2019-04-12-文献精读2

3′ UTR lengthening as a novel mechanism in regulating cellular senescence

Posted by DL on April 12, 2019

Title:

3′ UTR lengthening as a novel mechanism in regulating cellular senescence

3’UTR延长作为调节细胞衰老的新机制

Abstract:

  Cellular senescence has been viewed as a tumor suppression mechanism and also as a contributor to individual aging. Widespread shortening of 3′ untranslated regions (3′ UTRs) in messenger RNAs (mRNAs) by alternative polyadenylation (APA) has recently been discovered in cancer cells. However, the role of APA in the process of cellular senescence remains elusive. Here, we found that hundreds of genes in senescent cells tended to use distal poly(A) (pA) sites, leading to a global lengthening of 3′ UTRs and reduced gene expression. Genes that harbor longer 3′ UTRs in senescent cells were enriched in senescence-related pathways. Rras2, a member of the Ras superfamily that participates in multiple signal transduction path-ways, preferred longer 3′ UTR usage and exhibited decreased expression in senescent cells. Depletion of Rras2 promoted senescence, while rescue of Rras2 reversed senescence-associated phenotypes. Mechanistically, splicing factor TRA2B bound to a core “AGAA” motif located in the alternative 3′ UTR of Rras2, thereby reducing the RRAS2 protein level and causing senescence. Both proximal and distal poly(A) signals showed strong sequence conservation, highlighting the vital role of APA regulation during evolution. Our results revealed APA as a novel mechanism in regulating cellular senescence.

  细胞衰老被视为一种肿瘤抑制机制,也是致使个体衰老的原因。最近有研究发现,癌细胞的信使RNA(mRNA)会通过选择性多聚腺苷酸化(APA)来广泛缩短其3’非翻译区(3’UTR)。但是,APA在细胞衰老过程中所起到的作用仍然难以捉摸。在这里,我们发现衰老细胞中的数百个基因倾向于使用远端poly(A)(pA)位点,导致3’UTR的全局延长和基因表达的降低。在衰老细胞中含有较长3’UTR的基因,其主要富集于衰老相关途径。 Rras2是参与多种信号转导途径的Ras超家族的成员,其偏好使用更长的3’UTR并且在衰老细胞中表现出表达量的降低。 Rras2的敲除会促进衰老,而Rras2的rescue则可逆转衰老相关的表型。从机制上来说,剪接因子TRA2B与位于Rras2的选择性3’UTR中的核心“AGAA”基序结合,从而降低RRAS2蛋白水平并引起衰老。近端和远端poly(A)信号均显示出强烈的序列保守性,突出了APA调控在进化过程中的重要作用。我们的研究结果表明,APA是一种调控细胞衰老的新机制。

Introduction

  Cellular senescence was originally described as a process that limits the proliferation of cultured human cells. After extensive proliferation, senescence occurs because of telomere shortening and loss in the absence of endogenous telomerase activity (Hayflick and Moorhead 1961; Olovnikov 1996). In addition to telomere erosion, many stimuli and stresses can cause cellular senescence, including DNA double-strand breaks, strong mitogenic signals, oxidative stress, and ectopic expression of cyclin-dependent kinase inhibitors (CDKIs) (Xue et al. 2004; Rodier and Campisi 2011; Li et al. 2013). Numerous morphological and molecular markers of senescent cells have been identified in recent decades, which include a flattened and enlarged cell morphology, increased senescence-associated β-galactosidase (SA-β-gal) activity, reduced proliferation rate, and expression of tumor suppressors, cell cycle inhibitors, and DNA damage markers (Dimri et al. 1995; Busuttil et al. 2003; Lopez-Otin et al. 2013; Munoz-Espin and Serrano 2014). Cellular senescence is viewed as an important mechanism for preventing cancer (Campisi et al. 2001). It is also involved in normal embryonic development and tissue damage (Munoz-Espin et al. 2013; Storer et al. 2013; Munoz-Espin and Serrano 2014). Removing senescent cells expands the healthy lifespan of mice (Baker et al. 2016). These results demonstrate the significance of cellular senescence.

  细胞衰老最初被描述为限制培养的人类细胞增殖的过程。在广泛增殖后,由于端粒缩短而缺乏内源性端粒酶活性而发生衰老。除端粒侵蚀外,许多应激和胁迫均可导致细胞衰老,包括DNA双链断裂,强烈的有丝分裂信号,氧化应激和细胞周期蛋白依赖性激酶抑制剂(CDKI)的异位表达。近几十年来,研究人员已经鉴定出许多衰老细胞的形态学和分子标记,其包括扁平和扩大的细胞形态,增加的衰老相关的β-半乳糖苷酶(SA-β-gal)活性,降低的增殖速率和肿瘤抑制因子的表达,细胞周期抑制剂和DNA损伤标记物。细胞衰老被认为是预防癌症的重要机制。它还参与了正常的胚胎发育和组织损伤,去除衰老细胞则可延长小鼠的健康寿命。这些结果证明了细胞衰老的重要性。

  A number of studies have shown that dramatic changes in the transcriptome and/or proteome accompany the phenotypic alterations of senescent cells (Kim et al. 2013b; Mazin et al. 2013; Waldera-Lupa et al. 2014; Wei et al. 2015) and that the development of senescence-associated phenotypes can be regulated by stage-specific gene expression modules (Kim et al. 2013b). Therefore, understanding the regulation of gene expression and the corresponding regulatory networks is crucial to dissecting the mechanism of cellular senescence. Alternative polyadenyla-tion (APA) is recognized as a crucial contributor to the regulation of mammalian gene expression (Di Giammartino et al. 2011; Elkon et al. 2013; Chen et al. 2017; Tian and Manley 2017). Cleavage and polyadenylation of nascent RNA is essential for mat-uration of the vast majority of eukaryotic mRNAs and determines the length of 3′ UTRs (Sachs 1990). The process requires several cis-acting RNA elements and dozens of trans-factors (Millevoi and Vagner 2010). The key cis-element is a six-nucleotide (nt) motif (termed the poly[A] signal or PAS), the canonical form of which is “AAUAAA.” The PAS determines recognition and cleavage by the 3′ end-processing machinery (Proudfoot and Brownlee 1976). The mammalian 3′ end-processing machinery contains several subcomplexes as well as additional accessory factors which mediate the precise processing of mRNA precursors (pre-mRNAs) together (Di Giammartino et al. 2011; Elkon et al. 2013).

