04 for 0 LD50; 1.01 ± 0.12 for 0.5 LD50 and 1.22 ± 0.02 for 0.9 LD50); 3 months (1.0 ± 0.06 for 0 LD50; 1.24 ± 0.05 for 0.5 LD50 and 1.33 ± 0.11 for 0.9 LD50) and 6 months (1.0 ± 0.07 for 0 LD50; 1.53 ± 0.13 for 0.5 LD50 and 0.90 ± 0.03 for 0.9 LD50); (*n* = 8 to 10 per group). Significant differences were determined by one-way ANOVA (F = 5.54, *p* = 0.015 for 24H; F = 6.39, *p* = 0.0065 for 7D; F = 2.88, *p* = 0.095 for 1M; F = 4.46, *p* = 0.024 for 3M and F = 15.78, *p* < 0.0001 for 6M with Tukey's multiple comparisons test. (**c**) Neutrophil count at different timepoints after NIMP exposure relative to CTL: 24 h (1.0 ± 0.02 for 0 LD50; 0.93 ± 0.08 for 0.5 LD50 and 2.12 ± 0.23 for 0.9 LD50); 7 days (1.0 ± 0.06 for 0 LD50; 1.37 ± 0.11 for 0.5 LD50 and 1.76 ± 0.15 for 0.9 LD50); 1 month (1.0 ± 0.02 for 0 LD50; 1.77 ± 0.33 for 0.5 LD50 and 1.59 ± 0.29 for 0.9 LD50); 3 months (1.0 ± 0.01 for 0 LD50; 1.21 ± 0.04 for 0.5 LD50 and 1.04 ± 0.03 for 0.9 LD50) and 6 months (1.0 ± 0.03 for 0 LD50; 1.34 ± 0.11 for 0.5 LD50 and 0.92 ± 0.06 for 0.9 LD50); (*n* = 8 to 10 per group). Significant differences were determined by one-way ANOVA (F = 27.24, *p* < 0.0001 for 24H; F = 7.68, *p* = 0.0028 for 7D; F = 3.7, *p* = 0.067 for 1M; F = 12.19, *p* = 0.0003 for 3M and F = 9.50, *p* = 0.0009 for 6M with Tukey's multiple comparisons test. (**d**) Basophil count at different timepoints after NIMP exposure relative to CTL: 24 h (1.0 ± 0.03 for 0 LD50; 1.17 ± 0.04 for 0.5 LD50 and 1.60 ± 0.17 for 0.9 LD50); 7 days (1.0 ± 0.03 for 0 LD50; 1.40 ± 0.09 for 0.5 LD50 and 1.44 ± 0.08 for 0.9 LD50); 1 month (1.0 ± 0.01 for 0 LD50; 1.10 ± 0.06 for 0.5 LD50 and 0.99 ± 0.04 for 0.9 LD50); 3 months (1.0 ± 0.05 for 0 LD50; 0.77 ± 0.04 for 0.5 LD50 and 0.65 ± 0.03 for 0.9 LD50) and 6 months (1.0 ± 0.03 for 0 LD50; 0.77 ± 0.03 for 0.5 LD50 and 0.69 ± 0.03 for 0.9 LD50); (*n* = 8 to 10 per group). Significant differences were determined by one-way ANOVA (F = 7.218, *p* = 0.0041 for 24H; F = 9.08, *p* = 0.0016 for 7D; F = 1.84, *p* = 0.19 for 1M;

F = 14.12, *p* < 0.0001 for 3M and F = 20.04, *p* < 0.0001 for 6M with Tukey's multiple comparisons test. (**e**) Eosinophil count at different timepoints after NIMP exposure relative to CTL: 24 h (1.0 ± 0.09 for 0 LD50; 1.07 ± 0.09 for 0.5 LD50 and 0.54 ± 0.05 for 0.9 LD50); 7 days (1.0 ± 0.04 for 0 LD50; 0.76 ± 0.02 for 0.5 LD50 and 1.02 ± 0.08 for 0.9 LD50); 1 month (1.0 ± 0.02 for 0 LD50; 1.87 ± 0.30 for 0.5 LD50 and 1.84 ± 0.26 for 0.9 LD50); 3 months (1.0 ± 0.04 for 0 LD50; 0.70 ± 0.05 for 0.5 LD50 and 0.80 ± 0.04 for 0.9 LD50) and 6 months (1.0 ± 0.02 for 0 LD50; 0.71 ± 0.08 for 0.5 LD50 and 0.56 ± 0.11 for 0.9 LD50); (*n* = 8 to 10 per group). Significant differences were determined by one-way ANOVA (F = 8.17, *p* = 0.0045 for 24H; F = 7.52, *p* = 0.0042 for 7D; F = 3.95, *p* = 0.058 for 1M; F = 12.92, *p* = 0.0007 for 3M and F = 6.13, *p* = 0.0088 for 6M) with Tukey's multiple comparisons test. (\*\*\*\* *p* < 0.0001; \*\*\* *p* < 0.001; \*\* *p* < 0.01; \* *p* < 0.05 CTL vs. 0.5 LD50; +++ *p* < 0.001; ++ *p* < 0.01; + *p* < 0.05 CTL vs. 0.9 LD50; ΦΦΦΦ *p* < 0.01; ΦΦΦ *p* < 0.001; ΦΦ *p* < 0.01; Φ *p* < 0.05 0.5 LD50 vs. 0.9 LD50).

