Transcriptome-wide reprogramming of N⁶-methyladenosine modification by the mouse microbiome

Dear Editor,

Microbiome affects many aspects of human health and disease and elicits a wide range of host responses including remarkable epigenetic changes such as DNA methylation, histone modification and non-coding RNA expression.¹ A still poorly explored area of microbiome-host interaction is the response of host RNA modification. N⁶-methyladenosine (m⁶A) is the most abundant mRNA modification in mammalian cells, occurring at ~3 modified adenosine residues per transcript. The m⁶A mapping and biology have been extensively studied recently.² At the physiological level, m⁶A affects embryonic development, circadian clock, immuno-response, and others. At the cellular and molecular level, m⁶A affects all key aspects of mRNA processing, translation and decay. Importantly, m⁶A is a predominant, transcriptome-wide mark that is responsive to environmental changes; this dynamic m⁶A pattern is maintained by the writer enzyme complex containing the METTL3 and METTL14 proteins, and two eraser enzymes of FTO and ALKBH5.^3,4

We investigated the host response marked by m⁶A in the transcriptome to the presence of microbiome in mice (Fig. 1a). We employed one group of germ-free (GF) mice to identify the host response to the absence, and the other group of specific pathogen-free (SPF) mice to identify the host response to the presence of microbiome. We validated the absence of gut microbiota in our GF mice by PCR of the representative 16S genes (Supplementary information, Fig. S1a). 16S rRNA gene amplicon sequencing of the SPF mice showed that all three mice in this group had similar bacterial compositions at the genus level, which were mainly blautia and roseburia (Supplementary information, Fig. S1b).

Fig. 1.

m⁶A methylome and writer/eraser expression in the germ-free (GF) and specific pathogen-free (SPF) mouse tissues. a Schematic representation of the study. b QQQ LC/MS measurement of total m⁶A/A ratio of polyA-selected and ribo-minus treated RNAs. Values are the means ± standard deviation (SD), n = 3, *P < 0.05, Student’s t-test. c m⁶A pattern distribution across the mRNA regions in brain, intestine and liver. m⁶A peaks were mapped back to the corresponding gene, and assigned as originated from 5′ UTR, coding region (CDS) or 3′ UTR. d Motif analysis of m⁶A peaks. Upper panel, GF tissues; lower panel, SPF tissues. e Venn diagram showing the differences of m⁶A peaks between GF and SPF samples. f Principal component analysis of input (IN) and IP samples. The label is for Sample_tissue_Seq, e.g., GF_B_IP stands for GF mouse, brain, m⁶A-IP. Tissue labels are: B, brain; I, intestine; L, liver. g Representative sequencing coverage of an mRNA in the brain showing a differential m⁶A peak in GF and SPF samples. h Transcript counts containing different m⁶A peak numbers in the brain. i m⁶A peak and exon density in the brain. j Abundance of m⁶A-containing transcripts in the brain. k mRNA m⁶A peak positions in the brain. l Reactome analysis of biological pathways of m⁶A-containing transcripts in the brain. m Venn diagram comparing the 4-week-old GF/SPF brain m⁶A peak-containing transcripts with those in the E13.5 embryonic brain. n Western blots of m⁶A writer proteins METTL3, METTL14, and eraser proteins FTO, ALKBH5 in the brain tissues. o Quantitation of m⁶A writer and eraser protein levels in the brain. Values are the means ± SD, n = 3, *P < 0.05, **P < 0.01, ***P < 0.001, Student’s t-test. p Quantitation of m⁶A writer and eraser protein levels in the intestine and liver. Values are the means ± SD, n = 3, *P < 0.05, Student’s t-test