  许多研究表明,转录组和/或蛋白质组的显着变化伴随着衰老细胞的表型变化并且衰老相关表型的发展可以通过阶段特异性基因表达模块进行调控。因此,理解基因表达的调控和相应的调控网络对于剖析细胞衰老的机制至关重要。选择性多聚腺苷酸化(APA)被认为是哺乳动物基因表达调控的一大关键因素。新生RNA的切割和多聚腺苷酸化对绝大多数真核mRNA的研究非常重要,并决定了3’UTR的长度。该过程需要几种顺式作用RNA元件和数十种反式因子的参与。关键顺式元件是六核苷酸(nt)基序(称为poly [A]信号或PAS),其规范形式为“AAUAAA”。PAS决定3’末端加工机器的识别和切割。哺乳动物3’末端加工机器含有几个亚基复合物以及其他辅助因子,它们共同介导mRNA前体(前mRNA)的精确加工。

  With the increasing application of high-throughput sequencing technologies, genome-wide studies indicate that most eukary-otic mRNA genes have multiple polyadenylation (pA) sites (Tian et al. 2005; Wang et al. 2008; Zheng and Tian 2014). Alternative pA sites can reside in the 3′-most exon or upstream regions and can give rise to multiple mRNA transcripts that contain different coding sequences, 3′ UTRs, or both (Di Giammartino et al. 2011; Elkon et al. 2013). Importantly, both microRNAs (miRNAs) and RNA binding proteins (RBPs) targeting 3′ UTRs are able to regulate translational efficiency, degradation, and subcellular localization of mRNA or protein (Di Giammartino et al. 2011; Berkovits and Mayr 2015; Tian and Manley 2017). It is well known that APA plays important roles in a wide range of biological processes such as cell differentiation (Ji et al. 2009; Mangone et al. 2010; Hilgers et al. 2011; Li et al. 2012; Ulitsky et al. 2012; Fu et al. 2016; Hu et al. 2017), cell proliferation (Sandberg et al. 2008; Elkon et al. 2012; Hoffman et al. 2016), cell/tissue identity (Zhang et al. 2005; Derti et al. 2012; Smibert et al. 2012; Ni et al. 2013), and carcino-genesis (Mayr and Bartel 2009; Fu et al. 2011; Lin et al. 2012; Xia et al. 2014). However, whether APA is involved in senescence-associated gene expression and contributes to cellular senescence remains to be answered.

  随着高通量测序技术应用的不断应用,全基因组研究表明,大多数真核mRNA基因具有多个多聚腺苷酸化(pA)位点。选择性pA位点可以位于3’最外部或上游区域,并且可以产生包含不同编码序列,3’UTR或兼具两者的多个mRNA转录本。重要的是,靶向3’UTR的microRNA(miRNA)和RNA结合蛋白(RBPs)能够调节mRNA或蛋白质的翻译效率,降解和亚细胞定位。众所周知,APA在广泛的生物过程中起重要作用,比如细胞分化、细胞增殖、细胞/组织鉴定和致癌性。然而,APA是否参与衰老相关的基因表达并导致细胞衰老仍有待回答。

  We therefore examined the potential role and possible mechanism of APA in cellular senescence by applying our polyadenylation sequencing (PA-seq) approach (Ni et al. 2013) in two cellular senescence models, the passage of mouse embryonic fibroblasts (MEFs) and aortic vascular smooth muscle cells of rats (rVSMCs) at different ages.

  因此,我们通过应用我们的多聚腺苷酸化测序(PA-seq)方法在两种细胞衰老模型中研究了APA在细胞衰老中的潜在作用和可能的机制:小鼠胚胎成纤维细胞通道(MEFs) )和不同年龄的大鼠主动脉血管平滑肌细胞(rVSMCs)。

Results

Global lengthening of 3′ UTRs couples with decreased gene expression in senescent cells

3’UTR的全局延长伴随着衰老细胞中基因表达的降低

  To determine whether APA plays a role during cellular senescence, we first used MEFs undergoing replicative senescence in vitro and having population doubling (PD) times of 6, 8, 10, and 11 passages (Fig. 1A; Supplemental Fig. S1; Dimri et al. 1995; Parrinello et al. 2003; Tian and Li 2014; Tigges et al. 2014). We applied RNA sequencing (RNA-seq) and PA-seq (Ni et al. 2013) to discern the relationship between gene expression and 3′ UTR length patterns in these cells. After confirming the reliability of called pA sites, such as genomic location distribution, poly(A) signal enrichment, and overlap with polyA_DB (Supplemental Figs. S2, S3; Supplemental Table S1), the effective 3′ UTR length (or weighted mean of 3′ UTR length) was used to estimate the relative trend of pA site usage and changes in 3′ UTR length according to our previous method, which took pA site location, the distance to a stop codon, and tag number into consideration (Ni et al. 2013). Effective 3′ UTRs showed a global lengthening trend during MEFs senescence (Fig. 1B), indicating a tendency to use the distal pA sites. To further evaluate the changes in 3′ UTR length at the level of individual genes, we compared effective 3′ UTR length in later passaged (PD8, PD10, and PD11) cells with that in the earlier passaged (PD6) cells using different cut-offs. Cells of later passages always had many more genes with a longer 3′ UTR compared with cells of the earlier passage (Fig. 1C). Moreover, the number of genes with a lengthened 3′ UTR gradually increased from PD8 to PD11, while the number of genes with a shortened 3′ UTR continuously decreased (Fig. 1C). The same trend of global 3′ UTR lengthening was further confirmed by adopting a different methodology, the relative usage of distal polyadenylation sites (RUD) index (Ji et al. 2011), using separate RNA-seq data from the same MEFs used in PA-seq (Supplemental Figs. S4, S5; Supplemental Table S2). Biological replicates of PD11 MEFs had more genes with a longer 3′ UTR compared with those of PD6 MEFs (Supplemental Fig. S6; Supplemental Table S3). Together, these results demonstrate the global lengthening of 3′ UTRs in senescent MEFs.

