Expanding the repertoire of deadenylases

ABSTRACT

Deadenylases belong to an expanding family of exoribonucleases involved mainly in mRNA stability and turnover, with the exception of PARN which has additional roles in the biogenesis of several important non-coding RNAs, including miRNAs and piRNAs. Recently, PARN in C. elegans and its homolog PNLDC1 in B. mori were reported as the elusive trimmers mediating piRNA biogenesis. In addition, characterization of mammalian PNLDC1 in comparison to PARN, showed that is specifically expressed in embryonic stem and germ cells, as well as during early embryo development. Moreover, its expression is correlated with epigenetic events mediated by the de novo DNMT3b methyltransferase and knockdown in stem cells upregulates important genes that regulate multipotency. The recent data suggest that at least some new deadenylases may have expanded roles in cell metabolism as regulators of gene expression, through mRNA deadenylation, ncRNAs biogenesis and ncRNA-mediated mRNA targeting, linking essential mechanisms that regulate epigenetic control and transition events during differentiation. The possible roles of mammalian PNLDC1 along those dynamic networks are discussed in the light of new extremely important findings.

KEYWORDS:

• Deadenylases
• mRNA turnover
• non-coding RNAs
• PARN
• piRNAs
• PNLDC1
• poly(A) tail
• stem cells

Abbreviations

DEDD	=	Asp-Glu-Asp-Asp residues
EEP	=	Endoculease/Exonulcease/Phosphatase
PARN	=	Poly(A) specific Ribonuclease
PNLDC1	=	PARN-like Domain Containing Protein-1
ESC	=	Embryonic Stam Cell
MZT	=	Maternal to Zygotic Transition
EGA	=	Embryonic Genome Activation
TE	=	Transposon Element

Introduction

Developmental process in mammals, from oocyte fertilization to early embryogenesis, depends on the stability and translation rate of maternal mRNAs or mRNAs produced upon embryonic genome transcription activation.¹ Posttranscriptional gene expression regulation (PTGR) through translational control and stability of specific messages, either maintains pluripotency or eventually drives differentiation and proliferation.² During these transitions, the stability and decay of mRNAs depend on the length of their 3′ poly(A) tails, which are shaped by opposing activities from poly(A) polymerases and specific exoribonucleases, known as deadenylases.³ The first deadenylases were discovered during early studies on essential factors which drive oocyte maturation and early development.⁴ Today, many deadenylases are considered not only passive mediators of mRNA decay, but rather dynamic and highly specific modulators of gene expression, in each step of development and differentiation.⁵

Deadenylases: mRNA decay and beyond

Deadenylases are 3′–5′ exoribonucleases responsible for the removal of the 3′ poly(A) tail of mRNAs and are divided in 2 major subclasses (DEDD: CAF1, PAN2, PARN and PNLDC1 and EEP: CCR4, Nocturnin, ANGEL and PDE12). Many deadenylase representatives have been studied in model organisms like S. cerevisiae, C. elegans, Drosophila, Xenopus, zebrafish, mouse and human. However, the genomes of many organisms contain homologues genes encoding putative and still uncharacterized deadenylases.⁶ The number of active deadenylases in each organism, increases as we climb up the evolutionary ladder. The redundancy of deadenylases in higher eukaryotes reflect their biologic significance and regulatory role in PTGR mechanism, although their basic role is the simple hydrolysis of an ester bond.⁷ In mammals, CCR4-CNOT complexes and, in many cases, PAN2-PAN3 complex, are the major deadenylases responsible for mRNA turnover. Their role is prominent, among other functions, in maternal mRNA clearance during MZT and EGA, in the prevention of cell death, in senescence, in DNA damage response, cell cycle control, neuronal development and in the generation of induced iPSCs.^6,8,9 Their expression is generally ubiquitous, but some deadenylases can be tissue-specific and in addition, their deregulation has been linked to pathologic conditions, including carcinogenesis.^10,11 All the known deadenylases shuttle between nucleus (found also in nucleoli) and cytoplasm (found mainly in stress granules, P-bodies and Cajal bodies), with the exception of PDE12 which is localized in mitochondria.¹² Knowledge on the possible epigenetic or posttranscriptional regulation of deadenylases is extremely limited, but they can be posttranslationally modified via phosphorylation or proteolysis (in the case of PARN).⁶ Attempts to deplete the expression of each mammalian deadenylase in search of the most essential one, showed that some of them can be dispensable.¹³ Deadenylases achieve a high degree of activity modulation, and in some cases specificity, through interactions with a diverse group of RNA binding proteins and ncRNAs (like miRNAs and recently piRNAs) which can be guided by characteristic 3′ UTR sequence signatures.¹⁴ The dynamic interplay between deadenylases and their regulators is constant during embryogenesis and differentiation, and leads to fine-tune post-transcriptional regulation of gene expression.¹⁵ Most notably, PABP and the BTG/TOB family can either stimulate or inhibit deadenylation.^16,17,18,19 In addition, combinations of RBPs that belong to CPSF, CPEB and PUF families, regulate localization, stability and translation of mRNAs during MZT, self-renewal of stem cells and differentiation.^14,15 Finally, important trans-acting factors such as SMAUG-response elements (SREs) are involved on mRNA stability and degradation in embryos during MZT.¹⁹ On top of the above, 2 of the most important and highly dynamic RNA binding complexes related to targeting of 3′ UTRs are the miRISC and piRISC, which inhibit translation and promote mRNA decay through deadenylation.²⁰ Recent advances that allowed measurement of poly(A) tails, revealed that during embryogenesis the length of poly(A) tail of mRNAs is variable and regulates translational efficiencies which mark specific transitions during development.²¹ A similar correlation between poly(A) tail length and translational efficiencies was recently correlated with the progression of cell cycle in somatic cells.²²

