EBV infection of tree shrews
Six shrews were infected with 1 × 108 copies of EBV isolated from the B95-8 strain (NCBI: txid 10,377) via the femoral vein (Fig. 1A). qPCR was used to determine viral loads in the blood, plasma, and throat swabs. For blood and plasma, the peak level appeared on 3dpi(compared to 0 dpi, 7 dpi or 14 dpi, p < 0.05), while the peak level of throat swabs appeared on 7 dpi(compared to 0 dpi, 3 dpi or 14 dpi, p < 0.05)(Fig. 1C, Supplementary Table 7). Although viral loads decreased markedly on 14dpi, the positive signals of EBV-encoded protein1 (LMP1) and EBV-encoded small RNAs(EBERs) were detected in spleen tissues which implied a latent infection being established(Fig. 1B). Inflammation is typically induced in response to a microbial infection. We observed serum levels of the inflammatory marker C-reactive protein (CRP), interleukin (IL)-6 and Tumor Necrosis Factor (TNF)-α. Serum CRP and TNF-α levels peaked on 7 dpi(compared to 0 dpi or 14 dpi, p < 0.05), while serum IL-6 levels did not show an increase trend (Fig. 1D, Supplementary Table 7). Altogether, our results illustrate that tree shrews are susceptible to EBV experimental infection and concomitant inflammatory factor changes in the early phase of virus replication.
EBV infection alters the structure of the fecal microflora in tree shrews
The influence of EBV on the structure of the gut microflora was explored. Stool specimens were acquired before and during infection (six duplicate specimens at each time point), and 16 S rRNA sequencing was used to explore the microflora structure. Overall, 1,568,536 sequence reads (mean = 65,356; min = 53,997; max = 79,157; Supplementary Table 1) were obtained. There were no remarkable alterations in alpha diversity with respect to the number of reported species or Shannon indices over the course of infection (Supplementary Fig. 1A and 1B). Beta diversity analysis based on the Bray–Curtis index showed clustering per species and inter-individual variability (Supplementary Fig. 1C). Nonetheless, assessment of the Bray–Curtis distance on day 0 relative to the other time points exhibited a drift in the fecal microflora after infection from 3 dpi and a peak at 7 dpi (Fig. 2A). Notably, β diversity returned to basal levels at 14 dpi post-infection. These data illustrate that EBV infection triggers changes in gut microflora structure.
Alterations in the structure of bacterial gut microbiota over time. (A) Beta diversity was determined. Bray Curtis distance between specified time point and Day 0. The overall structure of the bacterial microflora (B) at the Phylum and Genus (C) levels was assessed for each time point during EBV infection. (D) A LEfSe assessment exhibits that diverse bacterial taxa changed over the EBV infection. Only taxa with a remarkably significant LDA score (log10) > 1.5 are shown. *p < 0.05, **p < 0.01
At the phylum level, Firmicutes were the most enriched bacteria, followed by Bacteroidetes, Proteobacteria, and Actinobacteria in the fecal microflora of tree shrews. At genus level, several bacteria(Streptococcus, Megamonas, Bifidobacterium) with the highest the relative abundance stood out (Fig. 2B). Visualization of variation in phylum level and genus level relative abundance across sampling time points can be found in Fig. 2C. It is clear that the composition of gut microbiota structure significantly altered during EBV infection. At the same time, the alterations in bacterial taxon richness during infection was further supported by linear discriminant analysis effect size(LEfSe) analysis. Some fecal microflora alterations remained present at 14 dpi, even though they varied from those reported at earlier time points. In particular, the abundance of several Syntrophococcus genus members (Firmicutes phylum) was higher at earlier time points. Even though a remarkable number of changes were reported at 7 dpi and resolved at later time points, other alterations were reported post-EBV infection at 14 dpi. Altogether, EBV infection in tree shrews was linked to alterations in the composition of the gut microflora, peaking at 7 dpi.
EBV infection influences the fecal microflora cross talk network
Gut microflora constitutes an ecosystem of cross-talking microorganisms rather than a simple group of individual microbes. Networks built on the correlation of microbial richness can be employed to assess the complex crosstalk of microbial communities, entailing possible interdependence or competition among taxa [34]. A global exploration of the established network yields remarkable information regarding the ecosystem, with a greater number of nodes and connections, illustrating a remarkable crosstalk at the community level. Therefore, we created bacterial richness correlation networks to explore the structure of the microbial ecosystem during EBV infection. To this end, we provided networks for four time points (Fig. 3A). Dense, inter-linked networks were established on day 0 and 3 dpi. In contrast, the network at 7 and 14 dpi was greatly atrophied, harboring few nodes and connections. These data were verified by a quantitative assessment of the degree of connectivity of the four networks (Fig. 3B). Taken together, these data illustrate that EBV infection triggers ecological alterations in the gut microflora.
