Multi-Omics Reveal Additive Cytotoxicity Effects of Aflatoxin B1 and Aflatoxin M1 toward Intestinal NCM460 Cells

Aflatoxin B1 (AFB1) is a common crop contaminant, while aflatoxin M1 (AFM1) is implicated in milk safety. Humans are likely to be simultaneously exposed to AFB1 and AFM1; however, studies on the combined interactive effects of AFB1 and AFM1 are lacking. To fill this knowledge gap, transcriptomic, proteomic, and microRNA (miRNA)-sequencing approaches were used to investigate the toxic mechanisms underpinning combined AFB1 and AFM1 actions in vitro. Exposure to AFB1 (1.25–20 μM) and AFM1 (5–20 μM) for 48 h significantly decreased cell viability in the intestinal cell line, NCM460. Multi-omics analyses demonstrated that additive toxic effects were induced by combined AFB1 (2.5 μM) and AFM1 (2.5 μM) in NCM460 cells and were associated with p53 signaling pathway, a common pathway enriched by differentially expressed mRNAs/proteins/miRNAs. Specifically, based on p53 signaling, cross-omics showed that AFB1 and AFM1 reduced NCM460 cell viability via the hsa-miR-628-3p- and hsa-miR-217-5p-mediated regulation of cell surface death receptor (FAS), and also the hsa-miR-11-y-mediated regulation of cyclin dependent kinase 2 (CDK2). We provide new insights on biomarkers which reflect the cytotoxic effects of combined AFB1 and AFM1 toxicity.


Introduction
In the natural environment, different crops are likely to be contaminated by mycotoxins during growth and harvesting, especially during environmental stresses such as flooding and assault by insects [1]. Of the different mycotoxins, aflatoxins (AFs) are serious and harmful molecules. Among AFs, the secondary metabolite produced by Aspergillus, aflatoxin B1 (AFB1), has attracted considerable research interest due to its universality and high toxicity. Recent studies have shown that colon inflammation and inhibition of embryonic development were associated with the exposure of AFB1 [2,3]. The International

NCM460 Cytotoxicity Is Induced by AFB1 and AFM1
To determine the intestinal cytotoxicity induced by AFB1 and AFM1, NCM 460 cells were exposed to different AFB1 and AFM1 concentrations (1.25, 2.5, 5, 10, 15, and 20 µM) either individually or in combination for 48 h. AFB1 significantly decreased (p < 0.05) NCM460 cell viability at a low concentration (1.25 µM), while AFM1 caused no effects at the same concentration, indicating that AFB1-induced cytotoxicity was stronger than AFM1 alone ( Figure 1A). In addition, half-maximal inhibitory concentration (IC 50 ) of 10.47 ± 2.40, 8.10 ± 1.44, and 5.50 ± 1.21 µM, respectively, were determined for AFM1, AFB1, and AFM1 + AFB1 using CalcuSyn software. Therefore, the toxicity order toward NCM460 cells was AFM1 + AFB1 > AFB1 > AFM1. tive effect being the most fundamental type of interaction, we believe that to understand the interaction mechanism of mycotoxins, we must firstly analyze the underlying mechanism of addition Therefore, to investigate the mechanism of combined additive effects from toxins, 2.5 μM AFB1 and 2.5 μM AFM1 were used for omics analysis. Additionally, 2.5 μM was chosen as we were interested in knowing if AFM1, which had no significant effect on cell viability, affected AFB1 toxicity in an additive manner.