We harvested three tissues of GF and SPF mice of the same genetic background at 4 weeks of age, brain, intestine, and liver, and performed m⁶A analysis in polyA-selected RNA by liquid chromatography/mass spectrometry (LC/MS) to determine the total m⁶A/A ratios and by the m⁶A-MeRIP sequencing to determine the transcriptomic m⁶A pattern and distribution. These three tissues were selected based on their pervasive studies in the literature on the GF and SPF mouse physiology. The m⁶A/A ratios of the polyA-selected RNA are in the expected range of 0.2%–0.6%; brain showed the highest m⁶A content for both GF and SPF mice, and brain and intestine showed higher m⁶A content in the GF mice (Fig. 1b). The polyA-selected RNA in kidney also showed higher m⁶A content in the GF mice (Supplementary information, Fig. S2a). The higher m⁶A content in the brain tissue was also observed in GF and SPF mice that were 10 weeks old (Supplementary information, Fig. S2b) and even 2 years old (Supplementary information, Fig. S2c). Our m⁶A-MeRIP results of all three tissues (Supplementary information, Table S1) showed the well-known m⁶A pattern across the mRNA transcripts such as the strong enrichment of m⁶A peaks at the junction of coding region (CDS) and 3′ UTR (Fig. 1c). We identified the m⁶A-containing transcripts that were present in all three GF or SPF mouse groups as “high confidence” data and used only these for further analysis (Supplementary information, Fig. S3). We recovered the known m⁶A installation consensus sequence, RRACH (R = A/G, H = A/C/U) among the m⁶A peaks with a preference of guanosine 5′ to the m⁶A site (Fig. 1d). We validated our sequencing results by quantitative RT-PCR of specific transcripts (Supplementary information, Fig. S4a). Our sequencing result was also consistent with the expected mRNA expression difference of GF versus SPF mouse reported in the literature^5,6 (Supplementary information, Fig. S4b). These results validated the high-quality nature of our m⁶A-MeRIP data.

We identified several differences in m⁶A patterns between the GF and SPF tissues. First, the m⁶A peak distributions had distinct shapes among these tissues (Fig. 1c). When benchmarked against the m⁶A cluster near the stop codon, the GF brain m⁶A occurrence was higher in the CDS region compared to the SPF brain. Second, the m⁶A installation consensus sequences in GF and SPF tissues were deviated in brain, but identical in intestine and liver (Fig. 1d). Third, only 25% of the m⁶A peaks in the GF brain overlapped with those in SPF brain, whereas > 59% of the m⁶A peaks overlapped in GF and SPF intestine and liver (Fig. 1e). The large brain m⁶A peak differences was also shown by principal component analysis and their statistical significance in gene ontology analysis across all tissues (Fig. 1f and Supplementary information, Fig. S5), and was not derived from global transcript expression differences (Supplementary information, Fig. S6). On the other hand, SPF brain may have higher m⁶A modification fraction for some common m⁶A peaks (Supplementary information, Fig. S7). All together, these results indicate that the presence of microbiome has a profound influence on the cellular mRNA m⁶A patterns in a tissue-dependent manner. Alteration of the m⁶A pattern is most pronounced in the brain, where the m⁶A methylome is substantially reduced in the presence of microbiome.