Correlations between the changes in the fecal microflora structure and infection-linked variables. (A) Correlation networks at the genus level for bacterial richness were created via Spearman’s correlation approach for four time periods. Each circle (node) designates a genus. (B) Violin plots show the degree of connectivity for the four time periods indicated. (C) The heat map shows the correlations between the representation of the various bacterial taxa and infection-linked variables. *p < 0.05, **p < 0.01
Fecal microflora taxa are linked to the features of EBV infection
Blood inflammatory factors are linked to the severity of virus infection [35, 36] and alterations in the relative richness of operational taxonomic units [37, 38]. Hence, we sought to explore whether alterations in the gut microflora structure are linked to EBV infection parameters, including viral load and inflammatory markers (Fig. 3C). Intriguingly, the relative richness of Butyricicoccus genus members positively linked to the presence of EBV in the blood and plasma; opposite correlation was observed for the genera Variovorax and Paramuribaculum. Only one taxon from the genus Syntrophococcus negatively correlated with throat EBV. Moreover, Syntrophococcus and Peptoclostridium were positively linked to TNF-α and CRP levels, whereas an inverse relationship was observed with Paramuribaculum. Additionally, a negative correlation was detected between Variovorax and IL6.
EBV infection alters the fecal metabolic profile
To explore the functional influences of infection-linked alterations in fecal microflora structure, we used a non-targeted metabolomics approach to detect metabolites at four time points. To date, no study has examined fecal metabolites following EBV infection. In the present study, 1684 secondary metabolites were identified in the fecal extracts of tree shrews, mainly compounds of the benzenoid class, lipids along with lipid-like molecules class, organic acids coupled with derivative class, organoheterocyclic compound class, phenylpropanoids, and polyketide class (Fig. 4A). Pairwise comparisons of the metabolomic changes in response to EBV infection by OPLS-DA are shown in Supplementary Fig. 3. The 3 dpi, 7 dpi, and 14 dpi clusters could be clearly separated from Day0 in the OPLS-DA model. This suggests that the fecal metabolome profile changed over the course of the infection. As shown in Fig. 4B, compared to Day0 after 7 dpi, EBV infection resulted in the most metabolites that were significantly affected, followed by 3 and 14 dpi. Specifically, there were 67, 116, and 44 differentially expressed metabolites(VIP > 1.5 and p < 0.05) at 3, 7, and 14 dpi, respectively. Most of the significantly differentially expressed metabolites can be classified into lipids, lipid-like molecules, organic acids, and derivatives. The specific metabolites at 3, 7, and 14 dpi are noted in Fig. 4C, D and E; the left panel shows the significant differential metabolites, while the right one shows the extracted ion chromatogram of the differential metabolites. These variations in the fecal metabolome profile might reflect alterations in their generation by bacteria and/or utilization by host cells.
Fecal metabolite generation is altered during an EBV infection. (A) Differential secondary metabolites statistics at different classes. Differential metabolites were defined by VIP > 1.5 & p-value < 0.05 compared with the Day 0. (B) Differential secondary metabolites statistics at different time points. The figure only shows classes which at least one differential individual per time point. The specific metabolites of the 3 dpi (C), 7 dpi (D) and 14 dpi (E) are noted, the left panel is the significant differential metabolites, and the right panel is the extracted ion chromatogram of the differential metabolites
The biological pathways affected at different time points were determined based on differentially expressed metabolites and genes. As shown in Fig. 5A, metabolic pathways were significantly enriched at all three time points, while the impact values of pathways decreased over time, until less than 0.1 at 14 dpi. Moreover, the citrate cycle (TCA cycle) and porphyrin and chlorophyll metabolism pathways were affected at both 3 and 7 dpi, indicating that they may serve as possible target pathways for EBV infection.
Correlations between the alterations in the fecal metabolites profile and infection-related variables. (A) The enrichment results of the KEGG pathway from the KEGG database. (B) Relationship of differential fecal metabolite levels with plasma cytokine levels and the representation of the diverse bacterial taxa. *p < 0.05, **p < 0.01. (C) The number of metabolites with significant bacterial taxa correlation
Fecal metabolites are linked to the fecal microflora taxa
The correlation between metabolites and infectious parameters indicated that differential metabolites were linked to some infectious parameters (Fig. 5B). Specifically, linoleic acid concentration in stool samples was positively linked to the presence of EBV in the blood and plasma. In contrast, tyramine concentration positively correlated with throat EBV load. The concentrations of estrone glucuronide, linoleic acid, protoporphyrin IX and tyramine negatively correlated with TNF-α levels. Moreover, geranic acid, tyramine, undecylenic acid, and estrone glucuronide positively correlated with CRP levels, while the opposite correlation was observed with hippuric acid and guanidinobutanoic acid. We subsequently analyzed the relationship between gut microbiota and metabolites. The microflora that affected most metabolites were the genera Blautia, Akkermansia, Acidithiobacillus, Lachnospiraceae_NK4A136_group, Peptoclostridium, Sulfobacillus, and Anaerostipes, especially Blautia (Fig. 5B and C). Interestingly, protoporphyrin IX, enriched in the porphyrin and chlorophyll metabolism pathway, was negatively related to multiple gut microbiota, including the genera Butyricicoccus and Anaerostipes. Baicalin concentration in stool samples positively correlated with the levels of Akkermansia and LachnospiraceaeNK4A136group (r > 0.7) and moderately correlated with the levels of Variovorax and Endozoicomonas. Altogether, these data illustrate those alterations in gut microflora structure during EBV infection are linked to changes in the fecal metabolic profile.