**Figure 6.** NIMP does not induce long-term systemic inflammation. Expression levels of the serum cytokines GM-CSF, M-CSF, G-CSF, KC, IL-1<sup>α</sup>, IL-10, IL-9 and IL-17 normalized to CTL values at 6 months post-intoxication [(GM-CSF: 1.0 ± 0.2 for 0 LD50; 2.8 ± 0.13 for 0.5 LD50 and 1.1 ± 0.01 for 0.9 LD50); (M-CSF: 1.0 ± 0.3 for 0 LD50; 0.72 ± 0.10 for 0.5 LD50 and 1.08 ± 0.3 for 0.9 LD50); (G-CSF: 1.0 ± 0.3 for 0 LD50; 1.63 ± 0.4 for 0.5 LD50 and 2.8 ± 1.46 for 0.9 LD50); (KC: 1.0 ± 0.2 for 0 LD50; 1.1± 0.08 for 0.5 LD50 and 1.46 ± 0.2 for 0.9 LD50); (IL-1α: 1.0 ± 0.1 for 0 LD50; 1.2 ± 0.3 for 0.5 LD50 and 1.7 ± 0.4 for 0.9 LD50); (IL-1α: 1.0 ± 0.07 for 0 LD50; 1.2 ± 0.3 for 0.5 LD50 and 1.06 ± 0.2 for 0.9 LD50); (IL-9: 1.0 ± 0.2 for 0 LD50; 0.68 ± 0.1 for 0.5 LD50 and 0.85 ± 0.07 for 0.9 LD50); (IL-17: 1.0 ± 0.3 for 0 LD50; 1.86 ± 0.6 for 0.5 LD50 and 1.43 ± 0.6 for 0.9 LD50)]. Statistical analyses were conducted by one-way ANOVA (*n* = 8 per group).

By combining other metadata related to anxiety-like behavior status and mononuclear blood cell percentages with gu<sup>t</sup> microbiota composition at the genus level, we identified multiple significant associations linking the host with specific bacterial taxa (Figure 9). Notably, blood monocyte levels were positively correlated with the abundance of Aestuariispira (r = 0.62) and Parasutterella (r = 0.71) and negatively correlated with the levels of Bilophila (r = −0.61), Flavonifractor (r = −0.61), Gemmiger (r = −0.61), Hydrogenoanaerobacterium (r = −0.69), Oscillibacter (r = −0.67) and Pseudoflavonifractor (r = −0.70). Neutrophil and lymphocyte percentages were positively and negatively associated with Ruminococcus abundance (r = 0.75 and −0.77, respectively). The Intestinimonas genera was associated with anxiety status parameters such as cumulative open area activity (r = 0.62).

**Figure 7.** Long-term morphological changes in the large intestine induced by NIMP exposure. (**a**) Large intestine pictures of CTL (top panel), 0.5 LD50 (middle panel) and 0.9 LD50 (bottom panel) mice. Scale bar = 100 μm. Mucin-producing goblet cells (arrow heads) were identified in colon sections of CTL mice. The double arrow indicates an example of villus size in a 0.5 LD50 NIMPexposed mouse; the asterisk indicates a lymphoid focus in a 0.9 LD50 NIMP-exposed mouse. Fold change in (**b**) goblet cell numbers (1.0 ± 0.03 for 0 LD50; 0.77 ± 0.08 for 0.5 LD50 and 0.79 ± 0.03 for 0.9 LD50), (**c**) villi size (1.0 ± 0.10 for 0 LD50; 0.76 ± 0.03 for 0.5 LD50 and 0.70 ± 0.03 for 0.9 LD50) and (**d**) lymphoid foci numbers (1.0 ± 0.28 for 0 LD50; 3.02 ± 1.12 for 0.5 LD50 and 5.46 ± 1.31 for 0.9 LD50) normalized to CTL values at 6 months post-intoxication. Significant differences were determined by one-way ANOVA (F = 5.5, *p* = 0.02 for goblet cell numbers; F = 6.87, *p* = 0.01 for villi size; F = 4.89, *p* = 0.028 for lymphoid foci numbers) with Tukey's multiple comparisons test. (\* *p* < 0.05 CTL vs. NIMP-treated mice). (**e**) Gram−stained stool smears in the large intestine of CTL (top panel), 0.5 LD50 (middle panel) and 0.9 LD50 (bottom panel) mice. Arrow heads illustrate Gram+ bacteria stained blue, while Gram− bacteria are stained pink (X100). (**f**) The ratio of Gram+/Gram− bacteria normalized to CTL animals (1.0 ± 0.7 for 0 LD50; 0.55 ± 0.09 for 0.5 LD50 and 0.47 ± 0.11 for 0.9 LD50). Significant differences were determined by one-way ANOVA (F = 7.6, *p* = 0.0047) with Tukey's multiple comparisons test. (\*\* *p* < 0.01; \* *p* < 0.05 CTL vs. NIMP-treated mice).
