2.1 Oral exposure to ZnONPs did not significantly alter the α-diversity of microbiota but change the gut compositions in mice
The workflow diagram for this study was showed in the Fig. 1. After orally exposed to ZnONPs by intragastric administration for 30 days, no significant alterations on the body weight in ZnONPs-treated mice in comparison to vehicle controls were found. Also, no mortality was observed during the treatment of ZnONPs (data not shown). The fecal pellets of two groups were then collected at the end of administration and subjected to the 16S rRNA sequencing immediately. The microbial community profiles were produced by clustering 16S rRNA sequences into operational taxonomic units (OUT, ≥ 97% sequence match). The coverage indexes of both two groups exceeded 99%, indicating a high level of diversity coverage in the samples. The fecal microbiota from the two groups showed a similar Shannon index (P = 0.73), Simpson index (P = 0.43), phylogenetic diversity (P = 0.52) and observed species (P = 0.75) (Fig. 2A-2D). These results indicated that there were no significant differences on within-sample (α) phylogenetic diversity analysis between the two groups. Intriguingly, the results of PCoA demonstrated ZnONPs-treatment obviously changed the gut microbial community compositions as compared with vehicle controls (Fig. 3A). We further observed the relative abundances of Actinobacteria were significantly increased in ZnONP-treated group in comparison to vehicle group (Fig. 3B). To identify the gut microbiota primarily responsible for the discrimination between the two groups, the linear discriminant analysis effect size (LEfSe) was performed. This analysis identified 15 different bacterial clades for the discrimination between two groups (Fig. 4A and 4B).
2.2 Oral exposure to ZnONPs caused the spatial learning and memory deficits and locomotor inhibition in mice
Morris water maze is a classic and widely used test for examining spatial learning and memory ability . To determine whether ZnONPs exposure affect the spatial learning and memory ability or not, the Morris water maze was carried out. Our results clearly showed the escape latency and swim distance in ZnONPs-administrated animals were much higher than those of vehicle controls in the navigation task (Fig. 5A and 5B). No significant change was found on the swim speed between two groups (Fig. 5C). On the day 5 of probe trial without hidden platform, we found the number of platform crossings was sharply reduced by ZnONPs and similar trend was depicted in the time spent in target quadrant (Fig. 5D and 5E). The representative track maps of two groups in the hidden platform and probe trial tests were showed in Fig. 5F. These results together indicate that exposure to ZnONPs by oral administration leads to the impairments in the spatial learning and memory.
To further detect the impact of ZnONPs on the locomotor activity, the open field test was performed because it is a commonly used test for the measurement of exploratory behavior and general activity in animals . As shown in Fig. 6A, a significant reduction on the total distance in ZnONPs group when compared with vehicle group was found. Meanwhile, both the central square duration and the distance moved in the center were dramatically decreased in ZnONPs-treated mice as compared with vehicle ones (Fig. 6B and 6C). The representative track maps of two groups in the open field test were shown in Fig. 6D. These findings together indicate oral exposure to ZnONPs is capable to induce the inhibitory effects on locomotor activity.
To reveal the potential molecular mechanisms involved, the candidate genes associated with the functions of learning and memory and locomotor were determined by qPCR assay. Our results revealed that the mRNA expressions of Bdnf and Dlg4 were both enhanced by ZnONPs in the hippocampus as compared with vehicle controls (Fig. 7A and 7B). To our surprise, no apparent alteration was observed on the expression of Grin2a between two groups (Fig. 7C), implying that ZnONPs-induced neurobehavioral dysfunctions might be regulated by the specific genes.
