In both animals and plants the preferred sequence consensus for methylation is GAC. Clancy et al.
) previously reported that m6
A is present within mRNA of sporulating yeast; we show here that such methylation is readily detectable following end labeling of T1 cut mRNA, indicating that, in yeast, m6
A is found in a GA and not in the CA or UA context. Thus the site of adenosine methylation in mRNA is conserved among evolutionary distant organisms such as plants animals and yeast, which further highlights the importance of this nucleotide modification.
In S. cerevisiae
certain small RNA molecules, such as tRNA and small non-coding RNAs may become polyadenylated (33–37
). Often this is the trigger for their degradation. In most cases the protein responsible for the polyadenylation is either the non-canonical poly(A) polymerase Pap2/Trf4 (34–36
), or Trf5 (38
), both are members of the TRAMP complex. Some of these short molecules may contain m6
A. Therefore, it remained a possibility that the m6
A observed in the poly(A) fraction in this and in earlier experiments was derived from an abundant methylated small non-coding RNA species that had been targeted for degradation, rather than from mRNA. To address this possibility, the poly(A) fraction was size-separated prior to digestion and labeling (). Nearly equal values of m6
A:A (~0.7–0.9%) were found in the three different poly(A) size fractions, suggesting that the m6
A may be distributed across the early meiotic poly(A) population and is not specifically associated with a short non-coding RNA class.
The ratio of labeled m6
A to A in our experiments is between 0.9–1% at 3 h after the shift to sporulation. This value is slightly lower, but comparable to other organisms, such as A. thaliana
, where the m6
A to A ratio varied between 0.4% and 1.5%, depending on developmental stage and tissue type (19
) or in mouse and human, where similar results have been found (our unpublished data). Using this experimental approach, only those adenosines in a GA sequence will be labeled, thus the observed m6
A:A ratio is really a measure of the Gpm6
A:GpA ratio. If all possible 16 dinucleotide pairs occurred with equal frequency, this would be equivalent to m6
A being 0.06% of the nucleotides in the mRNA sample as a whole, or 1 m6
A nucleotide per 1600 nt. If distributed evenly, this would equate to a little more than half of the mRNAs present being methylated. However, this may be an underestimate of true methylation levels. In plants and animals methylation can also occur in the sequence AAC, if this is also the case in yeast, then the actual frequency of m6
A will be higher.
The identification of methylated mRNA species is a pre-requisite for understanding how this base modification regulates gene expression in S. cerevisiae
and possibly other eukaryotes. We have had a monoclonal antibody developed, which recognizes the m6
A in single-stranded nucleic acids and we established a new method for immunoprecipitating individual mRNA molecules containing m6
A. Applying the immunoprecipitation approach, m6
A containing individual transcripts, expressed during the first 3 h of meiosis were isolated. In the first instance a mixture of mRNA from log phase, vegetative (does not contain m6
A) and sporulating yeast was subjected to IP. In this experiment we set out to specifically test if IME2
message is methylated. This transcript is an attractive target, as it is an abundant, meiosis-specific message that encodes a key early regulator of meiosis. In addition, Ime2 levels have to be tightly regulated for correct meiotic progression (40
), and Clancy et al.
) previously speculated that this might be a target for post-transcriptional regulation by Ime4. In this experiment, the enrichment of IME2
in the Ab bound fraction was nearly sevenfold higher relative to the Ab depleted fraction. Measurements were normalized relative to ACT1
, which can be assumed to be unmethylated in the RNA fraction derived from the vegetative cells, as methylation was not detectable in the mRNA from mitotic log phase cells (c). When mRNA from the meiotic cells alone was used, the enrichment of IME2
was still clear, but less pronounced (2-fold). This is consistent with IME2
being methylated and may also indicate that a small portion of ACT1
transcripts are also modified.