PARN and PNLDC1: 2 sides of the same coin?

PARN was among the first mammalian deadenylases described to play role in poly(A) shortening.²³ It was initially considered as the major deadenylase with role in maternal mRNA clearance during Xenopus oocyte maturation.²⁴ PARN is the only deadenylase which interacts with both the 5′m7G cap and the poly(A) tail. It is modulated by a large number of 3′ cis-elements and trans-acting factors and, until recently, it was the only deadenylase which extended its activity beyond mRNA decay.²³ PARN is expressed ubiquity and its subcellular localization to nucleoli and Cajal bodies lead to the observation that besides mRNAs, deadenylates also snoRNAs and scaRNAs of the H/ACA box type.²⁵ Moreover, human PARN participates in the maturation of small rRNA subunits and the telomerase RNA (TERC), linking PARN mutations with telomere diseases such as familial idiopathic pulmonary fibrosis (IPF) and dyskeratosis congenital.^26,27 Finally, mammalian PARN catalyzes miRNA-mediated degradation of TP53 and the Dicer-independent 3′ end trimming of the erythorpoietic pre-miR-451 which is conserved in all vertebrates.^28,29 These additional substrate specificities of PARN highlight not only the essential role of deadenylases in additional regulatory pathways, but also strongly suggest that depending on the trigger signal, deadenylases could more widely contribute in the regulation of gene expression.³⁰

Recently, Mello's laboratory identified in C. elegans PARN-1 as the elusive exonuclease that trims pre-piRNAs, an essential and very dynamic class of small non coding RNAs (26–31nt), first discovered in Drosophila germline.³¹ Acting together with PIWI proteins, piRNAs are mainly responsible for TEs repression through epigenetic modifications (H3K9me3 or CpG methylation) thus preserving genome integrity during reprogramming.^32,33,34 Besides germline, the presence of piRNAs and/or piRNAs/PIWI complexes has been reported in somatic cells (Drososphila) and in mammalian stem cells, oocytes and early embryos.^35,36,37 Besides their main role, piRNAs can trigger maternal mRNAs clearance through deadenylation during late spermiogenesis and in the early Drosophila embryo.^38,39,40 Current knowledge on the evolutionary origin, the exact pathways of biogenesis and the specific roles of piRNAs in various organisms, is still limited.^41,42

