胚胎干细胞具有mRNA异构体特异性的翻译调控

Embryonic Stem Cells Exhibit mRNA Isoform Specific Translational Regulation

|核心内容：

许多mRNA的多种变异的存在是蛋白质多样性的主要原因。

这些变异的处理以细胞类型特异性的方式受到严格控制，并对基因表达控制有重大影响。

在这里，我们使用多聚体谱结合RNA测序，研究了胚胎干细胞(ESCs)和ESC衍生的神经前体细胞(NPCs)中单个mRNA变异的差异转译率。

我们发现，在ESCs和NPCs中有大量可检测到的mRNA变异，其中许多显示出变异的特异性转译率。

这与变异的5'UTR占主导作用的差异有关。

我们认为，包含替代utr的mRNA变体受到不同的转录后控制。

这可能是由于调节翻译速率的miRNA和蛋白质结合位点的存在或缺失。

这强调了当使用mRNA水平作为蛋白质丰度的读取时，处理翻译效率的重要性。

另外的分析表明，在npc中，ESCs中存在许多注释的非编码mrna在多聚体组分上。

我们认为，使用多聚体分离结合RNA测序是一种有用的方法来分析细胞中许多不同RNA的翻译状态。原文摘要：

The presence of multiple variants for many mRNAs is a major contributor to protein diversity.

The processing of these variants is tightly controlled in a cell-type specific manner and has a significant impact on gene expression control.

Here we investigate the differential translation rates of individual mRNA variants in embryonic stem cells (ESCs) and in ESC derived Neural Precursor Cells (NPCs) using polysome profiling coupled to RNA sequencing.

We show that there are a significant number of detectable mRNA variants in ESCs and NPCs and that many of them show variant specific translation rates.

This is correlated with differences in the UTRs of the variants with the 5’UTR playing a predominant role.

We suggest that mRNA variants that contain alternate UTRs are under different post-transcriptional controls.

This is likely due to the presence or absence of miRNA and protein binding sites that regulate translation rate.

This highlights the importance of addressing translation rate when using mRNA levels as a read out of protein abundance.

Additional analysis shows that many annotated non-coding mRNAs are present on the polysome fractions in ESCs and NPCs.

We believe that the use of polysome fractionation coupled to RNA sequencing is a useful method for analysis of the translation state of many different RNAs in the cell.

Fig 1. Polysome profiling of ESCs and NPCs.

(A) Efficiency of NPC differentiation. Mouse Sox1-GFP ESCs were differentiated to NPCs for 6 days. Flow cytometry of the SOX1-GFP signal showing over 80% SOX1-GFP positive cells after differentiation.

(B) ESCs and NPCs were subjected to polysome profiling and 12 fractions were collected into four groups. No translation F1-5), Low translation (F6-8), High translation (F9-11) and the bottom of the gradient (F12).

Fig 2. Transcriptional and translational changes upon differentiation of ESCs to NPCs.

(A) Pie chart showing the degree of translational shift of mRNAs during ESC to NPC differentiation. 5% of mRNAs display a translation shift greater than 20% upon NPC.

(B) Pie chart showing transcriptional changes in the sub-fraction of translationally regulated genes. 58% of these genes are purely regulated translationally and show less than 2 fold changes in total mRNA levels.

(D) Western blot of TCHP protein in ESCs and NPCs. GAPDH is shown as a loading control.

Fig 3. Translational regulation of splice variants in ESCs and NPCs.

(A) Analysis of the translation rate of all mRNAs with more than one variant in ESCs and NPCs.

(B) Charts showing the correlation between variants with different translation rates and altered UTRs.

Fig 4. Translational regulation of splice variants with different 3’UTR in ESC.

(A) Diagram showing the gene structure for three genes, Ankrd27, Mpzl1 and Araf, whose variants are differentially loaded with ribosomes in ESCs.

(B) qRT-PCR analysis of the polysome distribution of two variants of three genes in ESCs.