Two further messages, IME1
, were also tested using IP of meiotic mRNA. Both of these were enriched in the Ab bound fraction, with IME4
messages showing the highest level of enrichment. Until recently only the four messages, bovine prolactin, R. sarcoma
virus, SV40 viral mRNA and the mouse dihydrofolate reductase were reported to contain m6
). The immunoprecipitation approach allowed us to double the number of the currently known m6
A harbouring messages and will allow further global analysis of transcriptomes. Following the immunoprecipitation, IME2
transcript was purified using a biotinilated complementing DNA sequence and streptavidin beads. The m6
A to A ratio for the analyzed part of the message was 0.48%. This value is somewhat lower than seen in the mRNA population as a whole, but is not necessarily unexpected. When looking at individual transcripts rather than the total mRNA population, the percentage of end labeled adenosines that are methylated will depend upon the absolute numbers of both Gpm6
A and GpA. Thus, within the non-protected fragment of IME2
, there are 119 GA sites, of these 12 are in a GAC consensus. The Gpm6
A: GpA ratio of 0.48% would mean that out of 208 GpA one would be Gpm6
A. Since the unprotected fragment of IME2
contains only 119 GpA sites it takes 1.7 of these fragments to harbor one Gpm6
A, which is equivalent of 58% of IME2
messages being methylated. This value is consistent with the observed average for the mRNA population as a whole.
The position of the m6
A in a particular message could be indicative of its function. However precise mapping has only been achieved for two highly abundant transcripts (R. sarcoma
virus and bovine prolactin) and the methods used are not applicable to the yeast system. In the IME2
mRNA molecule we narrowed down the position of the m6
A to a region in the 3′-end containing the last 11 GACs of the molecule [including 3′-UTR (44
)]. The 5′ 1291-nt fragment of the IME2
molecule [including 5′UTR (44
)] appeared to be free of m6
It is thought provoking, that the deletion of the C
-terminal 240aa from the Ime2 protein will lead to spore number reduction, as well as the stabilization of the otherwise labile Ime2 protein (40
). This deletion removes the last 10, potential m6
A harboring GAC consensus sites, in the 3′-region.
It has recently been published that Khd1p binds to IME2
mRNA as well as several other mRNA molecules (46
). Khd1p differentially affects gene expression, possibly due to combinatorial arrangement with additional factors. However, it is not impossible that the translatability of IME2
message could be affected by altered Khd1p binding, due to the presence of m6
A residues. We have manually searched for CNN repeats, the Khd1 binding domains (46
), in the IME2
transcript (with UTRs) and two near CNN repeats were found in the 3′-region (). However, neither of the repeats overlaps with any of the GAC consensus sites. On the other hand, two potentially methylated GACs are sandwiched between the two CNN repeats, therefore it remains possible that methylation of adenosine(s) in any of these sites could modify the binding of Khd1 or an interacting factor, resulting in altered translatability.
Figure 6. CNN repeat candidates in the IME2 message. The IME2 message contains two groups of CNN repeats, highlighted as red boxes. Additional C-rich regions adjacent to CNN repeats are accentuated by red Cs. Three GACs lie between the CNN repeats (bold and large (more ...)
A role for m6
A in regulating translation would be consistent with work previously reported by Tuck et al.
). These authors found that poly(A) RNA, isolated from methotrexate resistant mouse sarcoma cells treated with the methylation inhibitor, cycloleucine, translated less efficiently in an in vitro
translation assay, compared to mRNA from the control, untreated cells. However, in their system, cap-associated methylation was also reduced, making the interpretation of their results difficult. The presence of m6
A in mRNA might alter splicing during yeast meiosis, as proposed by Clancy et al.
). However the three transcripts IME1
identified here as methylated messages, are not spliced. Therefore we think, it is unlikely that the primary function of mRNA methylation would be to act as a positive signal for splicing in yeast.
We also would like to speculate that the presence of m6A in meiotic messages could be a flag for message recycling or storage, thus promoting a quick response to sudden changes in conditions.
Our results demonstrate that substantial levels of internal adenosine methylation are present in the GpA context in mRNA of sporulating yeast. This is consistent with the preferred methylation consensus of higher eukaryotes and is conserved between evolutionary distant species, which further highlights the importance of this nucleotide modification in the mRNA. Methylation is homogenously distributed across all mRNA size ranges, indicating that m6A is not limited to a small population of messages. If distributed evenly our measurements imply that slightly more than every second transcript in sporulating yeast is expected to contain m6A. Consistent with this, m6A is found once per 1.7 IME2 message. However, it is likely that transcripts of individual genes will vary in the degree or frequency of m6A. Using a novel immunoprecipitation approach we identify transcripts of three key, early regulators of meiosis, IME1, IME2 and IME4 itself, as being methylated. Methylation of these and other targets suggests mechanisms by which IME4 could control developmental choices leading to meiosis. For example, while a role in promoting splicing seems less likely, a function in translational control or message recycling remains a stronger possibility.