Supplementary Information Transcriptome wide Mapping Reveals …10.1038... · 2019-11-15 · 1 Supplementary Information Transcriptome-wide Mapping Reveals Reversible and Dynamic
Post on 10-Jul-2020
0 Views
Preview:
Transcript
1
Supplementary Information
Transcriptome-wide Mapping Reveals Reversible and Dynamic
N1-methyladenosine Methylome
Xiaoyu Li1,4, Xushen Xiong1,2,4, Kun Wang1, Lixia Wang1, Xiaoting Shu1,2,
Shiqing Ma1,2 & Chengqi Yi1, 3,*
1State Key Laboratory of Protein and Plant Gene Research, School of Life Sciences, and
Peking-Tsinghua Center for Life Sciences, Peking University, Beijing 100871, China.
2Academy for Advanced Interdisciplinary Studies, Peking University, Beijing 100871, PR
China
3Department of Chemical Biology and Synthetic and Functional Biomolecules Center, 10
College of Chemistry and Molecular Engineering, Peking University, Beijing 100871,
China.
4These authors contributed equally to this work.
*Correspondence: chengqi.yi@pku.edu.cn (C. Y.)
Nature Chemical Biology doi:10.1038/nbt.2040
2
Supplementary Results
Supplementary Table 1
Sequences of qRT-PCR primers used to validate m1A peaks.
Primer names Sequences
SUMO3-FWD ATCAACCTGAAGGTGGCCG
SUMO3-RVS CTTCATCAGCTTGCTCAGCG
SERP1-FWD CCAGTTCTCTTCCTCCTGCG
SERP1-RVS GAGGTCTGAGCACAAGCCG
TUBB4A-FWD CGGGAGATCGTGCACCTG
TUBB4A-RVS TCGATGCCATGTTCGTCACT
MFGE8-FWD TCGCCCTGGATATCTGTTCC
MFGE8-RVS TGCAGGTGTACGAGGGGAAG
HMGB1-FWD GGCAAGTGAGAGCCGGAC
HMGB1-RVS AGGATCTCCTTTGCCCATGT
LUC7L-FWD GATGCGGGCCCTGCTG
LUC7L-RVS TGTCTGGTTTCGTCTCCGTC
MRPL2-FWD CTCTGCGCTCTCTGAACCTG
MRPL2-RVS GGAGGCCATTGTTCATCATC
EEF1A1-FWD CGGTCTCAGAACTGTTTGTTTC
EEF1A1-RVS AAACCAAAGTGGTCCACAAA
18S-NC-FWD TTTCGAGGCCCTGTAATTGG
18S-NC-RVS GCTGCTGGCACCAGACTTG
MTRNR2L8-FWD AGTATTAGAGGCACCGCCTG
MTRNR2L8-RVS AAGTGATTATGCTACCTTTGCAC
Nature Chemical Biology doi:10.1038/nbt.2040
3
Supplementary Figures
Supplementary Figure 1. Quantitative MS analysis of mRNA m1A methylation.
(a) Representative LC-MS/MS chromatograms of purified mRNA samples. The dotted
chromatograms of C, U, G and A are scaled to the left y axis, and the solid chromatograms
of m1A (red), m6A (black) i6A (blue, indicated by the arrow) are scaled to the right y axis.
cps, counts per second. The MS parameters were optimized for m1A and A detection. (b)
Quantification of m1A and i6A during every step of the RNA purification procedure. The
stationary value of m1A/A ratio after the 2nd polyA+ selection step is indicated with the
dotted grey line. Values represent mean ± s.e.m. (n 4). (c) Quantification of m1A 10
methylation in different cell lines. Values represent mean ± s.e.m. (n = 4).
Nature Chemical Biology doi:10.1038/nbt.2040
4
Supplementary Figure 2. Specificity of the m1A antibody.
(a) Serial dilution of m1A, unmethylated and m6A oligos show the specificity of the m1A
antibody used in this study. m1A antibody can specifically recognize m1A oligos in
comparison to A oligos. Importantly, it showed no cross reactivity to the m6A oligos. The
well-characterized m6A antibody was used as control. (b) The ability of m1A antibody to
enrich m1A-containing RNA. Quantification was performed with qRT-PCR. m1A
immunoprecipitation was applied to fragmented total RNA (as input). m1A1322 in 28S
rRNA was used as the methylated RNA and 18S rRNA, which does not contain m1A, was
used as control. The ratio of methylated/control in input and in immunoprecipitated RNA 10
(IP) samples were calculated to measure the enrichment fold of m1A immunoprecipitation.
Values represent mean ± s.e.m. (n = 4).