2.3 Oral exposure to ZnONPs remarkably changed the levels of plasma and hippocampal metabolites in mice
Next, we carried out non-targeted metabolomics to investigate whether or which metabolisms modulated by gut microbiome were paralleled by an altered gut-brain axis. As shown in Fig. 8, 30 differentially expressed metabolites (n = 26 from the plasma and n = 4 from the hippocampus) were identified between the two groups. Among these metabolites, we found 18 metabolites were of particular relevance to lipids and lipid-like molecules. As compared to vehicle controls, QH(2), Cortisol, DG(22:5(7Z,10Z,13Z,16Z,19Z)/18:4(6Z,9Z,12Z,15Z)/0:0), Taurocholic acid, Turanose, Tauroursocholic acid, SM(d18:0/18:1(9Z)), Sphinganine, Camellianin A, sodium taurocholate, Sulfamerazine levels in ZnONP group were significantly elevated. On the contrary, the levels of 28-Galloylglucosylpomolate 3-arabinoside, 3-Oxocholic acid, Chenodeoxycholic acid disulfate, Cholic acid, DG(22:5(7Z,10Z,13Z,16Z,19Z)/15:0/0:0), Myzodendrone, LysoPE(22:1(13Z)/0:0), DG(15:0/0:0/22:4n6), PGP(18:3(6Z,9Z,12Z)/18:1(9Z)), PIP2(16:1(9Z)/16:0), (E,E,E)-Sylvatine, PG(16:0/16:1(9Z)), Octanoylglucuronide, PGP(18:0/18:0), erythromycin B(1+), LMST01070013, NPC, Piromidic acid and sodium globostellatate B in the ZnONPs were dramatically reduced in comparison to the vehicle group. Hierarchical clustering heat map constructed using these differential metabolites also showed a consistent clustering pattern within individual groups.
2.4 Relationships among metabolites and bacterium after oral exposure to ZnONPs
Increasing evidence indicates that perturbations of the gut microbiome and its influence on metabolic functions may play an essential role in the development of various disease [18, 19]. Therefore, to determine the relationships among metabolites and bacterium after oral exposure to ZnONPs, the correlation pairs of metabolite-bacterium were then analyzed. As shown in Fig. 9A, the loading values of each differential metabolite and bacterial clade were calculated by rCCA. The loading values of 14 differential metabolites (Cortisol, 3-Oxocholic acid, LMST01070013, (E,E,E)-Sylvatine, PGP(18:0/18:0), PGP#, LysoPE(22:1(13Z)/0:0), PG(16:0/16:1(9Z)), CAD, NPC, 28-G, DG(15:0/0:0/22:4n6), Octanoylglucuronide, Myzodendrone and seven differential bacterial clades (g_Sutterella, g_Adlercreutzia, c_Actinobacteria, p_Actinobacteria, f_Bifidobacteriaceae, g_Bifidobacterium and o_Bifidobacteriales) were higher than the others, indicating their greater contributions to the overall correlation between these differential metabolites and bacterial clades. The spearman correlation analysis also showed the similar results (Fig. 9B): these 14 differential metabolites and seven differential bacterial clades had greater contributions to the overall correlation. Our data also showed that four groups of “metabolite-bacterium” correlation pairs stood out, including: 1) cortisol with 13 differential bacterial clades; 2) LMST01070013 with 12 differential bacterial clades; 3) g_Sutterella with 26 differential metabolites; and 4) g_Adlercreutzia with 17 differential metabolites (Fig. 9B).
2.5 Relationships among Bdnf, Dlg4, metabolites and bacterium after oral exposure to ZnONPs
The spearman correlation analysis showed that Bdnf was significantly correlated with five differential bacterial clades (f_Bifidobacteriaceae, p_Actinobacteria, c_Actinobacteria, g_Bifidobacterium and o_Bifidobacteriales) and two differential metabolites ((E,E,E)-Sylvatine, DG(15:0/0:0/22:4n6)). In addition, we found Dlg4 was positively correlated with eight differential metabolites (DG(15:0/0:0/22:4n6), Taurocholic acid, Tauroursocholic acid, Piromidic acid, 3-Oxocholic acid, Turanose, erythromycin B(1+), DG(22:5(7Z,10Z,13Z,16Z,19Z)/18:4(6Z,9Z,12Z,15Z)/0:0)) (Fig. 10).