SmD 1 interplays with splicing , RNA quality control , and 1 post - transcriptional gene silencing in Arabidopsis 2 3

40 41 RNA quality control (RQC) eliminates aberrant RNAs based on their atypical structure, 42 whereas post-transcriptional gene silencing (PTGS) eliminates both aberrant and functional 43 RNAs through the sequence-specific action of short interfering RNAs (siRNAs). The 44 Arabidopsis thaliana mutant smd1b was identified in a genetic screen for PTGS deficiency, 45 revealing the involvement of SmD1, a component of the Smith (Sm) complex, in PTGS. The 46 smd1a and smd1b single mutants are viable, but the smd1a smd1b double mutant is embryo47 lethal, indicating that SmD1 function is essential. SmD1b resides in nucleoli and 48 nucleoplasmic speckles, co-localizing with the splicing-related factor SR34. Consistent with 49 this, the smd1b mutant exhibits intron retention of certain endogenous mRNAs. SmD1 binds 50 to RNAs transcribed from silenced transgenes but not non-silenced ones, indicating a direct 51 role in PTGS. Yet, mutations in the RQC factors UPFRAMESHIFT3 (UPF3), 52 EXORIBONUCLEASE2 (XRN2), XRN3 and XRN4 restore PTGS in smd1b, indicating that 53 SmD1 is not essential for but rather facilitates PTGS. Moreover, the smd1b mtr4 double 54 mutant is embryo-lethal, suggesting that SmD1 is essential for mRNA TRANSPORT 55 REGULATOR4 (MTR4)-dependent RQC. These results indicate that SmD1 affects splicing, 56 RQC, and PTGS. We propose that SmD1 facilitates PTGS by protecting transgene-derived 57 aberrant RNAs from degradation by RQC in the nucleus, allowing sufficient amounts to enter 58 cytoplasmic siRNA-bodies to activate PTGS. 59 60 61 62 63 64 65 66 67 68 69 3 70 71 72 73 74 75 Introduction 76 77 Post-transcriptional gene silencing (PTGS) controls a wide diversity of processes in 78 eukaryotes through mRNA degradation mediated by small 21-22 nt short interfering RNAs 79 (siRNAs) (Baulcombe, 2004; Voinnet, 2009; Martinez de Alba et al., 2013). PTGS starts with 80 the production of double-stranded RNAs (dsRNAs) and their processing into siRNAs. These 81 small siRNAs trigger the sequence-specific cleavage of mRNAs containing complementary 82 sequences. When PTGS is induced by viral or transgenic RNAs, siRNAs target the 83 degradation of the invading RNAs but also of homologous endogenous mRNAs, if any. A 84 forward genetic screen based on the transgenic Arabidopsis thaliana line L1, which carries a 85 post-transcriptionally silent p35S:GUS sense transgene, identified about 50 PTGS-deficient 86 mutants that defined 12 independent SUPPRESSOR OF GENE SILENCING (SGS) loci. 87 Mutations in these 12 SGS loci also impair PTGS in line 2a3, which carries a p35S:NIA2 88 sense transgene that triggers co-suppression of the endogenous genes NIA1 and NIA2. A 89 forward genetic screen directly based on line 2a3 identified three additional loci (SGS13, 90 SGS14, and SGS15) required for 2a3 but not L1 silencing (Jauvion et al, 2010). So far, 91 SGS2/RDR6, SGS3, SGS4/AGO1, SGS5/HEN1, SGS6/MET1, SGS7/SDE5, SGS8/JMJ14, 92 SGS9/HPR1, and SGS13/SDE3 have been characterized (Elmayan et al., 1998; Fagard et al., 93 2000; Mourrain et al., 2000; Morel et al., 2002; Boutet et al., 2003; Jauvion et al., 2010; Le 94 Masson et al., 2012). During PTGS triggered by sense transgenes (S-PTGS), primary siRNAs 95 are produced by an unknown mechanism, methylated at their 3' end by the methyltransferase 96 HEN1 (HUA ENCHANCER 1) (Boutet et al., 2003; Li et al., 2005) before loading into 97 AGO1 (ARGONAUTE1), which cleaves complementary target RNAs (Morel et al., 2002; 98 Baumberger and Baulcombe, 2005). AGO1-mediated cleavage generates RNA fragments that 99 escape degradation due to the protective activity of SGS3 (SUPPRESSOR OF GENE 100 SILENCING 3) and are transformed into dsRNA by RDR6 (RNA-DEPENDENT-RNA101 POLYMERASE 6 (Mourrain et al., 2000). These dsRNA are processed into siRNA duplexes 102 by DCL4 (DICER-LIKE4) to produce secondary siRNAs. These secondary siRNAs are also 103 4 loaded onto AGO1, which cleaves complementary transgene mRNAs, resulting in an 104 amplification loop that reinforces silencing. MET1 and JMJ14 encode a DNA 105 methyltransferase and an histone demethylase, respectively, which likely play a role in 106 remodeling chromatin to allow the transcription of transgene-derived aberrant RNAs that 107 induce PTGS (Le Masson et al., 2012). Also, SDE5 and HPR1 encode RNA trafficking 108 proteins, which likely play a role in bringing RNA molecules at the right place during PTGS 109 (Hernandez-Pinzon et al., 2007; Jauvion et al., 2010; Yelina et al., 2010). 