131I therapy changes the structure of the gut microbiota in patients with differentiated thyroid cancer

According to the flat rarefaction curves, sufficient sequencing data were gathered, and the amount of sequencing data was acceptable (Fig. S1A). In the diversity analysis, α-diversity assessed by the Ace and Chao1 indices of the gut microbiota was significantly decreased after 131I therapy (Fig. 2A). Both binary_jaccard and unweighted_uniFrac indices analyzed using principal coordinates analysis (PCoA) presented obvious changes (Fig. 2B, C; Fig. S1B, C), indicating that 131I therapy can greatly alter the composition of the gut microbiota. To further probe which bacterial taxa were distinct between the two groups, we compared the abundance distribution and found that the relative abundance of Lachnospiraceae was much higher at both family and genus levels (Fig. 2D, E). Additionally, linear discriminant analysis (LDA) effect size (LefSe) analysis at the genus level was performed to further corroborate the data on the representative taxa, and Lachnospiraceae was identified as the most predominant bacterial taxon after 131I treatment (Fig. 2F, G). Therefore, we speculated that Lachnospiraceae might play a crucial role in radiation protection after 131I treatment.

Fig. 2

131I therapy changes the structure of gut microbiota in patients with DTC (n = 102). A α-diversity analysis in Ace and Chao1 indices between patients before 131I therapy [131I ( −)] and after 131I therapy [131I ( +)]. β-diversity analysis on principal coordinates (PC) analysis for binary_jaccard (B) and unweighted_unifrac (C) indices [131I ( +) vs 131I ( −)]. Abundance distribution on family (D) and genus (E) level [131I ( +) vs 131I ( −)]. F Linear discriminate analysis (LDA) effect size (LEfSe) revealing differential microbiota on genus level [131I ( +) vs 131I ( −)]. G Relative abundance of three genera in Lachnospiraceae identified by LefSe [131I ( +) vs.131I ( −)]

131I therapy alters the intestinal metabolite landscape of patients with DTC

Next, we assessed the metabolome of the enteric flora to further explore the effects of 131I therapy on gut metabolites (Fig. 3). Orthogonal partial least-squares discrimination analysis (OPLS-DA) score plots (Fig. S2A-D) and principal components analysis (PCA) (Fig. 3A, B) revealed clear and separate clusters in the positive and negative ion modes between patients with DTC before and after 131I therapy. These results indicated that there were significant differences in the abundance of metabolites between the two groups, visualizing with volcano plots (Fig. 3C, D, more details in Supplemental Excel1). Compared with the before 131I therapy group, the differential metabolites with significantly decreased levels in positive ion mode were 9-HODE and 13-OxoODE in the after 131I therapy group (Fig. 3C). Pathway functional enrichment analysis showed that linoleic acid (LA) metabolism was robustly inhibited (Fig. 3E). In the negative ion mode, the metabolites with differentially decreased levels in the after 131I therapy group were 5,6-DHET and 20-COOH-LTB4 (Fig. 3D), which were found to be enriched in the ARA metabolism pathway by Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis (Fig. 3F).

Fig. 3

131I therapy alters the intestinal metabolite landscape of patients with DTC (n = 102). 3D principal components (PC) analysis for binary_jaccard index in positive (A) and negative (B) ion mode between patients before 131I therapy [131I ( −)] and after 131I therapy [131I ( +)]. Volcano plot of differential metabolites (fold change ≤ 0.5 or > 2) in positive (C) and negative (D) ion mode [131I ( +) vs 131I ( −)]. E Bubble plot of KEGG pathway analysis (most inhibited green marked) in positive ion mode [131I ( +) vs 131I ( −)]. F Differential abundance of KEGG pathways (most inhibited green marked) in negative ion mode [131I ( +) vs 131I ( −)]. G Spearman correlation network diagram among three core genera and four core lipid metabolites. H 3D scatter plot of three core lipid metabolites [131I ( +) vs 131I ( −)]. I KEGG metabolic pathways in LA metabolism [131I ( +) vs 131I ( −)]. J Heatmaps of targeted oxylipins contents [131I ( +) vs 131I ( −)]. K Box plots of alterations in ARA, ± 5,6 − DHET, 9-HODE, 13-OxoODE contents [131I ( +) vs.131I ( −)]. L Spearman correlation network diagram among ARA, ± 5,6 − DHET, 9-HODE, 13-OxoODE. *p < 0.05, ***p < 0.001