Transcriptomic Analysis of NCM460 Cells
Transcriptome analysis was performed in NCM460 cells exposed to 2.5 μM AFM1, 2.5 μM AFB1, and their combination for 48 h. Volcano plots showed that when compared with control group, the number of differentially expressed genes (DEGs) in AFM1, AFB1, and AFM1 + AFB1 treatments was 71 (67 up and 4 down), 578 (385 up and 193 down), and 2650 (1651 up and 999 down), respectively (Figure 2A-C). Combined toxins generated the most significant changes. A heatmap showed the AFM1 expression profile was similar to the control group, while the AFB1 profile was similar to AFM1 + AFB1 ( Figure 2D). This result was consistent with there being no DEGs between AFM1 + AFB1 and AFB1 ( Figure  S1). A Venn diagram showed that when compared with control group, 2083 unique genes were present in the AFM1 + AFB1 group ( Figure 2E).
Subsequently, Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analysis was performed. Compared with control group, GO terms were mainly enriched for cell proliferation, cell cycle processes, and cell cycle for individual and combined treatments ( Figure 3A). Compared with control group, KEGG analysis enriched by the DEGs in AFM1, AFB1, and AFM1 + AFB1 were shown ( Figure S2), which demonstrated that cell cycle and p53 signaling pathway were the most common enrichment pathways in all groups. For KEGG analysis of 2083 unique genes in AFM1 + AFB1 treatment, genes were enriched in focal adhesion, DNA replication, p53 signaling pathway, and the extracellular matrix (ECM) ( Figure 3B). A combination index (CI) curve of combined AFM1 and AFB1 is shown ( Figure 1B). Straight lines in graphs (CI = 1) indicate an additive toxicological effect, with points above (CI > 1) and below (CI < 1) the straight lines indicating antagonistic and synergistic effects, respectively. Exposure to AFM1 + AFB1 in NCM460 cells for 48 h showed an additive effect (CI: 1.06 ± 0.31) at 2.5 µM, a slightly synergistic effect (CI: 0.82 ± 0.15) at 5 µM, and an antagonistic effect (CI: 1.22-2.22) at other concentrations ( Figure 1B). Due to the additive effect being the most fundamental type of interaction, we believe that to understand the interaction mechanism of mycotoxins, we must firstly analyze the underlying mechanism of addition Therefore, to investigate the mechanism of combined additive effects from toxins, 2.5 µM AFB1 and 2.5 µM AFM1 were used for omics analysis. Additionally, 2.5 µM was chosen as we were interested in knowing if AFM1, which had no significant effect on cell viability, affected AFB1 toxicity in an additive manner.

Transcriptomic Analysis of NCM460 Cells
Transcriptome analysis was performed in NCM460 cells exposed to 2.5 µM AFM1, 2.5 µM AFB1, and their combination for 48 h. Volcano plots showed that when compared with control group, the number of differentially expressed genes (DEGs) in AFM1, AFB1, and AFM1 + AFB1 treatments was 71 (67 up and 4 down), 578 (385 up and 193 down), and 2650 (1651 up and 999 down), respectively (Figure 2A-C). Combined toxins generated the most significant changes. A heatmap showed the AFM1 expression profile was similar to the control group, while the AFB1 profile was similar to AFM1 + AFB1 ( Figure 2D). This result was consistent with there being no DEGs between AFM1 + AFB1 and AFB1 ( Figure S1). A Venn diagram showed that when compared with control group, 2083 unique genes were present in the AFM1 + AFB1 group ( Figure 2E).
Subsequently, Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analysis was performed. Compared with control group, GO terms were mainly enriched for cell proliferation, cell cycle processes, and cell cycle for individual and combined treatments ( Figure 3A). Compared with control group, KEGG analysis enriched by the DEGs in AFM1, AFB1, and AFM1 + AFB1 were shown ( Figure S2), which demonstrated that cell cycle and p53 signaling pathway were the most common enrichment pathways in all groups. For KEGG analysis of 2083 unique genes in AFM1 + AFB1 treatment, genes were enriched in focal adhesion, DNA replication, p53 signaling pathway, and the extracellular matrix (ECM) ( Figure 3B). the intestinal barrier in AFM1 + AFB1 were selected for qRT-PCR. These genes included ECM-receptor interaction (ITGA3, LAMA3, ITGB5, COL9A3, LAMC1, and ITGA10), focal adhesion (FLT1, CCND1, CCND3, SHC2, and ITGA7), and p53 signaling pathway (TP73, SERPINE1, TP53I3, TP53, SESN3, and SFN). Compared with control group, the expression data for these 17 genes in AFM1 + AFB1 treatment were consistent with RNA-sequencing data ( Figure 3C).
Furthermore, KEGG analysis showed that these 98 unique DEPs were mainly enriched in focal adhesion, ECM receptor interaction, and p53 signaling pathway ( Figure  4C). DEPs enrichment in AFM1/CTL, AFB1/CTL, and AFM1 + AFB1/CTL groups showed that p53 signaling pathway and ribosome biogenesis in eukaryotes were key pathways in AFB1-treated cells, while p53 signaling pathway, focal adhesion, and ribosome biogenesis in eukaryotes had vital roles in AFM1 + AFB1-induced cytotoxicity (Table S1). Combining KEGG pathways, p53 signaling pathway had important roles in the intestinal toxicology induced by AFM1 and AFB1.