Luciferase reporter assay

The UTR of candidate mRNAs were cloned into the psiCHECK-2 vector. ESCs were transfected using FuGENE HD (Promega) according to manufacturer’s instructions. Thirty hours after transfection, cells were lysed and Renilla (RL) and firefly (FL) luciferase activities were determined using the Dual-Luciferase Reporter Assay system (Promega).

For data analyses, all RL signals were normalized by the non-targeted control FL readings.

荧光素酶报告基因是指以荧光素(luciferin)为底物来检测萤火虫荧光素酶(fireflyluciferase)活性的一种报告系统。荧光素酶可以催化luciferin氧化成oxyluciferin，在luciferin氧化的过程中，会发出生物荧光(bioluminescence)。

Fig 5. Translational regulation of splice variants with different 5’UTR in ESC.

(A) Diagram showing the gene structure for two genes, Igf2 and Rnps1, whose variants are differentially loaded with ribosomes in ESCs. (B) qRT-PCR analysis of the polysomes distribution of two variants of Igf2 and Rnps1 in ESCs.

Fig 6. Polysomal association of ncRNA in ESC and NPC.

(A,B) Analysis of the distribution of annotated non-coding RNAs in the different polysome fractions of ESCs and NPCs. (C) Polysome profiles of ESCs, NPCs and puromycin treated ESCs (PURO) to selectively disrupt polysomes. (D) qRT-PCR of representative non-coding RNAs showing their distribution in polysome gradients from ESCs and NPCs with and without the addition of puromycin.

Discussion

We have performed RNA sequencing analysis on mRNA isolated from different fractions of a polysome gradient in order to analyse the translation rate of individual mRNA variants. We find that many mRNAs are translationally regulated during differentiation of ESCs to NPCs.

We have validated that a shift in ribosomal load has a consequence for protein production with Tchp. Tchp is translationally upregulated in NPCs resulting in an increase in protein levels.

TCHP was originally described as a keratin filament binding protein and plays a role in ciliogenesis and centrosomal function [35, 36]. TCHP has no known role in the nervous system but given its translational upregulation it may be a regulator of early neural differentiation.

We identified 31 genes expressing multiple variants that are translationally regulated in a variant specific manner on NPC differentiation.

In these cases multiple variants are detected but only one is under translational control on differentiation. In a number of cases there was a switch in the variant that was highly translated so that while the absolute levels of the RNA variants may be the same, their different translation rates will result in the protein products being present at different levels in ESCs and NPCs.

Interestingly the role of the majority of these variants is not known so it remains to be determined how this variant specific translation impacts on neural differentiation.

The method of RNAseq analysis of polysomal fractions is different from the alternative approach of ribosomal footprinting which provides information on the mRNA sequence physically bound by ribosomes and is used to infer translation rate [32, 37].

Despite the high resolution of the method the footprint sequences do not provide information on UTRs. In addition, currently the technique is not able to distinguish between different splice variants. Polysome profiling coupled to RNAseq provides information on variant specific ribosomal load and enables the analysis of corresponding UTR sequences.

Using polysome profiling coupled with RNAseq, we find that in ESCs ten percent of RNA isoforms display different translation rates.

This correlates with differences in the UTR sequences which likely drive the altered translation rate. This suggests that the relative protein levels of the two isoforms will not correlate with their mRNA levels.

Alternate UTRs can arise through a number of different mechanisms including alternate TSS, alternative splicing and alternative polyA site selection.

Our datasets were mapped to the RefSeq genome which has annotations for promoter start sites and known alternative splicing events. As such, our studies have focused only on variants arising from alternative transcription start site selection and alternative splicing.

Our analysis does not take into account any APA that may be occurring and it is possible that the variants identified in our studies may be further regulated by APA in different cell types. We find that in ESCs 10% of mRNAs with multiple variants display different translation rates for each variant inferred from their ribosomal load.