PARN-1 trimming activity was detected independently of interactions with auxiliary proteins and in vitro biochemical characterization showed that recombinant PARN-1 is a bona fide deadenylase localized in the germline and in P granules.³¹ In a concurrent study published in the same issue of Cell by Tomari's laboratory, the same Trimmer activity in B. mori was attributed to the putative deadenylase PNLDC1, a homolog of PARN.⁴³ Either one or both genes exist in almost all higher eukaryotes, with the exception of flies.^44,45 The co-existence of PNLDC1 and PARN is evident in all amniotic vertebrates but, it is also obvious that they form 2 distinct evolutionary branches, possibly separated after an early gene duplication event, when multipotency and cell differentiation mechanisms become the driving force for multicellular organisms. Although PNLDC1 was named after its homology to PARN, it remained until recently uncharacterized for its putative deadenylase activity. It must be noted, that in few databases PNLDC1 protein sequence (annotated also in some cases as PARN-like) is confused with that of PARN and therefore, for a long time PNLDC1 was overlooked. In the study by Tomari's laboratory, PNLDC1 knockdown inhibited pre-piRNA trimming leading to accumulation of piRNAs with 3′ extended ends. However, PNLDC1 pre-piRNA trimming activity was entirely depended on the presence of Papi (a Tudor protein homolog) which, anchored on the mitochondrial surface, supplies MIWI-loaded pre-piRNAs to PNLDC1. It must be noted, that in contrast to PARN-1 from C. elegans, recombinant BmPNLDC1 was insoluble and therefore its putative deadenylase activity on mRNA-like substrates was not verified. In addition, the antibody used was unsuitable for subcellular localization studies and therefore the presence of BmPNLDC1 in the mitochondrial fraction was verified based on the separation of cell extracts and subsequent pre-piRNA trimming activity assays of the fractions. Both studies in their conclusive remarks suggested that either PARN or PNLDC1 could be responsible for pre-piRNA trimming activity in mammals and more specifically in mouse. The suggestions were based on the reported elevated PNLDC1 expression in mouse testis (annotated in relative databases) and the reconstitution of the pre-piRNA trimming activity after expression of both mouse PNLDC1 and Tdrkh (the mouse homolog of BmPapi) in HEK293T cells.⁴³ In a very thorough report by Brennecke's laboratory that followed, Nibbler was identified as the 3′–5′ exonuclease which besides miRNAs, trims also pre-piRNAs in Drosophila, where both PARN and PNLDC1 are absent.^45,46,47 Interestingly, in C. elegans, Nibbler is required for 22G siRNA biogenesis.⁴⁸ All the described previously reports shed light on the largely unknown mechanisms of piRNA biogenesis, but the possible role of PARN-2 in C. elegans and of PARN in B. mori await further investigation. Given that the 2 fundamental functions of deadenylases is the balance of poly(A) tail length and the swift response to translational control of specific signaling pathways through decay of specific subsets of mRNAs, the recruitment of either PARN or PNLDC1 could facilitate specific adaptations during early development including, but not restricted to, piRNA biogenesis.

Mammalian PNLDC1: A link between epigenetic regulation, piRNA biogenesis and mRNA turnover?

Although PNLDC1 and PARN preserve the DEDD deadenylase subclass signature, they display limited sequence similarity.^44,49 As a consequence, it is not surprising that mammalian PNLDC1 exhibits different biochemical characteristics compared with PARN. It degrades poly(A) in a cap-independent fashion and has very strict poly(A) substrate specificity (Fig. 1). The differences with PARN extend also to the intracellular localization. PARN has ubiquitous expression and shuttles between nucleus and cytoplasm while mammalian PNLDC1 was found localized in the cytoplasm and mainly in the ER. Although clear and direct localization of PNLDC1 into mitochondrial surface was not detected, the possibility that PNLDC1 under specific circumstances can be localized there as well, cannot be excluded. Its B. mori homolog relies exactly on this localization for pre-piRNA trimming with the association of auxiliary proteins. Preliminary fractionation experiments in mESCs also verify this possibility (data not shown). The strong signal of PNLDC1 outside nucleus makes it the only known deadenylase to be exclusively localized in the cytoplasm, an observation suggesting a possible role in posttransciptional regulation (discussed below).

Figure 1. Illustration of verified and hypothetical roles of mammalian PNLDC1.