Cross-Omics Analysis of the Transcriptome and Proteome Induced by AFB1 and AFM1
To narrow down potential targets, a cross-omics analysis of transcriptome and proteome data was conducted. Scatter plot showed the distribution of corresponding proportions of transcripts to proteins ( Figure 5). As indicated, most transcript to protein ratios were focused on quadrant e, where both genes and proteins were not differentially expressed. Genes differential expression patterns were consistent with corresponding proteins in quadrant c (unanimous up-regulation) and g (unanimous down-regulation). We observed 5 (3 up and 2 down) proteins associated with DEGs in the AFM1 treatment (Figure 5A). In the AFB1 treatment, 45 (36 up and 9 down) proteins were identified in quadrants c and g ( Figure 5B). In the combined treatment, 66 (56 up and 10 down) proteins were associated with DEGs in the transcriptome ( Figure 5C). Venn diagram results showed that when compared with the control group, 29 unique proteins were present in the AFM1 + AFB1 group ( Figure 5D).
A protein-protein interaction (PPI) network, which included cell viability regulation, was constructed using DEGs and DEPs identified from cross-omics analysis in the AFM1 Furthermore, KEGG analysis showed that these 98 unique DEPs were mainly enriched in focal adhesion, ECM receptor interaction, and p53 signaling pathway ( Figure 4C). DEPs enrichment in AFM1/CTL, AFB1/CTL, and AFM1 + AFB1/CTL groups showed that p53 signaling pathway and ribosome biogenesis in eukaryotes were key pathways in AFB1treated cells, while p53 signaling pathway, focal adhesion, and ribosome biogenesis in eukaryotes had vital roles in AFM1 + AFB1-induced cytotoxicity (Table S1). Combining KEGG pathways, p53 signaling pathway had important roles in the intestinal toxicology induced by AFM1 and AFB1.
To confirm our proteomic results, paxillin (PXN) production, which was related to focal adhesion and the ECM pathway, was examined by western blotting. When compared with the control group, AFM1 + AFB1 significantly (p < 0.05) increased PXN production. Importantly, western blotting data were consistent with proteome data ( Figure 4D).

Cross-Omics Analysis of the Transcriptome and Proteome Induced by AFB1 and AFM1
To narrow down potential targets, a cross-omics analysis of transcriptome and proteome data was conducted. Scatter plot showed the distribution of corresponding proportions of transcripts to proteins ( Figure 5). As indicated, most transcript to protein ratios were focused on quadrant e, where both genes and proteins were not differentially expressed. Genes differential expression patterns were consistent with corresponding proteins in quadrant c (unanimous up-regulation) and g (unanimous down-regulation). We observed 5 (3 up and 2 down) proteins associated with DEGs in the AFM1 treatment ( Figure 5A). In the AFB1 treatment, 45 (36 up and 9 down) proteins were identified in quadrants c and g ( Figure 5B). In the combined treatment, 66 (56 up and 10 down) proteins were associated with DEGs in the transcriptome ( Figure 5C). Venn diagram results showed that when compared with the control group, 29 unique proteins were present in the AFM1 + AFB1 group ( Figure 5D).