Besides, heatmaps of the correlation between the levels of the top 45 differential metabolites (identified in the negative ion mode and plotted in a volcano plot) and the top 25 differential microbes by LefSe were analyzed (Fig. S2E). The abundance of Lachnospiraceae was negatively associated with the levels of LA metabolites (9-HODE and 13-OxoODE) and ARA metabolites (5,6-DHET and 20-COOH-LTB4) (Fig. 3G, Fig. S2F). The strong connection between LA and ARA metabolites was subsequently shown by a 3D scatter plot (Fig. 3H) and validated by the KEGG database (Fig. 3I in LA metabolism, Fig. S2G in ARA metabolism).

To further validate the results of non-targeted metabolomic analysis, especially those regarding ARA-/LA-related metabolites, we performed targeted metabolomic analysis of oxylipins in patients with DTC (n = 5) and identified 62 metabolites (Fig. 3J). Surprisingly, not only the content of LA metabolites (9-HODE and 13-OxoODE) and ARA metabolites (5,6-DHET) but also that of ARA was significantly lower in DTC patients after 131I therapy than in patients who were not treated (Fig. 3K). As expected, ARA metabolites (5,6-DHET) were strongly associated with LA metabolites (13-OxoODE) (Fig. 3L). Since ARA had a higher content in vivo and higher availability than ARA and LA metabolites (Fig. 3K) [13] and served as downstream of LA metabolism (Fig. 3I), we intended to explore the radioprotective effects of ARA instead of ARA metabolites or LA metabolites in our subsequent study.

Oral gavage of ARA improves survival, decreases inflammatory response, and ameliorates hematopoietic and small intestine system injury in mice

We then performed animal studies to verify whether ARA, identified as a potential radioprotective metabolite in our clinical studies, might have relevant effects in mice. The first goal of our study was to determine whether ARA supplementation could decrease the mortality rate associated with radiation (Fig. 4A) [14]. Body weight loss is an effect of radiation-related toxicity [15], and oral administration of ARA did not decrease the survival rate or body weight, suggesting that ARA caused no detectable harm to the mice (Fig. 4B, C). Following 6 Gy whole-body irradiation, the survival rate and body weight were much higher in the ARA treatment group than in the control (Con) vehicle group (Fig. 4B, C), validating that radiation-induced mortality and weight loss can be prevented by ARA replenishment.

Fig. 4

Oral gavage of ARA improves survival (6 Gy), decreases inflammatory response (4 Gy), and ameliorates hematopoietic and small intestine system injury (4 Gy) in mice. A Schematic of ARA treatment under 6 Gy irradiation. KM survival curves (B) and weight (C) for mice in 3 cohorts (n = 5 for ARA cohort; n = 7 for 6 Gy and 6 Gy + ARA cohorts, day 0 = the day of the irradiation). D Schematic of ARA treatment under 4 Gy irradiation. Photographs (E) and weight (F) of removed spleens from mice in 3 cohorts (n = 5 per cohort). Photographs (G) and weight (H) of removed thymuses from mice in 3 cohorts (n = 4 per cohort). I Spleens stained with H&E (× 200 magnification, congestion, yellow arrow). The content of IL-6 in plasma (J) and small intestine tissues (K) for mice in 3 cohorts assessed by ELISA (n = 5 per cohort). The IL-6 (L) and TNFα (M) levels in small intestine tissues for mice in 3 cohorts assessed by q-PCR (n = 5 per cohort). The content of MDA in plasma (N) and small intestine tissues (O) for mice in 3 cohorts assessed by ELISA (n = 5 per cohort). The Nrf2 (P), HO1 (Q), ZO1 (R), occludin (S), and claudin1 (T) levels in small intestine tissues for mice in 3 cohorts assessed by q-PCR (n = 5 per cohort). The small intestines stained with H&E (× 200 magnification, broken intestinal epithelium, black arrow), AB-PAS (× 200 magnification, goblet cells, black arrow) (U), and IHC (× 200 magnification, stained with antibodies, black arrow) (V). *p < 0.05, **p < 0.01; WP, white pulp