  为了确定APA是否在细胞衰老过程中发挥作用,我们首先使用MEFs在体外进行复制衰老,并且设置群体倍增(PD)时间分别为6,8,10和11次(图1A;补充图S1)。我们应用RNA sequencing(RNA-seq)和PA-seq来识别这些细胞中的基因表达和3’UTR长度模式之间的关系。在确认了pA位点的可靠性后(如基因组位置分布,poly(A)信号富集,以及与polyA_DB重叠)(补充图S2,S3;补充表S1),我们利用有效的3’UTR长度(或3’UTR长度的加权平均值)来估计PA位点使用的相对趋势和3’UTR长度的变化,该方法考虑了PA位点的位置、到终止密码子的距离和标签。有效的3’UTR在MEFs衰老期间显示出全局延长的趋势(图1B),这表明使用使用远端pA位点的趋势。为了进一步评估单个基因水平上3’UTR长度的变化,我们比较了后期传代(PD8,PD10和PD11)细胞中的有效3’UTR长度与使用不同截止值的早期传代(PD6)细胞中的有效3’UTR长度。与早期传代的细胞相比,后来传代的细胞总是具有更多具有更长3’UTR的基因(图1C)。此外,具有延长的3’UTR的基因数量从PD8逐渐增加至PD11,而具有缩短的3’UTR的基因数量连续减少(图1C)。通过采用不同的方法,使用不同的远端多聚腺苷酸化位点(RUD)指数,并使用和PA相同来源的MEF的单独RNA-seq数据,我们进一步证实了全局3’UTR延长的相同趋势。(补充图S4,S5;补充表S2)。与PD6 MEF相比,PD11 MEF的生物重复具有更多的长3’UTR的基因(补充图S6;补充表S3)。总之,这些结果证明了衰老MEF中3’UTR的全球延长。

  To determine whether cell cycle affects the length of 3′ UTRs during senescence, PD6 MEF cells were subjected to serum starvation to drive them into G0 phase (Gustincich and Schneider 1993). The global pattern of 3′ UTR length in G0 cells was most similar to that in PD6 compared with other passages (Supplemental Figs. S5A, S7; Supplemental Tables S1, S2), implying that the global lengthening of 3′ UTRs resulted from senescence rather than from cell cycle alterations.

  为了确定细胞周期是否会影响衰老过程中的3’UTR长度,我们对PD6 MEF细胞进行血清饥饿以使它们进入G0期。 与其他传代相比,G0细胞中3’UTR长度的全局模式与PD6中的最相似(补充图S5A,S7;补充表S1,S2),这意味着3’UTR的全局延长是由衰老引起的,而不是细胞周期的改变。

  A longer 3′ UTR region could provide more opportunities for regulation by miRNAs and/or RBPs, which would influence mRNA and/or protein abundance at the post-transcriptional level (Mayr and Bartel 2009; Zheng and Tian 2014). In line with this hypoth-esis, a global decline in gene expression was observed for genes preferring distal pA sites during senescence (Supplemental Fig. S8A,C–E). In contrast, control genes with a single pA site or favor-ing proximal pA sites during senescence did not show such a trend (Supplemental Fig. S8B,F,G). Meanwhile, changes in APA and gene expression at the individual gene level were also analyzed by comparing PA-seq data from PD11 and PD6 MEFs using TSI (tandem 3′ UTR isoform switch index), a higher value of which indicates 3′ UTR lengthening (Fu et al. 2011; Li et al. 2012). Genes with a significantly lengthened 3′ UTR outnumbered those containing a shortened one (Fig. 1C,D; Supplemental Fig. S9). Within 3′ UTR lengthened genes, more genes showed reduced expression than elevated expression (P < 5.6 × 10−12, binomial test) (Fig. 1D, right half), while no such difference was observed for genes with shorter 3′ UTRs (Fig. 1D, left half). These data suggest that global 3′ UTR lengthening is associated with decreased gene expression during cellular senescence.

  较长的3’UTR区域可以为miRNA和/或RBP提供更多的调节机会,这将影响转录后水平的mRNA和/或蛋白质丰度(Mayr和Bartel 2009; Zheng和Tian 2014)。与该假设一致,在衰老过程中观察到优选远端pA位点的基因的基因表达的全局下降(补充图S8A,C-E)。相反,在衰老期间具有单个pA位点或有利于近端pA位点的对照基因未显示出这种趋势(补充图S8B,F,G)。同时,还通过使用TSI(串联3’UTR同种型转换指数)比较来自PD11和PD6 MEF的PA-seq数据来分析个体基因水平的APA和基因表达的变化,其中较高的值表示3’UTR延长( Fu et al.2011; Li et al.2012)。具有显着延长的3’UTR的基因数量超过包含缩短的基因的数量(图1C,D;补充图S9)。在3’UTR延长的基因内,更多基因表达减少而不是表达升高(P <5.6×10-12,二项式测试)(图1D,右半),而对于具有较短3’UTR的基因没有观察到这种差异(图1D,左半部分)。这些数据表明全球3’UTR延长与细胞衰老过程中基因表达降低有关。

  To explore whether APA-induced global lengthening of 3′UTRs occurred in other senescence systems, we performed PA-seq in VSMCs derived from young and aged rats (Supplemental Table S4). The results showed that senescent rVSMCs and MEFs had a similar ratio between numbers of genes with longer and shorter 3′ UTRs (Supplemental Fig. S9). A significant overlap of genes with longer 3′ UTR usage was observed in these two senes-cence models (Supplemental Figs. S10, S11). The observation that more genes with lengthened 3′ UTRs tended to be down-regulated was also verified in rVSMCs (P < 1.5 × 10−5, binomial test) (Fig. 1E). Together, the results indicate that APA-mediated 3′ UTR lengthening is involved in gene expression regulation in multiple cellular senescence systems.