Nature Chemical Biology doi:10.1038/nbt.2040
5
Supplementary Figure 3. Optimizations of the in vitro demethylation reaction
conditions
(a) Optimized pH conditions for the in vitro demethylation reactions. The results show that
demethylation fraction is highest under pH 6.5 condition. Values represent mean ± s.e.m.
(n = 2). (b) Optimized protein/RNA ratio for the in vitro demethylation reaction.
Demethylation efficiency is highest when protein/RNA ratio is 2:1. Values represent mean
± s.e.m. (n = 2). (c) Optimized demethylation recipe. As commercial BSA contains RNA
contamination, we performed demethylation with or without BSA. We found that BSA did
not affect the demethylation efficiency. Hence, we choose to avoid the addition of BSA to 10
the demethylation reaction for the immunoprecipitated m1A-containing RNA. Values
represent mean ± s.e.m. (n = 2).
Nature Chemical Biology doi:10.1038/nbt.2040
6
Supplementary Figure 4. m1A-ID-seq readily detects known m1A sites in rRNAs.
(a) Peak harboring the known site m1A1322 in human 28S rRNA. The dashed line
indicates its precise location. “Fold enrichment” was calculated by dividing the sequencing
depth of one given window of the IP samples with that of the input samples, hence
reflecting the enrichment of IP experiments. (b) Peaks harboring the known sites m1A645
and m1A2142 on Yeast 25S rRNA. Again, the dashed lines indicate their locations.
Nature Chemical Biology doi:10.1038/nbt.2040
7
Supplementary Figure 5. Demethylation sensitivity of sequencing profiles is
common to two widely-used RTases and can also be optimized for m1A mRNA
detection. (a) Demethylase-sensitive regions were also observed for m1A1322 detection
when Superscript III was used. The dotted line represents the position of m1A1322.
Although m1A-ID-Seq using Superscript III could also detect the known m1A1322 site, the
peak is wider and the background is higher compared to that of AMV. It showed that AMV,
which demonstrated higher signal to noise ratio, worked better for m1A detection in the
much less abundant mRNA. (b) Optimizations of AMV RT conditions. The stop rate of
m1A1322 under different dNTP and Mg2+ concentrations were measured. The stop rates 10
are almost similar under these conditions; however, the background is lowest under 0.2
mM dNTP and 10 mM Mg2+ conditions. Hence, this condition was used further in
m1A-ID-seq.
Nature Chemical Biology doi:10.1038/nbt.2040
8
Supplementary Figure 6. m1A-ID-seq reveals the distribution of m1A in the human
transcriptome. (a) A representative view of typical m1A peaks on non-coding RNA
(SPRNP1). (b) The histogram depicts relative enrichment of each mRNA segment. The
region “Near Start codon” was defined as 25 nucleotides upstream and 25 nucleotides
downstream the start codon.
Nature Chemical Biology doi:10.1038/nbt.2040
9
Supplementary Figure 7. m1A can be converted to m6A by Dimroth Rearrangement.
(a) Scheme of Dimroth Rearrangement. (b) Total RNA was divided into two parts: one part
was fragmented using regular fragmentation buffer (NEB) as the untreated sample, and
the other part was treated with with 0.1 M Na2CO3 /NaHCO3, pH 10.2 at 65 °C for 3 h as
the alkaline treated sample. LC-MS/MS chromatograms showed that after alkaline
treatment, signals according to m1A was significantly decreased while that of m6A
increased at the same time, indicating that most m1A was converted to m6A. (c) Under the
alkaline conditions, greater than 85% of endogenous m1A in mRNA was converted to m6A.
Nature Chemical Biology doi:10.1038/nbt.2040
10
Considering that m1A content is 5%-10% of m6A, the net m6A increase appears modest
after alkaline treatment. Values represent mean ± s.e.m. (n = 4).
Nature Chemical Biology doi:10.1038/nbt.2040
11
Supplementary Figure 8. Validation of m1A peaks identified by m1A-ID-Seq.
(a) Analyses of fragmentation size of untreated (regular fragmentation) and alkaline
treated samples using Bioanalyzer 2100. RNA was fragmented either by using regular
fragmentation buffer (as untreated) or by alkaline treatment. The fragmented RNA were of
very similar size for the untreated and alkaline treated samples. (b) The gel
electrophoresis of Supplementary Fig. 8a using Bioanalyzer 2100. (c) Procedure of m1A
peak validation utilizing Dimroth Rearrangement and subsequent m6A
Nature Chemical Biology doi:10.1038/nbt.2040
12
immunoprecipitation. Detailed experimental procedures can be found in Online Methods.