Since the hematopoietic system is particularly susceptible to radiation, we subsequently explored the radioprotective function of ARA in the hematopoietic system (Fig. 4D) [14]. With exposure to 4 Gy irradiation, the size and weight of the thymus and spleen of mice were significantly reduced but could be partially restored by ARA treatment (Fig. 4E–H). H&E pathological analyses implied that irradiation contributed to damage to the splenic white pulp (WP), but no tissue damage of the splenic WP area was showed in the ARA group (Fig. 4I). Analysis of routine blood test confirmed that white blood cell (WBC) counts, platelet (PLT) counts, and percentage of lymphocytes (LY%) were obviously decreased, but the percentage of neutrophil granulocytes (NE%) was increased in the irradiation group. The opposite results were observed after ARA replenishment (Fig. S3A-D).

In addition to the hematopoietic system, moderately high doses of radiation affect systemic inflammation and the gastrointestinal system [14]. Subsequently, we evaluated the protective effects of ARA on radiation-induced inflammation and gastrointestinal injury (Fig. 4D). Following ARA administration, mice exposed to 4 Gy irradiation exhibited reduced levels of inflammatory cytokines in the plasma and small intestine tissues (Fig. 4J–M), suggesting that ARA ameliorated radiation-induced systemic and intestinal inflammation. We next investigated whether ARA mediated radioprotection in mice by preventing irradiation-induced oxidative stress [16]. After ARA treatment, high oxidative stress levels induced by radiation were recovered, with decreased level of MDA (oxidative stress marker) and increased level of Nrf2 and HO1 (antioxidant markers) (Fig. 4N–Q). We further measured the expression of ZO1, occludin, and claudin1, which are regulators of epithelial proliferation and survival after toxic stimuli [17,18,19]. It was shown that the level of ZO1, occludin, and claudin1 increased after ARA treatment, indicating that ARA could promote intestinal regeneration and maintenance of the mucosal epithelial barrier (Fig. 4R–T), as verified by H&E (of intestinal epithelial barrier integrity), AB-PAS (for intestinal epithelial regenerative ability), and immunohistochemistry (IHC) (for intestinal epithelial tight junction proteins) staining in small intestine tissues (Fig. 4U, V).

ARA retains gut bacteria and metabolite composition influenced by irradiation

Based on the close association between intestinal microecology and radiation progression mentioned above, we assessed the therapeutic effects of ARA treatment on alterations in gut bacteria (metagenomics) and metabolites (non-targeted metabolomics) under 4 Gy irradiation in mice (Fig. 5A). The α-diversity analysis of the ACE and Chao1 indices showed that the abundance of gut microbiota was significantly decreased after irradiation, but this damage was partly attenuated by ARA replenishment (Fig. S4A at the genus level and Fig. 5B at the species level). We next compared the microbiota abundance distribution among the three cohorts and found that Lachnospiraceae was the most significant bacteria that showed low abundance in the irradiation group and recovered its abundance in the ARA-applying group (Fig. S4B, C on genus level; Fig. 5C, D on species level; statistical tests in Fig. S4D, E). For β-diversity, the sample distance among the three cohorts was analyzed and plotted using PCA, and Lachnospiraceae was defined as the major contributor to the subject distribution, indicating that ARA might change the intestinal flora composition after radiation by regulating the abundance of Lachnospiraceae (Fig. S4F at the genus level and Fig. 5E at the species level). We enrolled clinical factors in mice for redundancy analysis (RDA), and the results showed that both inflammation and hematopoietic function factors were essential for the sample distribution (Fig. S4G at the genus level, Fig. 5F on species level). In metabolomic analysis, PCA in both positive and negative ion modes showed that ARA treatment could robustly alter the metabolite profiles after irradiation (Fig. 5G, H). KEGG enrichment analysis showed that differential metabolites between irradiation and control groups were enriched in ARA metabolism (8,9-DiHETrE and 6-ketoprostaglandin E1 decreased in the irradiation group) pathway (Fig. 5I); differential metabolites between ARA therapy and irradiation groups were enriched in ARA metabolism (11,14,15-THETA and dinoprost increased in ARA therapy group) and LA metabolism (9,10,13-TriHOME, 13(S)-HpODE and 13-HODE increased in ARA therapy group) pathways (Fig. 5J), consistent with the metabolomic results analyzed in patients (Fig. 3E, F). The correlation heatmap also implied a highly positive connection between Lachnospiraceae and ARA metabolites (Fig. S4H at the genus level; Fig. 5K at the species level), showing that Lachnospiraceae as well as ARA and LA metabolism was the most predominant bacteria and metabolites affected by ARA treatment under 4 Gy irradiation.