Mammalian PNLDC1 is also the first deadenylase to be regulated at the transcription level through epigenetic modifications introduced by the DNMT family of DNA methylatransferases.⁵⁰ The observation that PNLDC1 expression is reversibly linked to the expression of the de novo DNMT3b methyltrasnsferase, brings it in the center of extremely important events known to be regulated by DNMT methylatransferases, including development and genome reprogramming.^51,52,53 Interestingly, in cancer DNMTs appear deregulated and they are targeted by nucleoside analogs which act as demethylating agents, like 5′AZACdR, which in the case of PNLDC1, released its expression in HEK293 cells.⁵⁴ A more intriguing linked emerges from the fact that members of the DNMT family (including DNMT3b) are important in both germ and somatic cells and in addition, they are involved in the TEs silencing by CpG de novo methylation, in the mouse embryo germline via a piRNA related pathway which involves MIWI2 loaded with secondary piRNAs.^32,33 This is the first evidence of piRNA-mediated transcriptional silencing which involves DNMTs and correlates directly with the regulation of a deadenylase expression (Fig. 1). This hypothesis is substantially supported by the fact that PNLDC1 is practically undetectable in differentiated cells and is expressed only in stem or germ cells. Even more interestingly, the observation that PNLDC1 is highly expressed in human and mouse meiotic spermatocytes, where pachytene piRNAs drive a massive elimination of mRNAs through deadenylation, are supportive of the hypothesis that PNLDC1 is an additional regulator during this process.³⁸ Further experimentations is required to clarify whether PNLDC1 mediates piRNAs biogenesis per se or participates also in mRNA decay as deadenylase which is recruited by piRNA-mediated complexes. Given the strict in vitro specificity of mammalian PNLDC1 [piRNAs do not possess conserved 3′ poly(A) trailing signatures] and the fact that previous immunoprecipitation of TDRKH, MIWI and MILI proteins (all associated with piRNA binding) from mouse testes failed to co-precipitate PNLDC1, the second option is more favorable (Fig. 1).^55,56 However, the involvement of PNLDC1 in pre-piRNA trimming in early mammalian embryos and germ cell linage commitment mechanisms cannot be excluded, especially given a recent report showing that PIWI proteins and piRNAs bearing characteristic poly(A) marks exist in mammalian oocytes and early (2 to 4 cells) bovine embryos.³⁷

The intrinsic expression of PNLDC1 in mESCs allowed its effective knockdown which showed that depletion of PNLDC1 expression does not affect the remaining deadenylases, as previously observed.¹³ Downregulation of PNLDC1 was always correlated with DNMT3b expression during mESCs differentiation, an observation that suggest PNLDC1 participation at least in some of the pathways that maintain multipotency during early development and/or pathways that require or induce genome reprogramming and are piRNA-mediated.^57,58 Accordingly, transcriptomic analysis showed significant up- or downregulation of genes that shape cell cycle, are important for DNA replication repair and chromatin remodeling, transcription modification and translational regulation (Fig. 2).^59,60 All these events occur also during germline biogenesis and depend on dynamic processes from a very complex molecular landscape, which awaits further investigation. Given the in vitro biochemical character of PNLDC1, its putative participation in more than one events that affect transcriptome and epigenome during early differentiation, must be also considered.

Figure 2. Networks related to cell assembly and organization, DNA replication, recombination and repair, and posttranscriptional modification (A) and cell-cycle regulation (B) according to Ingenuity Pathway (IPA). Differentially expressed genes after PNLDC1 knockdown experiments were uploaded into IPA and the most significant enriched pathways were identified. Genes were overlaid onto a global molecular network developed from information contained in the Ingenuity® Knowledge Base. The obtained networks were algorithmically generated based on the connectivity of differentially expressed genes. Red symbols denote upregulated genes and green symbols denote downregulated genes.

Conclusions

Deadenylases are evolutionary old and important enzymes that have recently expanded their repertoire beyond mRNA turnover. Among them, mammalian PNLDC1 represents a new and unique member which does not act as PARN's alter ego. Its exclusive expression in stem and germ cells, its tight epigenetic regulation and strict substrate specificity suggest that PNLDC1 could be a master regulator of early developmental pathways that include, but not limited to, piRNA-mediated epigenetic silencing, piRNA-triggered mRNA decay and in certain cases, piRNA biogenesis. Given that piRNAs are not only present in germline but also in embryonic and adult stem cells, in mammalian oocytes and early embryos, the possible roles of mammalian PNLDC1 could shift according to specific cellular needs during development. The roles of mammalian PNLDC1 as a possible link between common mechanisms in germ and stem cells have just started to unveil.

Disclosure of potential conflicts of interest

No potential conflicts of interest were disclosed.

Funding

This work was supported in part and implemented under the “ARISTEIA I” (EXCELLENCE I) Action of the “OPERATIONAL PROGRAMME EDUCATION AND LIFELONG LEARNING” and is co-funded by the European Social Fund (ESF) and National Resources (MIS 1225, No D608 to C.S.). A-N.S is a recipient of “Bodossakis Foundation” postgraduate fellowship, which is gratefully acknowledged.