Discussion
AFs are fungal metabolites in food and feed, with humans likely to be simultaneously exposed to both AFB1 and AFM1. In recent years, the adverse effects of AFB1 and AFM1 on the intestinal barrier have attracted considerable research attention. However, the underlying mechanisms of interactions between toxins are lacking. Therefore, we used multi-omics (transcriptome, proteome, and miRNA-sequencing approaches) to clarify such mechanisms in NCM460 cells-an intestinal epithelial cell line.
AFM1 and AFB1 decreased NCM460 cell viability, with AFB1 toxicity being higher than AFM1 (Figure 1). These results were consistent with our previous study showing that AFB1 cytotoxicity was higher than AFM1 in differentiated and undifferentiated Caco-2 cells [19]. Furthermore, we clarified the interactive effects of AFB1 and AFM1 on cell viability via isobologram analysis (Figure 1), and the interaction of toxins including AFM1, OTA, and zearalenone (ZEN) at different concentrations on cell viability of intestine in vitro were also revealed in our previous study [20]. As the study reported, the type of interaction depends on the concentrations of mycotoxins [21].
From transcriptome, proteome, and miRNA sequencing results, DEGs, DEPs, and DEmiRNAs in combined mycotoxin treatment were higher than in individual mycotoxin treatments (Figures 2A, 4A and 6A). Thus, AFM1 + AFB1 induced more severe intestinal toxicity, via additive effects. Additionally, KEGG pathway analysis of DEGs, DEPs, and DEmiRNAs from AFM1 and AFB1 treatments showed that these toxins affected NCM460 cell viability mainly through p53 signaling pathway. Previously, it was reported that cell viability reduction, induced by various chemical compounds, was directly correlated with p53 induction, and the induction peak appeared at the chemical concentration that caused cell viability to be lower than 80% [22]. In our study, a peak of p53 induction may have occurred due to the fact that 2.5 µM AFB1 and AFM1 maintained the cell viability below 80%. P53 regulates different signaling pathways to prevent cell damage or deterioration. Various cellular stresses could activate p53 signaling pathway [23], which then undergoes post-translational modifications [24] and trans-activates several genes required for cell growth inhibition. P53 activation also results in cell cycle arrest and is associated with the down regulation of cell cycle-related genes/proteins, including CDKN1A, CDK1, and CDC25 [25]. P53 pathway activation also has important roles in the transcriptional regulation of networks responding to cell apoptosis and cell proliferation [26]. The pathway generally responds to mycotoxins, including AFs and OTA. Previously, it was reported that AFB1 caused a G-to-T transversion mutation, which was related to a high frequency of p53 mutation, therefore, the majority of AFB1-associated hepatocellular cancer cases were shown to carry TP53 mutant DNA [27]. P53/p21-mediated cell cycle control was involved in OTA-induced carcinogenic effects [28].
We also screened for key regulators involved in reduced cell viability induced by AFB1 and AFM1. A cross-omics approach using transcriptomics and proteomics was performed and 43 key proteins were selected ( Figure 5). Among these regulators, CDK1, CDK2, CDKN1A, DDB2, RRM2, TP53I3, PCNA, and FAS were related to p53 signaling pathway. CDKs are vital cell cycle regulators [29]. AFB1 blocks the cell cycle in S phase by activating p53 [30]. CCND1 overexpression imbalances CDK activity and induces out-ofcontrol cell growth [31]. A previous study showed that AFM1 induced intestinal toxicity by cell cycle arrest in Caco-2 cells via changes in CDK1, SOS1/AKT, and AMPK signaling molecules [32]. ZEN arrested cell cycle progression at the G2/M phase by altering CDK1, CCNB1, CDC25A, and CDC25C expression in porcine granulosa cells [33]. CDK inhibitor p21 (CDKN1A) was involved in regulating the cell process, with p21 knockdown leading to the decrease in cell survival [34]. CDC2/CCNB1 complex inhibition was also involved in citrinin-induced G2/M phase arrest in HEK293 cells [35]. Additionally, in this study, the expression of pro-arrest CDKN1A was increased after AFB1 and AFM1 treatment, consistent with previous data [36]. Additionally, RRM2 was reported as closely linked with cell survival [37,38].
Growing evidence suggests that miRNAs fine-tune gene expression based on different external stimuli, such as mycotoxins [39]. MiRNAs appear to simultaneously interact with various mRNAs, which are involved in key signaling pathways, such as cell processing [40]. For clarification, we jointly analyzed DEmiRNAs with DEGs in combination with DEPs and identified 15 key DEmiRNAs implicated in the regulation of NCM460 cell viability, including hsa-miR-628-3p, hsa-miR-217-5p, and hsa-miR-184 ( Figure 6C). These miRNAs have been reported to perform different functions. A new finding suggested that plasma miR-628-3p level may represent a marker to predict the presence of severe atopic keratoconjunctivitis in atopic dermatitis patients [41]. Although miR-628-3p is not yet implicated in intestinal function, we hypothesize it could regulate CDK1 expression. MiR-217-5p could induce apoptosis in colorectal cancer cells by regulating multiple target genes, including PRKCI, BAG3, ITGAV, and MAPK1 in the ERK-MAPK signaling pathway [42]. Similarly, miRNA-217-5p may function as a therapeutic intervention against inflammatory bowel disease [43]. MiR-548u was identified in stored human leukocyte-depleted platelets [44]. In addition, cell function assay and western blotting showed that miR-184 was critical for hepatocellular carcinoma cell proliferation, migration, apoptosis, and autophagy [45].
In conclusion, for the first time, we showed that AFM1 and AFB1 exerted additive effects toward NCM460 cell viability via p53 signaling pathway, which appeared to induce changes in downstream regulators, including apoptosis, cell cycle arrest, and DNA repair genes/proteins. Based on p53 signaling pathway, we co-analyzed DEmiRNAs, DEmRNAs, and DEPs, and found that AFM1 + AFB1 reduced NCM460 cell viability via hsa-miR-628-3p and hsa-miR-217-5p FAS regulation, and CDK2 regulation via hsa-miR-11-y (Figure 7). These findings provide a theoretical molecular basis for future mycotoxin risk assessments. Further research is required to comprehensively explore the additive effects of these mycotoxins in the human food chain, and to accurately reflect the potential risk to human intestinal function.