  为了探究APA诱导的3’UTR全局延长是否发生在其他衰老系统中,我们在源自年轻和年老大鼠的VSMC中进行PA-seq(补充表S4)。 结果显示,衰老的rVSMC和MEF在具有更长和更短的3’UTR的基因数量之间具有相似的比例(补充图S9)。 在这两个senes-cence模型中观察到具有较长3’UTR使用的基因的显着重叠(补充图S10,S11)。 在rVSMCs中也证实了更多具有延长的3’UTR的基因倾向于下调的观察结果(P <1.5×10-5,二项式检验)(图1E)。 总之,结果表明APA介导的3’UTR延长参与多种细胞衰老系统中的基因表达调控。

Genes preferred distal pA sites in senescent cells enrich in senescence-associated pathways

衰老细胞中偏好使用远端pA位点的基因主要富集在衰老相关途径

  To further understand the correlation between pA site selection and cellular senescence, we performed pathway analysis on genes preferring distal pA sites in PD11 compared with PD6 MEFs. Genes containing a significantly longer 3′ UTR in PD11 compared with PD6 MEFs were enriched in pathways that are highly pertinent to cellular senescence (Fig. 1F), as were the genes whose 3′ UTRs pro-gressively lengthened in four different passages of MEFs (PD6, PD8, PD10, and PD11) (Supplemental Fig. S12). We also found similar enrichment in senescence-associated pathways between old and young rVSMCs (Fig. 1G). Notably, four of the shared pathways in senescent MEFs and rVSMCs, including ubiquitin-mediated pro-teolysis, the Wnt signaling pathway, cell cycle, and regulation of the actin cytoskeleton (Fig. 1F,G), are linked to cellular senescence (Amberg et al. 2012; Chandler and Peters 2013; Deschenes-Simard et al. 2014; Hofmann et al. 2014). Further examination of the genes preferring distal pA sites in mouse and rat revealed that these two species also possessed common genes undergoing APA regulation (Fig. 1H), which may serve as good candidates to study the function of 3′ UTR lengthening in cellular senescence.

  为了进一步了解pA位点选择和细胞衰老之间的相关性,我们对PD11中偏好远端pA位点的基因进行了通路分析(相比于PD6 MEFs)。与PD6 MEF相比,在PD11中含有更长的3’UTR的基因显着富集在与细胞衰老高度相关的途径中(图1F),在4个不同的MEF传代(PD6,PD8,PD10和PD11)的那些3’UTR逐渐延长的基因中也是如此(补充图S12)。我们还发现了衰老的和年轻的rVSMC之间也有着类似的衰老相关通路的富集(图1G)。值得注意的是,衰老MEF和rVSMC中四种共有的途径均与细胞衰老有关,包括泛素介导的促分裂作用,Wnt信号通路,细胞周期和肌动蛋白细胞骨架的调节(图1F,G)。我们进一步检查了在小鼠和大鼠中检查了具有远端pA位点偏好的基因,发现这两个物种在AP调控中也具有一些共同基因(图1H),这些基因可能是研究细胞衰老中3’UTR延长功能的良好候选基因。

APA-induced longer 3′ UTR of Rras2 reduces protein production and promotes cellular senescence

APA诱导的Rras2较长的3’UTR可减少蛋白质产生并促进细胞衰老

  To identify candidate genes that can affect cellular senescence through alternate pA site usage, we applied the following criteria:(1) favoring distal pA site usage both in senescent mouse and rat cells; (2) belonging to the shared four senescence-associated path-ways (Fig. 1H); and (3) exhibiting decreased expression during senescence. Based on these criteria, we focused on Rras2. The tendency of Rras2 to use distal pA sites in senescent cells was further confirmed. The UCSC Genome Browser displayed a higher usage of distal pA tags in both senescent MEFs (Fig. 2A) and rVSMCs (Supplemental Fig. S13). 3′ Rapid Amplification of cDNA Ends (3′ RACE) showed a reduced amplicon intensity of the proximal pA site in PD11 MEFs compared with PD6 MEFs (Fig. 2B). Real-time reverse transcription polymerase chain reaction (qRT-PCR) demonstrated a higher usage of longer 3′ UTRs in senescent MEFs (Fig. 2C). Rras2 mRNA and protein levels were both down-regulated in senescent MEFs (Fig. 2D,E), consistent with the idea that longer 3′ UTRs tended to display a decreased mRNA abun-dance. To investigate how APA-induced increase in Rras2 3′ UTR length contributes to gene expression changes, we evaluated the RNA degradation rate of Rras2 transcripts containing long and short 3′ UTRs. After blocking transcription followed by qRT-PCR, we found that Rras2 transcripts with a longer 3′ UTR were less stable than those with a shorter 3′ UTR (Fig. 2F). We then inserted the shorter 3′ UTR (represented as 3′ UTR_S) and the longer 3′ UTR (3′ UTR_L) with a mutated proximal PAS of Rras2 into a dual-luciferase reporter system, respectively. The reporter gene contain-ing 3′ UTR_L exhibited significantly reduced luciferase activities compared with that containing 3′ UTR_S (Fig. 2G), implying that the APA-induced longer 3′ UTR of Rras2 down-regulated RRAS2 protein abundance.