Fig. 5

ARA retains gut bacteria and metabolite composition influenced by irradiation (4 Gy). A Schematic of fecal collection for bacteria and metabolite sequencing after ARA treatment under 4 Gy irradiation (n = 3 per cohort). B α-diversity analysis in Ace and Chao1 on species level in 3 cohorts. C Abundance distribution on species level (top 20) in 3 cohorts (s_Lachnospiraceae red marked). D Heatmap analysis on species level (top 30) in 3 cohorts (s_Lachnospiraceae red marked). E Principal components (PC) analysis (PCA) on species level in 3 cohorts (s_Lachnospiraceae red marked). F Redundancy analysis (RDA) and clinical factors on species level in 3 cohorts. G PCA analysis in positive ion mode in 3 cohorts. H PCA analysis in negative ion mode in 3 cohorts. KEGG Enrichment Analysis (I, 4 Gy vs control; J, 4 Gy + ARA vs 4 Gy; inhibited pathways green marked; activated pathways red marked). K Spearman correlation between seven metabolites enriched in ARA metabolism and LA metabolism and top 20 species (s_Lachnospiraceae red marked). *p < 0.05, **p < 0.01, ***p < 0.001

The gut microbiota impacts the radioprotective effects of ARA treatment in mice

Given the aforementioned radioprotective effects of ARA treatment, we established an ABX treatment model to investigate whether ARA-moderated radiation toxicity is dependent on the gut microbiota (Fig. 6A). In the quality of life analysis, ARA treatment no longer alleviated body weight loss induced by 4 Gy irradiation in ABX-challenged mice (Fig. 6B). Furthermore, we evaluated the radioprotective function of ARA in the hematopoietic system. The results showed that ARA treatment failed to improve the mass of the spleen or promote hematopoietic recovery without the participation of intestinal flora (Fig. 6C–E). As expected, the loss of gut microbes made it difficult to modulate oxidative stress and inflammation via ARA treatment after irradiation (Fig. 6F–M). Moreover, ABX treatment eliminated the radioprotective effects of ARA replenishment by enhancing the mucosal barrier and protecting the intestinal epithelium (Fig. 6N–R). These findings suggest that ARA protects against radiation damage, possibly through the gut microbiota.

Fig. 6

Gut microbiota impacts the radioprotective effects (4 Gy) of ARA treatment in mice. A Schematic of antibiotics (ABX) intervention before ARA treatment. B The weight for mice in 2 cohorts (n = 4 for 4 Gy + ARA cohort; n = 5 for ABX + 4 Gy + ARA cohort, day 0 = the day of the irradiation). Photographs (C) and weight (D) of removed spleens from mice in 4 cohorts (n = 5 per cohort). E Spleens stained with H&E (× 200 magnification, congestion, yellow arrow). The content of plasma MDA (F) and IL-6 (G); MDA (H) and IL-6 (I) in small intestine tissues for mice in 2 cohorts assessed by ELISA (n = 5 per cohort). The IL-6 (J), TNFα (K), Nrf2 (L), HO1 (M), ZO1 (N), occludin (O), and claudin1 (P) levels in small intestine tissues for mice in 4 cohorts assessed by q-PCR (n = 5 per cohort). Small intestine stained with H&E (× 200 magnification, broken intestinal epithelium, black arrow), AB-PAS (× 200 magnification, goblet cells, black arrow) (Q),and IHC (× 200 magnification, stained with antibodies, black arrow) (R). *p < 0.05, **p < 0.01, ***p < 0.001