Cell Culture and Treatment
NCM460 cells from the American Type Culture Collection (Manassas, VA, USA) were cultured in RPMI 1640 medium supplemented with 4.5 g/L glucose, antibiotics, 1% NEAA, and 10% FBS. Cells were cultured at 37 • C in 5% CO 2 . When the confluence reached approximately 90%, cells were digested with trypsin and then inoculated into 96 well plate with 6 × 10 3 cells/well for 24 h. Cells were then treated with different AFB1 and AFM1 concentrations (1.25, 2.5, 5, 10, 15, and 20 µM) either individually or in combination for 48 h.

Cytotoxicity Assay
Cell viability of NCM460 cells was examined using MTT assay. After 48 h of individual and combined AFB1 and AFM1 treatments (1.25, 2.5, 5, 10, 15, and 20 µM) in 96-well plates, supernatants were discarded and 0.5 mg/mL 100 µL MTT solution added to wells and plates incubated for 4 h. This solution was removed and 100 µL dimethyl sulfoxide added to dissolve methylazan crystals. The plate was gently shaken for 10 min to completely dissolve crystals, after which absorbance was recorded on a spectrophotometer (Thermo Company, Waltham, MA, USA) at 570 nm and 630 nm. The cell viability was calculated as follows: Cell viability (%) = (treatment group at 570 nm − treatment group at 630 nm)/(control group at 570 nm − control group at 630 nm) × 100

Isobologram Analysis
CI values were calculated using isobologram analysis according to a formula in our previous study [20]. The interaction type between mixed mycotoxins was quantitatively analyzed by the CI index; CI = 1, <1, and >1 indicated additivity, synergy, and antagonism actions, respectively. Isobologram analysis was performed using CalcuSyn software (Biosoft, Cambridge, UK).

RNA Sequencing and Validation
The transcriptome of RNA sequencing was conducted according to our previous study [13]. In the current study, DEGs were identified as genes with a fold change (FC) ≥2 and false discovery rate <0.05. KEGG pathways, enriched by these DEGs, were investigated. Biological pathways from GO and KEGG analyses were analyzed in a human genome background. To reduce the sample variability, three replicates of harvested NCM460 cells were pooled into one sample. A total of three samples were prepared from each treatment.
To validate RNA-sequencing reliability, qRT-PCR was performed. The conditions were previously described [14]. Primer sequences are shown (Table S3). The 2 −∆∆Ct method was used to calculate relative gene expression, with GAPDH used as a reference gene.

Proteome Analysis and Validation
Proteomic analysis was conducted as previously reported [13]. Cell supernatants were enzymatically hydrolyzed using reagents from the iTRAQ kit. 114 was labeled as the CTL group, 115 as the AFM1 group, 116 as the AFB1 group, and 117 as the AFB1 + AFM1 (M1-B1) group. Proteins with p ≤ 0.05 and FC > 1.2 values were considered to be DEPs. The KEGG analysis enriched by DEPs also conducted as transcriptome. To reduce the sample variability, three replicates of harvested NCM460 cells were pooled into one sample. Each treatment was run in three replicates.
DEPs were validated by western blotting. Lysed NCM460 cell proteins were blocked and incubated with a PXN primary and secondary antibody. Band densities were analyzed in Image J2 software and normalized to human β-actin.

miRNA Sequencing
MiRNA sequencing was conducted as previously described [46]. After deriving total RNA, rRNA, scRNA, snRNA, snoRNA, and tRNA were identified and removed using GeneBank (Release 209.0) and Rfam databases (Release 11.0). The criteria for identifying DEmiRNAs were p < 0.05 and FC ≥ 2 values. Target mRNAs for DEmiRNAs were predicted by miRanda (Version 3.3a) and TargetScan (Version 7.0) with Spearman Rank Correlation Coefficient (SRCC) ≤ −0.5. The relationship between miRNA-mRNA was illustrated by Sankey plot. KEGG analyses were also performed to functionally annotate target mRNAs. To reduce the miRNA-sequencing sample variability, three replicates of harvested NCM460 cells were pooled into one sample. A total of three samples were prepared from each treatment.

Statistical Analysis
Cell viability data were analyzed in GraphPad Prism 8.0 (San Diego, CA, USA), and one-way analysis of variance and t-test methods were used for statistical analyses. A p < 0.05 value was regarded as statistically significant.