We performed in-depth analysis for the GF and SPF brain tissues to further elucidate the m⁶A alteration in the mRNA (a representative read coverage plot shown in Fig. 1g). Among the 8643 and 2750 m⁶A-containing transcripts, 67 and 80% had only one m⁶A peak in the GF and SPF brains, respectively, and the GF/SPF transcript ratio was 2.6 (Fig. 1h). However, this GF/SPF ratio for the transcripts containing two or more m⁶A peaks steadily increased to 4.4 for two, 6.7 for three, and 9.2 for more than three m⁶A peaks. GF brain transcripts also had a broader distribution of m⁶A peak/exon ratios (Fig. 1i). These results indicate that more m⁶A clusters are present in individual GF than SPF brain transcripts. The abundance of the m⁶A-containing transcripts was lower in GF than SPF brain (Fig. 1j), which might be associated with a major known role of m⁶A in accelerating mRNA decay.⁷ More m⁶A peaks were present in all three mRNA regions in GF than SPF brain (Fig. 1k). The m⁶A location in different mRNA regions has been associated with different functions. For example, m⁶A in the 5′ UTR enhances translation through eIF3-dependent recruitment of the ribosome;⁸ m⁶A in the 3′ UTR regulates mRNA stability and translation efficiency that depend on the m⁶A reader proteins YTHDF1, YTHDF2 and YTHDF3; m⁶A in the CDS regulates splicing that involves the m⁶A reader proteins YTHDC1, hnRNPC and hnRNPG,^9,10 and codon-dependent translational efficiency.¹¹ The large increase of the m⁶A peaks in the GF brain therefore could affect the m⁶A-dependent mRNA function in several different ways. This multifaceted m⁶A effects in the brain were consistent with the reactome analysis of biological pathways that showed the top enriched categories for m⁶A-containing transcripts, including mRNA splicing, cell cycle and signaling (Fig. 1l).

The GF brain may represent an under-developed state due to the lack of its microbiome exposure.^12–14 To obtain insight into whether this idea applies to m⁶A in the transcriptome, we compared our results with the published m⁶A patterns from the mouse embryonic brain¹⁵ using only the high-confidence m⁶A-containing transcripts from both studies (Fig. 1m). The E13.5 embryonic brain (7701) had a comparable amount of m⁶A-containing transcripts to the 4-week-old GF brain (8,643); the overlap between them was 60% for embryonic brain and 54% for GF brain. The embryonic brain (7701) had a much higher amount of m⁶A-containing transcripts than the 4-week-old SPF brain (2750); the overlap between them was 20% for embryonic brain and 57% for SPF brain. Ninety-seven percent of all GF and SPF brain m⁶A-containing transcripts overlapped, which explained the similar overlapping percentage of GF and SPF with the embryonic transcripts. These results suggest that in regards to m⁶A modification, the GF brain more closely resembles the embryonic brain than the SPF brain of the same age.

To obtain mechanistic understanding of m⁶A changes in GF and SPF tissues, we measured the levels of the mRNA m⁶A writer proteins METTL3 and METTL14, and the m⁶A eraser proteins FTO and ALKBH5 by western blot. We found that both m⁶A writer proteins and both m⁶A eraser proteins were highly overexpressed in the GF brain compared to the SPF brain (Figs. 1n, o, and Supplementary information, Fig. S8). In contrast to brain, the differential expression of these proteins in the intestine and liver was much less noticeable without a uniform trend (Fig. 1p). These results correlate well with the finding that the brain has the largest difference in the m⁶A pattern among the three tissues examined here. Furthermore, the simultaneous overexpression of the m⁶A writer complex and the erasers in the same tissue should increase the ability to rapidly tune the m⁶A pattern upon environmental changes.

In summary, here we show that the microbiome has a strong effect on host m⁶A mRNA modification. Among the brain, intestine and liver tissues, the largest effect is present in the brain, which is associated with overexpression of both m⁶A writer and eraser proteins; this result suggests that the brain tissue may be more sensitive to adjust the m⁶A methylome in response to the microbiome than other tissues. Future studies will reveal the specific microbial species and the molecular mechanisms that regulate the host m⁶A methylome.

Author contributions

X.Y.W. and T.P. designed and initiated the study. Y.L. and G.Z.L. analyzed sequencing data. X.Y.W. and W.J.C. performed experiments. H.L.S. helped prepare m⁶A sequencing libraries. A.M. performed 16S rRNA gene sequencing. A.M.E. analyzed 16S rRNA gene sequencing data. C.H., helped design the study. C.H. and A.M.E. helped interpret the data. X.Y.W., G.Z.L. and T.P. wrote the paper.

Competing interests

The authors declare no competing interests.

Supplementary information

Supplementary information accompanies this paper at 10.1038/s41422-018-0127-2.