110 111 Components of RNA processing complexes that counteract PTGS also have been identified. 112 Known endogenous PTGS suppressors include 5’->3’ EXORIBONUCLEASE2 (XRN2), 113 XRN3, XRN4, and their regulator FIERY1 (FRY1) (Gazzani et al., 2004; Gy et al., 2007), 114 exosome components HUA ENHANCER2 (HEN2), mRNA TRANSPORT REGULATOR4 115 (MTR4), RIBOSOMAL RNA PROCESSING4 (RRP4), RRP6L1, RRP41, RRP44a, 116 SUPERKILLER3 (SKI3) (Moreno et al., 2013; Lange et al., 2014; Yu et al., 2015), decapping 117 components DECAPPING1 (DCP1), DCP2, VARICOSE (VCS) (Thran et al., 2012; Martinez 118 de Alba et al., 2015), nonsense-mediated decay (NMD) components UPFRAMESHIFT1 119 (UPF1) and UPF3 (Moreno et al., 2013), and 3’ end processing factors ENHANCED 120 SILENCING PHENOTYPE1 (ESP1), ESP4, ESP5, CARBONE CATABOLITE 121 REPRESSOR4a (CCR4a) and 3’->5’ POLY(A)-SPECIFIC RIBONUCLEASE (PARN) (Herr 122 et al., 2006; Moreno et al., 2013). This revealed the diversity of RNA regulation processes 123 intertwined with siRNA-mediated PTGS in all types of compartments (nucleolus, 124 nucleoplasm, cytoplasm). Likely, a tug of war between RNA quality control and RNA 125 silencing contributes to determine the final transcriptome of the cell by addressing aberrant 126 RNAs to one or the other degradation pathway. However, cellular factors that influence the 127 partitioning of aberrant RNAs to one or the other pathway remain unknown. 128 129 Here, we show that the PTGS-defective mutant sgs14 recovered from a genetic screen based 130 on the p35S:NIA2 sense transgene carries a deletion of the SmD1b gene, which encodes one 131 of the two orthologs of the yeast Sm domain-containing protein SmD1, a small nuclear 132 ribonucleoprotein of the conserved Smith (Sm) complex (Wang and Brendel, 2004). The Sm 133 group of proteins was named after Stephanie Smith, the first patient in which the systemic 134 lupus erythematosus-associated anti-Sm autoimmune antibodies were identified. Sm proteins 135 are highly conserved among protists, fungi, animals and plants, and can be classified in 136 several groups. A first group comprises the canonical proteins SmB, SmD1, SmD2, SmD3, 137 5 SmE, SmF, and SmG; a second group comprises related LSM proteins LSM1 to LSM8. 138 SmB/D1/D2/D3/E/F/G form the core particles of the U1, U2, U4 and U5 spliceosomal 139 ribonucleoproteins (RNPs) while LSM2-8 is part of the U6 small nuclear ribonucleoprotein 140 (snRNP) also involved in pre-mRNA splicing. LSM1-7 proteins form a different complex, 141 which participates in mRNA decapping in cytoplasmic processing bodies (P-bodies). 142 Additional components (up to LSM16) play various roles, including maturation of U3 small 143 nucleolar RNA (snoRNA), participation to the U7 RNP involved in the maturation of histone 144 mRNA, degradation of mRNA precursors in the nucleus, mRNA translational control, and 145 formation of P-bodies [(Golisz et al., 2013) and references therein]. The Arabidopsis genome 146 contains 42 Sm and LSM genes (Cao et al., 2011), among which very few have been 147 characterized (Perea-Resa et al., 2012; Golisz et al., 2013). In particular, the role of 148 Arabidopsis SmD1 identified here is not known, although its implication in splicing could be 149 suspected based on the function of yeast SmD1 in this process (Zhang et al., 2001). As we 150 report here, localization studies revealed that Arabidopsis SmD1b co-localizes with the 151 splicing-related factor SR34 in nuclear speckles. Consistent with this, the smd1b mutation 152 affects the splicing of several endogenous mRNAs. Arabidopsis SmD1b also binds to RNAs 153 transcribed from silenced transgenes but not non-silenced ones, indicating a connection 154 between splicing and PTGS. Nevertheless, PTGS is restored in smd1b upf3, smd1b xrn2, 155 smd1b xrn3 and smd1b xrn4 double mutants, indicating that SmD1b is not essential for PTGS. 156 Moreover, smd1b mtr4 mutants are not viable, indicating that SmD1b also participates to 157 RQC, at least MTR4-dependent RQC. Together, these results indicate that SmD1 influences 158 splicing and the partitioning of aberrant RNAs between RNA quality control and RNA 159 silencing pathways, revealing a broad role of SmD1 in the regulation of gene expression. 160 161 162 163