ARA reconstructs the small intestinal protein expression profile and protects against radiation via Hmgcs1 in mice

To systematically assess the molecular mechanism of ARA treatment in radiation, we performed proteomic analysis of the small intestine tissues of mice from the Con and 4 Gy irradiation with or without ARA treatment groups (Con, 4 Gy, ARA + 4 Gy groups) to distinguish key regulatory proteins. Based on the screening criteria for differentially expressed proteins (DEPs) of p < 0.05 and fold change ≤ 0.6 or > 1.67, 64 significantly DEPs in the Con vs 4 Gy comparison and 86 significant DEPs in the ARA + 4 Gy vs 4 Gy comparison were identified (Fig. S5A, B; more details in Supplemental Excel2). According to the protein alterations identified by hierarchical cluster analysis, hydroxy-3-methylglutaryl-coenzyme A (HMG-CoA) synthase 1 (Hmgcs1) was the only protein downregulated in the 4 Gy group compared to the Con group, whereas it was upregulated in the ARA treatment group compared to the group treated with irradiation only (Fig. 7A, B), which was validated by qPCR (Fig. S5C), revealing that Hmgcs1 may serve as a target protein for ARA supplement. KEGG pathway enrichment analysis of DEGs revealed lower activation of ARA and LA metabolism in the irradiation-treated group than in the Con group (Fig. 7C), which was similar to the KEGG pathway results obtained through untargeted metabolomics (Figs. 3E, F and 5I). In the ARA + 4 Gy vs 4 Gy comparison, the PI3K–Akt signaling pathway and pathways in cancer were significantly stimulated by ARA treatment, suggesting that ARA might function as a radioprotectant in mice by promoting intestinal epithelial cell proliferation (Fig. 7D).

Fig. 7

ARA reconstructs the small intestinal protein expression profile and protects against radiation (4 Gy) via Hmgcs1 in mice. Hierarchical cluster analysis for the differential proteins (fold change ≤ 0.6 or > 1.67) in the small intestine of mice in three cohorts (Hmgcs1 red marked) (A, control vs 4 Gy; B, 4 Gy + ARA vs 4 Gy; n = 3 per cohort). KEGG pathway enrichment analysis (most inhibited green marked) of differentially expressed proteins in the three cohorts (C, 4 Gy vs control; D, 4 Gy + ARA vs 4 Gy). E Schematic representation of ursolic acid (UA) intervention with ARA treatment. Photographs (F) and weight (G) of the spleens were removed from mice in four cohorts (n = 5 per cohort). H Spleens and small intestine stained with H&E (× 200 magnification, congestion, yellow arrow; broken intestinal epithelium, black arrow) and AB-PAS (× 200 magnification, goblet cells, black arrow). IL-6 (I), TNFα (J), Nrf2 (K), HO1 (L), ZO1 (M), occludin (N), and claudin1 (O) levels in the small intestine tissues of mice in the four cohorts assessed by q-PCR (n = 5 per cohort). P Small intestine stained with IHC (× 200 magnification, stained with antibodies, black arrow). *p < 0.05, **p < 0.01

Based on the above analyses, we re-evaluated the radioprotective function of ARA treatment by inhibiting the expression of Hmgcs1 via UA (Fig. 7E, Fig. S5D) [20]. Interestingly, compared with only ARA treatment, the irradiated mice with Hmgcs1 silencing and ARA replenishment exhibited worse hematopoietic system and small intestine injuries, higher inflammation, and oxidative stress (Fig. 7F–P), suggesting that ARA provided radioprotection possibly dependent on the Hmgcs1 target.