This strongly correlates with the variants having different 5’ and 3’UTR sequences. This represents over 67 genes with different detectable isoforms in ESCs. This demonstrates the importance of analysing the translation rates of RNA variants when studying gene expression patterns. We confirmed the differential ribosomal load of five variants by PCR and performed luciferase assays to confirm that the regulation was mediated through sequences in the UTRs of these mRNAs.

We tested two variants that had altered 5’UTRs and three that had altered 3’UTRs. In all cases the UTR sequence was shown to regulate luciferase protein levels in a similar way to that predicted by the polysome profiling data. Where the variant had a decreased ribosomal load following polysome profiling, the UTR promoted a decrease in luciferase activity suggesting a role in translational repression. It is likely that the UTRs that are associated with a decreased ribosomal load have cis-acting sequences that confer translational repression.

These could take the form of binding sites for miRNA or RNA binding proteins. Alternate 5’UTRs could also contain uORFs which can function to repress translation [38, 39]. The variants that do not have these sequences are likely not targeted for repression. These different UTR sequences have likely evolved to confer additional levels of regulation on specific RNA variants and have the potential to precisely regulate their translation both spatially and temporally.

Analysis of the ORF size of the translationally regulated variants suggested that for most variants the differences in ribosomal load were not a consequence of altered ORF size.

There were a minority of variants that had significantly longer ORFs that correlated with increased ribosomal load and in these cases the ORF length is likely the cause. For variants that did show a correlation between ORF length and ribosomal load the difference in size was not large enough (less than 25% larger) to result in a shift in the polysome fractions in most cases (Table 1).

Interestingly we identified a selection of splice variants that had different translation rates but no change in the UTRs. While some of these candidates will have a decreased ribosomal load due to a decreased ORF length it is likely that the remaining transcripts contain cisacting regulatory sequences within their ORF.

miRNAs have been shown to target the ORF of mRNAs [40, 41] so it is possible that the changes in ORF seen in these variants results in altered translational control in addition to altered protein function. Exon 4 and 5 in Ctage5 variant 1 are associated with a decreased ribosomal load compared to variant 3 that skips these exons.

While there was no enrichment for rare codons in these exons it is possible that these regions could bind to trans-acting factors or that the sequence itself is inhibiting translation. It will be interesting to determine if these exons can function to regulate translation from within the UTR or if they have to be in the ORF to function.

We assessed the ribosomal load of annotated non-coding RNAs and found that a significant number of long non-coding RNAs (lncRNAs) and non-coding RNAs were associated with the ribosomes. The lnc-RNAs include predominantly long intergenic non-coding RNAs (lincRNAs) while the non-coding RNAs include predominantly pseudogenes.

The presence of the pseudogenes on the polysomes is in agreement with recent reports that analysed ribosomal footprinting data from a number of different species. [42] Recent studies have demonstrated the presence of many non-coding RNAs on polysomes.

Our studies confirm that many noncoding RNAs are in the polysome fraction but we cannot distinguish between RNAs that are actively being translated by the ribosome and those that may be binding the ribosome or other mRNA molecules that are being translated.

It is possible that many non-coding RNAs could be associating with the ribosome in a regulatory capacity and so while associated with the ribosome would show no ribosomal footprint. It has been suggested that non-coding RNAs could be being translated to give short peptides but additional studies are needed to elaborate on this idea [32, 42, 43].

Our data confirms the presence of non-coding RNAs on the ribosome and further investigation is needed to determine why they are there.

Taken together our data illustrates the importance of addressing the translation rate of individual mRNA variants.

We demonstrate that different mRNA variants can have very different translation rates.

This confirms and expands on previous reports from human cells and demonstrates variant specific translation rates in ESCs and NPCs.

Additional work is needed to determine the mechanisms by which these mRNAs are regulated and their significance for ESC self-renewal and pluripotency.

参考文献：Embryonic Stem Cells Exhibit mRNA Isoform Specific Translational Regulation

--------------------------------------------------------------

本站仅提供存储服务，所有内容均由用户发布，如发现有害或侵权内容，请点击举报。