(d) The m6A-IP/input ratio of the target RNA region after alkaline treatment was
normalized by that of the untreated sample, generating the fold change (plotted in the
Y-axis). It showed that the m6A-IP/input ratio increased significantly after alkaline
treatment for m1A-containing regions, while no obvious changes were observed for
endogenous m6A-containing regions (EEF1A1) and the regions devoid of both m1A and
m6A (18S rRNA). Values represent mean ± s.e.m. (n = 3). (e) For an abundant
m1A-containing transcript, MTRNR2L8, both m1A immunoprecipitation and m6A
immunoprecipitation were performed to the untreated and alkaline treated samples. The
procedure of m1A immunoprecipitation was similar to that of Supplementary Fig. 5d, 10
besides the use of anti- m1A antibody. After alkaline treatment, the m6A-IP/input ratio of
MTRNR2L8 increased while its m1A-IP/input ratio decreased, again suggesting true m1A
modification in MTRNR2L8. Values represent mean ± s.e.m. (n = 3).
Nature Chemical Biology doi:10.1038/nbt.2040
13
Supplementary Figure 9. GO analysis of m1A-containing transcripts.
Gene Ontology analysis was performed for the human transcripts containing
high-confidence m1A peaks of wild type HEK293T, using DAVID bioinformatics database.
The GO terms enriched in Molecular Function (MF) were shown. (p-value provided by
DAVID)
Nature Chemical Biology doi:10.1038/nbt.2040
14
Supplementary Figure 10. Demethylase-sensitive regions and DS scores effectively
improved the resolution of m1A detection.
The red line represents the size of m1A peaks called from the (-) demethylase samples.
The blue line represents the size of demethylase-sensitive regions defined using DS score
by comparison the (-) demethylase samples and the corresponding (+) demethylase
samples. Density plot reveals that the resolution of m1A detection was improved from
~105 nt to ~55 nt when the demethylase-sensitive regions were calculated and defined.
More details about m1A peak calling and demethylase-sensitive region definition can be
found in Online Methods. 10
Nature Chemical Biology doi:10.1038/nbt.2040
15
Supplementary Figure 11. Additional sequence motifs found in m1A peaks by
MEME.
The other enriched motif discovered by MEME (E-value=2.4e-006, provided by MEME).
The presence of multiple motifs could be due to the presence of multiple
methyltransferases that have distinct specificities for mRNA targets.
Nature Chemical Biology doi:10.1038/nbt.2040
16
Supplementary Figure 12. Sequence and structure traits of m1A peaks.
(a) GC content comparisons between m1A peaks and control sequences. GC%: the
percentage of G+C contents within the m1A peaks. The control sequences were randomly
selected from the 5’ UTR or “start codon”, with the same length and number.
(Mann-Whitney U-test, two-sided, n=766, p < 2.2e-16). (b) Regions containing m1A tend
to be more structured compared to the control regions. MFE: minimum free energy. A
lower number of MFE indicates a more stable secondary structure. (Mann-Whitney U-test,
two-sided, n=766, p < 2.2e-16).
Nature Chemical Biology doi:10.1038/nbt.2040
17
Supplementary Figure 13. ALKBH3 exhibits robust in vitro demethylation activity to
m1A-containing RNA and DNA oligos.
(a) SDS-PAGE analysis of purified ALKBH3 protein. (b) Time curves of ALKBH3-mediated
(1 µM) demethylation of m1A in synthetic RNA or DNA oligos (10 µM). The RNA
sequence is: 5’ UGUUCCUUAA m1A UAGGGACUU 3’. A similar single-stranded DNA
sequence, which contains a central m1dA, was used for comparisons. Values represent
mean ± s.e.m. (n = 2).
Nature Chemical Biology doi:10.1038/nbt.2040
18
Supplementary Figure 14. ALKBH3 demethylates m1A in mRNA .
(a) The fold change of m1A/A ratio of the ALKBH3 knockout HEK293T cell line (comparing
to that of wildtype HEK293T cell line) for mRNA and small RNA (size below 200 nt).
Values represent mean ± s.e.m. (n = 4).The m1A/A ratio of mRNA in the ALKBH3
knockout HEK293T cell line increased while no obvious change for that of small RNA. (b)
The fold change of m1A/A ratio of HEK293T cells overexpressing ALKBH3 (comparing to
that of mock control, which was transfected with the empty vector) for mRNA and small
Nature Chemical Biology doi:10.1038/nbt.2040
19
RNA (size blew 200 nt). Values represent mean ± s.e.m. (n = 4). (c) The relative m1A/A
ratio of mRNA in the ALKBH3 knock down WPMY-1 and DU145 cells (comparing to that
of mock control, which was infected with virus packaged with a scrambled shRNA). Values
represent mean ± s.e.m. (n = 4). (d) The relative m1A/A ratio of mRNA in ALKBH3
overexpressing WPMY-1 and DU145 cells (comparing to that of mock control, which was
transfected with the empty vector). Values represent mean ± s.e.m. (n = 4). In all panels,
the level of ALKBH3 was indicated by western blot. -actin was used as a loading control.