  为了鉴定出一些可以通过选择性pA位点的使用来影响细胞衰老的候选基因,我们使用了以下标准:(1)在衰老的小鼠和大鼠细胞中均偏好使用远端pA位点; (2)属于重叠的四个衰老相关的路径(图1H); (3)在衰老过程中表现出表达量的减少。基于这些标准,我们着眼于Rras2。我们进一步证实了Rras2在衰老细胞中使用远端pA位点的趋势。 通过UCSC基因组浏览器提供的信息,在衰老的MEF(图2A)和rVSMC(补充图S13)中均显示出更多的远端pA标签使用。 3’cDNA末端的快速扩增(3’RACE)显示,与PD6 MEF相比,PD11 MEF中近端pA位点的扩增子强度降低(图2B)。实时逆转录聚合酶链反应(qRT-PCR)显示出在衰老的MEF中的更强的3’UTRs使用偏好(图2C)。 Rras2 mRNA和蛋白质水平在衰老的MEF中均下调(图2D,E),这与更长的3’UTR倾向于具有较少的mRNA丰度的观点相一致。为了研究APA诱导的Rras2 3’UTR长度增加如何促进基因表达变化,我们评估了含有长和短3’UTR的Rras2转录本的RNA降解速率。在阻断转录然后进行qRT-PCR后,我们发现具有较长3’UTR的Rras2转录本比具有较短3’UTR的转录本稳定性要差(图2F)。然后我们将较短的3’UTR(表示为3’UTR_S)和较长的3’UTR(3’UTR_L)与Rras2的突变近端PAS分别插入双荧光素酶报告系统。含有3’UTR_L的报告基因与含有3’UTR_S的荧光素酶活性相比,显示出荧光素酶活性显着降低(图2G),这意味着APA诱导的Rras2的较长3’UTR会下调RRAS2蛋白丰度。

  To determine whether decreased levels of RRAS2 trigger senescence-associated phenotypes, we depleted RRAS2 with two short hairpin RNAs (shRNA-mediated knockdown; Rras2-KD) in mouse NIH3T3 cells (Fig. 2H; Supplemental Fig. S14A) and observed delayed cell proliferation and increased SA-β-gal staining (Fig. 2I, J; Supplemental Fig. S14B,C). Cell cycle analysis showed that RRAS2-depletion reduced the percentage of S phase cells (Fig. 2K, L). Cdkn1a, which encodes cyclin-dependent kinase inhibitor 1A (P21), showed increased expression upon Rras2-KD (Fig. 2M). Moreover, Rras2-KD in primary MEFs has similar effects to those in NIH3T3 cells (Fig. 2N; Supplemental Fig. S15). In addition, rein-troduction of RRAS2 (Rras2-Rescue) into Rras2-KD cells rescued the expression level of RRAS2 (Fig. 2O) and gave rise to reduced SA-β-gal signals (Fig. 2Q). Through RNA-seq profiling of Rras2-KD and Rras2-Rescue NIH3T3 cells, we found that altered RRAS2 expression caused expression changes in a variety of genes (Fig. 2S). Gene Ontology (GO) and pathway analysis revealed that those genes showing opposite expression trends were enriched in senes-cence-relevant biological processes, including cell cycle, cell adhe-sion, and DNA replication (Fig. 2T,U; Supplemental Table S5). More importantly, overexpression of RRAS2 (Fig. 2P) was able to re-verse the SA-β-gal staining (Fig. 2R) and rescue Cdkn1a expression levels in senescent MEFs (Supplemental Fig. S16). Taken together, these data indicate that RRAS2 plays a crucial role in delaying cellular senescence.

  为了确定RRAS2水平的降低是否引发了与衰老相关的表型,我们在小鼠NIH3T3细胞中用两个短发夹RNA(shRNA介导的敲除; Rras2-KD)敲除了RRAS2(图2H;补充图S14A)并观察到细胞增殖的延迟和SA-β-gal染色的增加(图2I,J;补充图S14B,C)。细胞周期分析显示,RRAS2的敲除会降低了S期细胞的百分比(图2K,L)。编码细胞周期蛋白依赖性激酶抑制剂1A(P21)的Cdkn1a在Rras2-KD上表达增加(图2M)。此外,初级MEF中的Rras2-KD具有与NIH3T3细胞中相类似的效果(图2N;补充图S15)。此外,将RRAS2(Rras2-Rescue)重新导入Rras2-KD细胞回补了RRAS2的表达水平(图2O)并致使SA-β-gal信号减少(图2Q)。通过RNA-seq分析Rras2-KD和Rras2-Rescue NIH3T3细胞,我们发现RRAS2的表达量变化会导致多种基因的表达变化(图2S)。GO和Pathway分析显示,那些表现出相反表达趋势的基因主要富含在与衰老相关的生物学过程中,包括细胞周期,细胞粘附和DNA复制(图2T,U;补充表S5) 。更重要的是,RRAS2的过表达(图2P)能够逆转SA-β-gal染色(图2R)并在衰老的MEF中回补Cdkn1a的表达水平(补充图S16)。总之,这些数据表明,RRAS2在延迟细胞衰老中起着至关重要的作用。

  To address whether APA in Rras2 is evolutionarily conserved between rodents and human, we verified the existence of its two pA sites in both human embryonic kidney (HEK) 293T cells and human umbilical vein endothelial cells (HUVECs) (Kim et al. 2007; Muck et al. 2008; Cardus et al. 2013; Zhang et al. 2014). Our PA-seq data from 293T cells (Ni et al. 2013) and public PolyA-seq data from human tissues confirmed the existence of both pA sites in human RRAS2 (Fig. 3A). The two pA sites were val-idated by 3′ RACE (Fig. 3B), and their PCR products were cloned and sequenced by the Sanger method to confirm the existence of poly(A) or poly(T) sequence at the end of the amplicon (Supplemental Fig. S17). Of note, senescent HUVEC cells also favored the distal pA site (Fig. 3C), which generated less protein (Fig. 3D,E). Knockdown of RRAS2 (Fig. 3F–I) increased SA-β-gal activity (Fig. 3J), slowed cell proliferation rate (Fig. 3K,L), and induced CDKN1A expression in both 293T cells and HUVECs (Fig. 3M,N). These findings indicate an evolutionarily conserved mechanism that longer 3′ UTR of Rras2 causes reduced protein production and then results in senescence in both rodents and human cells.