The full gel is shown in Supplementary Figure 18.
Nature Chemical Biology doi:10.1038/nbt.2040
20
Supplementary Figure 15. Western blot analysis of ALKBH3 protein level indifferent
cell lines.
Protein level of different cell lines was analysed by western blot, using a commercially
available ALKBH3 antibody. -actin was used as a loading control. The full gel is shown in
Supplementary Figure 18.
Nature Chemical Biology doi:10.1038/nbt.2040
21
Supplementary Figure 16. Motifs found in ALKBH3-/- specific m1A peaks.
MEME algorithm was utilized to discover enriched motifs in ALKBH3-/- specific m1A peaks.
Top two motifs (a), (b), with lowest E-values are shown.
Nature Chemical Biology doi:10.1038/nbt.2040
22
Supplementary Figure 17. m1A is dynamically regulated by stress conditions.
(a) m1A is inducible in response to various stimuli in human mRNA. The m1A/A ratio was
measured by LC-MS/MS, and the fold change (relative to the untreated control) is plotted
on the y axis. Values represent mean ± s.e.m. (n = 4). (b) For three other cell lines, m1A
level in mRNA also increased in response to serum starvation. Values represent mean ±
s.e.m. (n = 4). (c), (d) Distribution of H2O2-induced and serum starvation-induced m1A
peaks along mRNA segments (5’ UTR, CDS, 3’ UTR), respectively. Transcript architecture
is shown beneath, with thin and thick bars representing UTRs and CDS, respectively. (e),
(f) Comparison of the expression level (RPKM) of the transcripts with inducible m1A sites. 10
Such transcripts showed very similar expression level before and after stress (left: H2O2,
Nature Chemical Biology doi:10.1038/nbt.2040
23
Mann-Whitney U-test, two-sided, n=206 p = 0.82; right: serum starvation, Mann-Whitney
U-test, two-sided, n=470, p = 0.86), indicating that the stress-induced pattern of m1A is not
due to the change of mRNA expression.
Nature Chemical Biology doi:10.1038/nbt.2040
24
Nature Chemical Biology doi:10.1038/nbt.2040
25
Supplementary Figure 18. Complete Western blot images
(a) From Supplementary Figs. 14a,b. (b) From Supplementary Fig. 14c. (c) From
Supplementary Figure 14d. (d) From Supplementary Figure 15.
Nature Chemical Biology doi:10.1038/nbt.2040
26
Supplementary Data Set 1
Contain the following four sheets, followed with explanation to each column.
1. m1A peaks of wild type HEK293T cells
The sheet contains the list of the overlapping m1A peaks (from two replicates)
identified by MACS2 in wild type HEK293T cells, as well as the
demethylase-sensitive regions of these peaks.
Table column explanation:
1) Gene name
2) Refseq gene ID 10
3) Peak start called by MACS2
4) Peak end
5) Peak length
6) Peak summit in absolute coordinate called by MACS2
7) Pileup height at peak summit
8) -log10(p-value)
9) Fold enrichment
10) -log10(q-value)
11-19) The same as in (2-10) for replicate 2
20) Intersecting length for the peaks shared by the two replicates. 20
21) If_demethylase-sensitive (if “no”, the following column would all be “.” or “-1”).
22) Refseq gene ID
23) Demethylase-sensitive region start
24) Demethylase-sensitive region end
25) Width of the Demethylase-sensitive region
26) Extended region start
27) Extended region end
28) Max DS score in the region
29-36) The same as in (22-28), for replicate 2.
30
2. m1A peaks of ALKBH3-KO cells
The sheet contains the list of the overlapping m1A peaks called by MACS2 in
ALKBH3-/- HEK293T cells.
The information in each column is the same as in columns (1-20) of Sheet 1 “m1A peaks of
wildtype HEK293T cells”
3. H2O2-induced m1A
The sheet contains the H2O2-induced m1A peaks called by MACS2.
The information contained in each column is the same as (1-20) of Sheet 1 “m1A peaks of
wild type HEK293T cells” 40
4. Serum starvation-induced m1A
The sheet contains the serum starvation-induced m1A peaks called by MACS2.
Nature Chemical Biology doi:10.1038/nbt.2040
27
The information contained in each column is the same as (1-20) of Sheet 1 “m1A peaks of
wild type HEK293T cells”
Nature Chemical Biology doi:10.1038/nbt.2040
top related