  为了评估Rras2中的APA是否在啮齿动物和人类之间进化保守,我们验证了其在人胚肾(HEK)293T细胞和人脐静脉内皮细胞(HUVEC)中存在的两个pA位点。我们使用来自293T细胞的PA-seq数据和来自人组织的公共PolyA-seq数据证实了人RRAS2中存在两个pA位点(图3A)。通过3’RACE对两个pA位点进行了鉴定(图3B),并且通过Sanger方法克隆并测序它们的PCR产物,以确认在扩增结束时poly(A)或poly(T)序列的存在。(补充图S17)。值得注意的是,衰老的HUVEC细胞也偏好远端pA位点(图3C),其会产生较少的蛋白质(图3D,E)。敲除RRAS2(图3F-I)会增加SA-β-gal的活性(图3J),减缓细胞增殖速率(图3K,L),并诱导293T细胞和HUVEC中CDKN1A的表达(图3M, N)。这些发现暗示出一种进化上的保守机制,即Rras2的较长3’UTR会导致蛋白质的产生减少,然后导致啮齿动物和人类细胞的衰老。

Splicing factor TRA2B represses RRAS2 protein level through binding to its alternative 3′ UTR and contributes to cellular senescence

剪接因子TRA2B通过与其选择性3’UTR的结合而抑制RRAS2蛋白水平并促进细胞衰老

  We reasoned that cis-elements recognized by either miRNAs or RBPs in the alternative 3′ UTR of Rras2 contribute to decreased RRAS2 protein production. To this end, we divided the 3′ UTR of Rras2 into four regions by deleting sequences of different lengths (shown as R1–R4 in Fig. 4A) and inserted them separately into a luciferase reporter, which was then transfected into mouse cells. Subsequent luciferase assays demonstrated significantly reduced luciferase activity from the construct containing a long 3′ UTR (re-sulting from mutated proximal PAS, labeled “M”) compared with the short 3′ UTR (labeled “S”) and the R1–R4 constructs (Fig. 4B). Thus, the sequence of last 148 base pairs (bp) located in the alternative 3′ UTR of Rras2 contains key motifs or elements essential for protein production.

  我们推断,Rras2的选择性3’UTR中的miRNA或RBP识别的顺式元件有助于降低RRAS2蛋白的产生。 为此,我们通过删除不同长度的序列(在图4A中显示为R1-R4)将Rras2的3’UTR分成四个区域,并将它们分别插入荧光素酶报告基因中,然后将其转染到小鼠细胞中。 随后的荧光素酶测定显示,与短3’UTR(标记为“S”)和R1-R4构建体相比,含有长3’UTR(从突变的近端PAS重新标记,标记为“M”)的构建体的荧光素酶活性显着降低。 (图4B)。 因此,位于Rras2的选择性3’UTR中的最后148个碱基对(bp)的序列含有蛋白质产生所必需的关键基序或元件。

  To further narrow the range of key elements, four more reporter constructs harboring the truncated 3′ UTRs (presented as R5–R8 in Fig. 4A) were generated. The construct lacking a 34-bp element showed significantly reduced luciferase activity, indicating that this element is critical for down-regulation of protein produc-tion (Fig. 4A,B). No miRNA binding sites were predicted in this 34-bp region; therefore, we searched RBPmap (Paz et al. 2014) for potential RBPs that might recognize this element and identified four RBPs (SPSF2, ZCRB1, TRA2B, and MBNL1). Following individual knockdown of the four RBPs, we measured the luciferase activities of S and M constructs (Fig. 4C). Depletion of TRA2B, a known splicing factor involved in mRNA processing, cell proliferation, and migration (Yang et al. 2015), eliminated the difference in luciferase activities between the S and M fragments-containing reporters (Fig. 4C), suggesting that TRA2B played an important role in down-regulating protein production. These data implied that the 34-bp element in the 3′ UTR of Rras2 is required for TRA2B-mediated down-regulation of RRAS2.

  为了进一步缩小关键元件的范围,我们构建了另外四个包含截短的3’UTR(在图4A中表示为R5-R8)的报告构建体。缺乏34-bp细胞的构建体显示出荧光素酶活性的显著降低,这表明该元件对于蛋白质产量的下调是至关重要的(图4A,B)。在这个34-bp区域,我们并没有预测到miRNA的结合位点;因此我们搜索了RBPmap可能识别该元件的潜在RBP,并确定了四个RBP(SPSF2,ZCRB1,TRA2B和MBNL1)。在对四种RBP进行单独敲除后,我们测试了S和M构建体的荧光素酶活性(图4C)。TRA2B(一种参与mRNA加工,细胞增殖和迁移的已知剪接因子)的敲除,可以消除含有S和M片段的报告基因之间荧光素酶活性的差异(图4C) ,表明TRA2B在下调蛋白质产量中发挥了重要作用。这些数据暗示出,RAS2的3’UTR中的34-bp元件是TRA2B介导的RRAS2下调所必需的。

  The “AGAA” element may serve as the core sequence of the TRA2B binding motif (Fig. 4D; Grellscheid et al. 2011). Considering that “AGAA” shared 1 nt with the downstream PAS “ATTAAA” in the alternative 3′ UTR of Rras2, we mutated the first 3 nt to avoid disruption of the PAS (“AGAA” to “CCCA”) (designated Mut3 in Fig. 4D). All 4 nt were also mutated (“AGAA” to “CCCC”) (denoted Mut4 in Fig. 4D) to validate the effects that are dependent on the whole core sequence. According to luciferase reporter assays, the Mut3 construct partially restored luciferase activity compared with that for M, while Mut4 showed a higher capability of restoring the repressed luciferase activity than Mut3 (Fig. 4D). To explore whether TRA2B directly binds to the alterna-tive 3′ UTR of Rras2, we carried out RNA immunoprecipitation coupled with both regular and quantitative PCR (RIP-PCR and RIP-qPCR) assays. The results showed an enrichment of TRA2B-binding signal (Fig. 4E), suggesting direct binding of TRA2B to the alternative 3′ UTR of Rras2. To further evaluate the role of “AGAA,” we moved this motif to two upstream positions in the alternative 3′ UTR of Rras2. The “AGAA” motif at a more proximal position did not repress protein production (Supplemental Fig. S18), implying that the “AGAA” motif in the alternative 3′ UTR of the Rras2 gene was required but not sufficient to repress protein levels. Together, these data suggest that the “AGAA” motif within the alternative 3′ UTR of Rras2 is essential for TRA2B binding and repression of RRAS2 protein production.

  “AGAA”元件可以作为TRA2B结合基序的核心序列(图4D)。考虑到“AGAA”与Rras2的选择性3’UTR中的下游PAS“ATTAAA”共享1nt,我们突变了前3nt以免破坏PAS(“AGAA”到“CCCA”)(designated的Mut3在图4D)中。所有4nt也被突变(“AGAA”至“CCCC”)(在图4D中表示为Mut4)以验证依赖于整个核心序列的效应。根据荧光素酶报告基因分析,与M相比,Mut3构建体可以部分恢复荧光素酶活性,而Mut4在恢复抑制的荧光素酶活性方面,能力比Mut3更强(图4D)。为了探究TRA2B是否能直接结合Rras2的选择性3’UTR,我们进行了RNA免疫沉淀以及常规和定量PCR(RIP-PCR和RIP-qPCR)测定。结果显示出TRA2B结合信号的富集(图4E),这表明了TRA2B与Rras2的选择性3’UTR的直接结合。为了进一步评估“AGAA”的作用,我们将这个基序移到了Rras2的选择性3’UTR中的两个上游位置。位于更近端位置的“AGAA”基序,不能抑制蛋白质的产生(补充图S18),这暗示出Rras2基因的选择性3’UTR中的“AGAA”基序是必需的,但其不足以抑制蛋白质水平。总之,这些数据表明,Rras2的选择性3’UTR内的“AGAA”基序对于TRA2B结合和抑制RRAS2蛋白的产生是必需的。

  To further confirm the contribution of TRA2B-RRAS2 regulation to cellular senescence, we overexpressed TRA2B in 293T cells, which show a higher usage of the distal pA site in RRAS2 and can serve as an appropriate cell model to perform the test, given that TRA2B bound to the alternative 3′ UTR of Rras2 (Fig. 3A). As expected, ectopic expression of TRA2B led to decreased RRAS2 protein level (Fig. 4F,G). Notably, neither RNA stability (Fig. 4H, I) nor mRNA steady-state levels of RRAS2 (Fig. 4J) showed considerable changes upon TRA2B overexpression, suggesting that the reduced RRAS2 protein level was caused by repressed translation through binding of TRA2B to the alternative 3′ UTR. Up-regulation of TRA2B led to a reduced proliferation rate (Fig. 4K), higher SA-β-gal staining level, and increased CDKN1A expression (Fig. 4L,M). Recovery of RRAS2 expression by additional overexpression of RRAS2 in TRA2B-overexpressed cells attenuated the CDKN1A expression and SA-β-gal staining to basal levels (Fig. 4N,O). Taken together, these data demonstrated that binding of TRA2B to the alternative 3′ UTR of Rras2 results in decreased RRAS2 ex-pression, which in turn accelerated cellular senescence.

  为了进一步证实TRA2B-RRAS2调控对细胞衰老的贡献,我们在293T细胞中过表达TRA2B,其显示出RRAS2中远端pA位点的更高的使用率。考虑到TRA2B可以与Rras2的选择性3’UTR结合,所以该细胞可以作为一个合适的细胞模型进行验证(图3A)。和我们预期相一致,TRA2B的异位表达导致RRAS2的蛋白水平降低(图4F,G)。值得注意的是,当TRA2B过表达时,RNA稳定性(图4H,I)和RRAS2的mRNA稳态水平(图4J)均未显示出较大的变化,表明RRAS2蛋白水平降低是由TRA2B与选择性3’UTR相结合导致的翻译抑制而引起的。TRA2B的上调导致增殖速率降低(图4K),SA-β-gal染色水平升高以及CDKN1A的表达增加(图4L,M)。通过在TRA2B过表达的细胞中额外过表达RRAS2来恢复RRAS2表达,可以使CDKN1A的表达和SA-β-gal的染色减弱至基础水平(图4N,O)。总之,这些数据证明,TRA2B与Rras2的选择性3’UTR的结合会导致RRAS2的表达减少,这反过来又会加速细胞衰老。

Key elements involved in APA regulation of Rras2 are evolutionarily conserved

参与APA调控Rras2的关键元件在进化上是保守的

  Since Rras2 prefers the distal pA site in senescent mouse, rat, and human cells (Figs. 2A–C, 3C; Supplemental Fig. S13), it would be interesting to have an evolutionary view on related Rras2 sequences. Comparative genomic analysis revealed that the 3′ UTR of Rras2 was extremely conserved in representative species from galliformes to primates (Supplemental Fig. S19), implying that regulation of 3′ UTR length by APA is of evolutionary significance. Notably, all listed animals contained a canonical PAS (AA/TTAAA) near the proximal pA site (Fig. 4P). This strongly indicated that APA-regulated Rras2 expression plays a crucial role during evolution. The sequence adjacent to the distal pA site within the 3′UTR of Rras2 also contained the canonical PAS “ATTAAA” and the TRA2B binding motif “AGAA,” which were conserved from birds to mammals (Fig. 4P), indicating that a strong selection pressure is likely to have driven the molecular evolution for TRA2B-mediated regulation of APA in Rras2.

  由于Rras2偏好衰老小鼠,大鼠和人细胞中的远端pA位点(图2A-C,3C;补充图S13),因此以进化的观点来看待相关的Rras2序列将是十分有趣的。比较基因组分析显示,Rras2的3’UTR在从鸡形目到灵长类的一些代表性物种中极其保守(补充图S19),这意味着APA对3’UTR长度的调控具有进化意义。值得注意的是,所有列出的动物在近端pA位点附近都含有经典的PAS(AA / TTAAA)(图4P)。这强烈地暗示出,APA调控的Rras2表达在进化过程中起着至关重要的作用。与Rras2的3’UTR内的远端pA位点相邻的序列还包含经典的PAS“ATTAAA”和TRA2B结合基序“AGAA”,其从鸟类到哺乳动物中均保守(图4P),这表明强烈的选择压力很可能推动了RAS2中TRA2B介导的APA调控的分子进化。

Discussion

  In this study, we discovered that APA-mediated 3′ UTR lengthening played a role in cellular senescence. As exemplified by Rras2 (Fig. 5), we extended the functional importance of APA to the aging field and provided a novel perspective for understanding the mechanism underlying cellular senescence. However, upstream factors controlling 3′ UTR lengthening during cellular senescence need to be further explored. Given that certain core factors in the 3′processing machinery were known to play a role in APA regulation (Takagaki et al. 1996; Zheng and Tian 2014), we therefore surveyed the gene expression of 24 important polyadenylation trans-factors. Most of these factors underwent a trend of decreased expression during MEFs senescence (Supplemental Fig. S20), consistent with recent findings that up-regulation of polyadenylation factors was associated with 3′ UTR shortening (Mayr and Bartel 2009; Xia et al. 2014). These results provided a possible mechanistic explanation for 3′ UTR lengthening during cellular senescence and deserved further investigation.

  在这项研究中,我们发现APA介导的3’UTR延长在细胞衰老中起到作用。如Rras2(图5)所示的那样,我们将APA功能的重要性扩展到衰老领域,并为理解细胞衰老的机制提供了新的视角。然而,那些在细胞衰老过程中控制3’UTR延长的上游因子,还需要被我们进一步探索。鉴于3’加工机制中的某些核心因素可以在APA调控中发挥作用,因此我们检查了24种重要的多聚腺苷酸化反式因子的基因表达。这些因素中的大多数在MEFs衰老过程中经历了表达量降低的趋势(补充图S20),这与最近的发现相一致,即多聚腺苷酸化因子的上调与3’UTR缩短相关。这些结果为细胞衰老过程中3’UTR延长提供了可能的机制解释,并且值得进一步的研究。

  Genes preferring distal pA sites showed a global lengthening of the 3′ UTR and a trend of decreased mRNA levels in both senescent MEFs and rVSMCs (Fig. 1), supplementing the observation that genes favoring proximal pA sites tended to have increased mRNA levels in cancer cells (Mayr and Bartel 2009; Xia et al. 2014). We also found that many genes displayed opposite pA site usage preference in senescent cells compared with cancer cells. Eighty-two and 166 genes that favored proximal pA sites in seven tumor types were prone to use distal pA sites in senescent MEFs and rVSMCs, respectively (Supplemental Fig. S21). In addition, 35 genes preferring shorter 3′ UTRs in multiple cancer cells tended to use longer 3′ UTRs in both senescent MEFs and rVSMCs (Supplemental Fig. S21). These findings supported a model that interaction between condition-specific trans-acting factors and dynamic changes in 3′ UTR length determined by APA could contribute to opposite biological processes, such as cellular senescence and tumor development.

  在MEF和rVSMCs中,偏好远端pA位点的基因均显示出3’UTR的全局延长和mRNA水平降低的趋势(图1),补充了前人在癌细胞中观察到偏好近端pA位点的基因倾向于增加mRNA水平的现象。我们还发现,与癌细胞相比,许多基因在衰老细胞中表现出与pA位点使用偏好相反的特征。在7种肿瘤类型中偏好使用近端pA位点的82个和166个基因分别在衰老的MEF和rVSMC中偏好使用远端pA位点(补充图S21)。此外,在多个癌细胞中偏好较短的3’UTR的35个基因倾向于在衰老的MEF和rVSMC中使用更长的3’UTR(补充图S21)。这些发现支持了一个模型,即条件特异性反式作用因子与APA确定的3’UTR长度的动态变化之间的相互作用可以归因于相反的生物过程,例如细胞衰老和肿瘤发展。

  There are two major categories of cellular senescence, developmentally programmed senescence and stress-induced premature senescence (SIPS) (Munoz-Espin and Serrano 2014). Here, we showed that MEFs in replicative senescence underwent global 3′UTR lengthening. VSMCs derived from old rats were likely to undergo a combination of replicative senescence and varieties of stress-induced senescence. Whether SIPS itself will induce global 3′ UTR lengthening needs to be further determined. Thus, more senescence models are required to fully understand the prevalence and func-tional relevance of 3′ UTR lengthening. In conclusion, our results provide evidence that APA contributes to the regulation of gene expression during cell senescence in multiple species, implying APA-regulated gene expression may be evolutionarily conserved.

  细胞衰老有两大类,程序性发育衰老和应激诱导的早衰(SIPS)。 在这里,我们发现复制衰老的MEF经历了全局3’UTR延长。 源自老年大鼠的VSMC可能经历复制衰老和多种应激诱导的衰老的组合。SIPS本身是否会引起全局3’UTR延长还需要进一步确定。 因此,我们需要更多的衰老模型来充分了解3’UTR延长的普遍性和功能相关性。 总之,我们的结果表明,APA有助于在多个物种的细胞衰老过程中调控基因表达,这意味着APA调控的基因表达可能在进化上是保守的。